[clang] [llvm] [SimplifyCFG] Emit SelectInst when folding branches to common dest with different PHI incoming values (PR #144434)
via llvm-commits
llvm-commits at lists.llvm.org
Fri Jul 4 19:08:17 PDT 2025
https://github.com/HighW4y2H3ll updated https://github.com/llvm/llvm-project/pull/144434
>From 20a06ff80f4a97a1d065c2a11e98f17f9b8dcffc Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Mon, 16 Jun 2025 00:12:43 -0700
Subject: [PATCH 1/8] [SimplifyCFG] Emit SelectInst when folding branches to
common dest with different PHI incoming values
---
llvm/lib/Transforms/Utils/SimplifyCFG.cpp | 82 ++++-
llvm/test/CodeGen/AArch64/rm_redundant_cmp.ll | 56 ++--
.../CodeGen/AArch64/tailmerging_in_mbp.ll | 6 +-
.../CodeGen/AArch64/typepromotion-cost.ll | 49 ++-
.../test/CodeGen/PowerPC/ppc-ctr-dead-code.ll | 32 +-
llvm/test/CodeGen/X86/loop-search.ll | 27 +-
llvm/test/Transforms/LICM/hoist-phi.ll | 28 +-
llvm/test/Transforms/LICM/sinking.ll | 56 ++--
.../LoopIdiom/AArch64/byte-compare-index.ll | 47 +--
.../LoopIdiom/RISCV/byte-compare-index.ll | 47 +--
.../LoopStrengthReduce/AArch64/pr53625.ll | 67 ++--
.../AArch64/small-constant.ll | 40 +--
.../LoopUnroll/runtime-loop-multiple-exits.ll | 290 ++++--------------
llvm/test/Transforms/LoopUnroll/scevunroll.ll | 18 +-
.../LoopUnroll/unroll-after-peel.ll | 8 +-
...iting-with-phis-multiple-exiting-blocks.ll | 61 +---
.../LoopVectorize/early_exit_legality.ll | 4 +-
.../single_early_exit_live_outs.ll | 127 ++------
.../LoopVectorize/unsupported_early_exit.ll | 53 +++-
.../inject-invariant-conditions.ll | 4 +-
.../2008-07-13-InfLoopMiscompile.ll | 2 +-
.../SimplifyCFG/X86/SpeculativeExec.ll | 2 +-
.../SimplifyCFG/X86/switch_to_lookup_table.ll | 5 +-
.../fold-branch-to-common-dest-phis.ll | 125 +++++---
.../SimplifyCFG/fold-branch-to-common-dest.ll | 92 +++---
.../Transforms/SimplifyCFG/hoist-dbgvalue.ll | 2 +-
.../SimplifyCFG/speculate-derefable-load.ll | 2 +-
.../SimplifyCFG/switch-on-const-select.ll | 2 +-
28 files changed, 606 insertions(+), 728 deletions(-)
diff --git a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
index a75f29000ca18..7304bdaab7f88 100644
--- a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -1164,6 +1164,9 @@ static void cloneInstructionsIntoPredecessorBlockAndUpdateSSAUses(
// Note that we expect to be in a block-closed SSA form for this to work!
for (Use &U : make_early_inc_range(BonusInst.uses())) {
auto *UI = cast<Instruction>(U.getUser());
+ // Avoid dangling select instructions
+ if (!UI->getParent())
+ continue;
auto *PN = dyn_cast<PHINode>(UI);
if (!PN) {
assert(UI->getParent() == BB && BonusInst.comesBefore(UI) &&
@@ -3909,7 +3912,8 @@ shouldFoldCondBranchesToCommonDestination(BranchInst *BI, BranchInst *PBI,
static bool performBranchToCommonDestFolding(BranchInst *BI, BranchInst *PBI,
DomTreeUpdater *DTU,
MemorySSAUpdater *MSSAU,
- const TargetTransformInfo *TTI) {
+ const TargetTransformInfo *TTI,
+ SmallDenseMap<PHINode*, SelectInst*, 8> &InsertNewPHIs) {
BasicBlock *BB = BI->getParent();
BasicBlock *PredBlock = PBI->getParent();
@@ -3996,6 +4000,28 @@ static bool performBranchToCommonDestFolding(BranchInst *BI, BranchInst *PBI,
ValueToValueMapTy VMap; // maps original values to cloned values
cloneInstructionsIntoPredecessorBlockAndUpdateSSAUses(BB, PredBlock, VMap);
+ if (!InsertNewPHIs.empty()) {
+ // Fixup PHINode in the commong successor
+ for (PHINode &PN : CommonSucc->phis()) {
+ auto It = InsertNewPHIs.find(&PN);
+ if (It != InsertNewPHIs.end() && It->first == &PN) {
+ Instruction *SI = It->second;
+ // Oprands might have been promoted to bonous inst
+ RemapInstruction(SI, VMap,
+ RF_NoModuleLevelChanges | RF_IgnoreMissingLocals);
+ // Insert SelectInst as the new PHINode incoming value
+ SI->insertBefore(PredBlock->getTerminator()->getIterator());
+ // Fix PHINode
+ PN.removeIncomingValue(PredBlock);
+ PN.addIncoming(SI, PredBlock);
+ // Remove map entry
+ InsertNewPHIs.erase(It);
+ }
+ }
+ // Cleanup dangling SelectInst
+ for (SelectInst *SI : InsertNewPHIs.values())
+ delete SI;
+ }
Module *M = BB->getModule();
@@ -4053,15 +4079,48 @@ bool llvm::foldBranchToCommonDest(BranchInst *BI, DomTreeUpdater *DTU,
// With which predecessors will we want to deal with?
SmallVector<BasicBlock *, 8> Preds;
+ struct InsertPointTy {
+ InstructionCost Cost;
+ Value *TValue; // True Value
+ Value *FValue; // False Value
+ PHINode *Phi;
+ };
+ SmallDenseMap<BranchInst*, SmallVector<InsertPointTy, 8>, 8> InsertPts;
for (BasicBlock *PredBlock : predecessors(BB)) {
BranchInst *PBI = dyn_cast<BranchInst>(PredBlock->getTerminator());
- // Check that we have two conditional branches. If there is a PHI node in
- // the common successor, verify that the same value flows in from both
- // blocks.
- if (!PBI || PBI->isUnconditional() || !safeToMergeTerminators(BI, PBI))
+ // Check that we have two conditional branches.
+ if (!PBI || PBI->isUnconditional())
continue;
+ // If there is a PHI node in the common successor, verify that the same value flows in from both
+ // blocks. Otherwise, check whether we can create a SelectInst to combine the incoming values
+ if (!safeToMergeTerminators(BI, PBI)) {
+ if (BI == PBI)
+ continue;
+ for (BasicBlock *Succ : BI->successors()) {
+ if (llvm::is_contained(PBI->successors(), Succ)) {
+ for (PHINode &Phi : Succ->phis()) {
+ Value *IV0 = Phi.getIncomingValueForBlock(BB);
+ Value *IV1 = Phi.getIncomingValueForBlock(PredBlock);
+ InstructionCost PCost;
+ if (TTI) {
+ PCost = TTI->getCmpSelInstrCost(Instruction::Select, Phi.getType(),
+ CmpInst::makeCmpResultType(Phi.getType()),
+ CmpInst::BAD_ICMP_PREDICATE, CostKind);
+ }
+ auto &IP = InsertPts[PBI];
+ if (PBI->getSuccessor(0) == BB)
+ IP.emplace_back(InsertPointTy{PCost, IV0, IV1, &Phi});
+ else
+ IP.emplace_back(InsertPointTy{PCost, IV1, IV0, &Phi});
+ }
+ }
+ }
+ if (InsertPts.empty())
+ continue;
+ }
+
// Determine if the two branches share a common destination.
BasicBlock *CommonSucc;
Instruction::BinaryOps Opc;
@@ -4080,6 +4139,9 @@ bool llvm::foldBranchToCommonDest(BranchInst *BI, DomTreeUpdater *DTU,
!isa<CmpInst>(PBI->getCondition())))
Cost += TTI->getArithmeticInstrCost(Instruction::Xor, Ty, CostKind);
+ for (auto const &InsertPoints : InsertPts.values())
+ for (auto &InsertInfo : InsertPoints)
+ Cost += InsertInfo.Cost;
if (Cost > BranchFoldThreshold)
continue;
}
@@ -4145,7 +4207,15 @@ bool llvm::foldBranchToCommonDest(BranchInst *BI, DomTreeUpdater *DTU,
// Ok, we have the budget. Perform the transformation.
for (BasicBlock *PredBlock : Preds) {
auto *PBI = cast<BranchInst>(PredBlock->getTerminator());
- return performBranchToCommonDestFolding(BI, PBI, DTU, MSSAU, TTI);
+ SmallDenseMap<PHINode *, SelectInst *, 8> newPhis;
+ if (InsertPts.contains(PBI)) {
+ Value *PC = PBI->getCondition();
+ for (auto const InsertInfo: InsertPts[PBI]) {
+ SelectInst *newPhi = SelectInst::Create(PC, InsertInfo.TValue, InsertInfo.FValue);
+ newPhis.insert(std::make_pair(InsertInfo.Phi, newPhi));
+ }
+ }
+ return performBranchToCommonDestFolding(BI, PBI, DTU, MSSAU, TTI, newPhis);
}
return false;
}
diff --git a/llvm/test/CodeGen/AArch64/rm_redundant_cmp.ll b/llvm/test/CodeGen/AArch64/rm_redundant_cmp.ll
index 9e3bb8ce8efc0..3240ef0a40ddc 100644
--- a/llvm/test/CodeGen/AArch64/rm_redundant_cmp.ll
+++ b/llvm/test/CodeGen/AArch64/rm_redundant_cmp.ll
@@ -49,16 +49,16 @@ define void @test_i16_2cmp_signed_2() {
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: adrp x8, :got:cost_s_i8_i16
; CHECK-NEXT: ldr x8, [x8, :got_lo12:cost_s_i8_i16]
-; CHECK-NEXT: ldrsh w9, [x8, #2]
-; CHECK-NEXT: ldrsh w10, [x8, #4]
-; CHECK-NEXT: cmp w9, w10
-; CHECK-NEXT: b.gt .LBB1_2
-; CHECK-NEXT: // %bb.1: // %if.else
-; CHECK-NEXT: mov w9, w10
-; CHECK-NEXT: b.ge .LBB1_3
-; CHECK-NEXT: .LBB1_2: // %if.end8.sink.split
+; CHECK-NEXT: ldrh w10, [x8, #2]
+; CHECK-NEXT: ldrh w11, [x8, #4]
+; CHECK-NEXT: sxth w9, w10
+; CHECK-NEXT: cmp w9, w11, sxth
+; CHECK-NEXT: csel w9, w10, w11, gt
+; CHECK-NEXT: cmp w10, w11
+; CHECK-NEXT: b.eq .LBB1_2
+; CHECK-NEXT: // %bb.1: // %if.end8.sink.split
; CHECK-NEXT: strh w9, [x8]
-; CHECK-NEXT: .LBB1_3: // %if.end8
+; CHECK-NEXT: .LBB1_2: // %if.end8
; CHECK-NEXT: ret
entry:
%0 = load i16, ptr getelementptr inbounds (%struct.s_signed_i16, ptr @cost_s_i8_i16, i64 0, i32 1), align 2
@@ -125,13 +125,11 @@ define void @test_i16_2cmp_unsigned_2() {
; CHECK-NEXT: ldrh w9, [x8, #2]
; CHECK-NEXT: ldrh w10, [x8, #4]
; CHECK-NEXT: cmp w9, w10
-; CHECK-NEXT: b.hi .LBB3_2
-; CHECK-NEXT: // %bb.1: // %if.else
-; CHECK-NEXT: mov w9, w10
-; CHECK-NEXT: b.hs .LBB3_3
-; CHECK-NEXT: .LBB3_2: // %if.end8.sink.split
+; CHECK-NEXT: csel w9, w9, w10, hi
+; CHECK-NEXT: b.eq .LBB3_2
+; CHECK-NEXT: // %bb.1: // %if.end8.sink.split
; CHECK-NEXT: strh w9, [x8]
-; CHECK-NEXT: .LBB3_3: // %if.end8
+; CHECK-NEXT: .LBB3_2: // %if.end8
; CHECK-NEXT: ret
entry:
%0 = load i16, ptr getelementptr inbounds (%struct.s_unsigned_i16, ptr @cost_u_i16, i64 0, i32 1), align 2
@@ -204,16 +202,16 @@ define void @test_i8_2cmp_signed_2() {
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: adrp x8, :got:cost_s
; CHECK-NEXT: ldr x8, [x8, :got_lo12:cost_s]
-; CHECK-NEXT: ldrsb w9, [x8, #1]
-; CHECK-NEXT: ldrsb w10, [x8, #2]
-; CHECK-NEXT: cmp w9, w10
-; CHECK-NEXT: b.gt .LBB5_2
-; CHECK-NEXT: // %bb.1: // %if.else
-; CHECK-NEXT: mov w9, w10
-; CHECK-NEXT: b.ge .LBB5_3
-; CHECK-NEXT: .LBB5_2: // %if.end8.sink.split
+; CHECK-NEXT: ldrb w10, [x8, #1]
+; CHECK-NEXT: ldrb w11, [x8, #2]
+; CHECK-NEXT: sxtb w9, w10
+; CHECK-NEXT: cmp w9, w11, sxtb
+; CHECK-NEXT: csel w9, w10, w11, gt
+; CHECK-NEXT: cmp w10, w11
+; CHECK-NEXT: b.eq .LBB5_2
+; CHECK-NEXT: // %bb.1: // %if.end8.sink.split
; CHECK-NEXT: strb w9, [x8]
-; CHECK-NEXT: .LBB5_3: // %if.end8
+; CHECK-NEXT: .LBB5_2: // %if.end8
; CHECK-NEXT: ret
entry:
%0 = load i8, ptr getelementptr inbounds (%struct.s_signed_i8, ptr @cost_s, i64 0, i32 1), align 2
@@ -280,13 +278,11 @@ define void @test_i8_2cmp_unsigned_2() {
; CHECK-NEXT: ldrb w9, [x8, #1]
; CHECK-NEXT: ldrb w10, [x8, #2]
; CHECK-NEXT: cmp w9, w10
-; CHECK-NEXT: b.hi .LBB7_2
-; CHECK-NEXT: // %bb.1: // %if.else
-; CHECK-NEXT: mov w9, w10
-; CHECK-NEXT: b.hs .LBB7_3
-; CHECK-NEXT: .LBB7_2: // %if.end8.sink.split
+; CHECK-NEXT: csel w9, w9, w10, hi
+; CHECK-NEXT: b.eq .LBB7_2
+; CHECK-NEXT: // %bb.1: // %if.end8.sink.split
; CHECK-NEXT: strb w9, [x8]
-; CHECK-NEXT: .LBB7_3: // %if.end8
+; CHECK-NEXT: .LBB7_2: // %if.end8
; CHECK-NEXT: ret
entry:
%0 = load i8, ptr getelementptr inbounds (%struct.s_unsigned_i8, ptr @cost_u_i8, i64 0, i32 1), align 2
diff --git a/llvm/test/CodeGen/AArch64/tailmerging_in_mbp.ll b/llvm/test/CodeGen/AArch64/tailmerging_in_mbp.ll
index 675380787af4d..75791f6589128 100644
--- a/llvm/test/CodeGen/AArch64/tailmerging_in_mbp.ll
+++ b/llvm/test/CodeGen/AArch64/tailmerging_in_mbp.ll
@@ -2,10 +2,10 @@
; CHECK-LABEL: test:
; CHECK-LABEL: %cond.false12.i
-; CHECK: b.gt
-; CHECK-NEXT: LBB0_8:
+; CHECK: b.le
+; CHECK-LABEL: LBB0_9:
; CHECK-NEXT: mov x8, x9
-; CHECK-NEXT: LBB0_9:
+; CHECK-NEXT: LBB0_10:
define i64 @test(i64 %n, ptr %a, ptr %b, ptr %c, ptr %d, ptr %e, ptr %f) {
entry:
%cmp28 = icmp sgt i64 %n, 1
diff --git a/llvm/test/CodeGen/AArch64/typepromotion-cost.ll b/llvm/test/CodeGen/AArch64/typepromotion-cost.ll
index 3aed4cb671c02..945588e5d2824 100644
--- a/llvm/test/CodeGen/AArch64/typepromotion-cost.ll
+++ b/llvm/test/CodeGen/AArch64/typepromotion-cost.ll
@@ -6,41 +6,28 @@
define i32 @needless_promotion(ptr nocapture noundef readonly %S, i64 noundef %red_cost) {
; CHECK-O2-LABEL: needless_promotion:
; CHECK-O2: // %bb.0: // %entry
-; CHECK-O2-NEXT: ldrsh w8, [x0, #4]
-; CHECK-O2-NEXT: tbnz w8, #31, .LBB0_3
-; CHECK-O2-NEXT: // %bb.1: // %lor.rhs
-; CHECK-O2-NEXT: cbz x1, .LBB0_5
-; CHECK-O2-NEXT: // %bb.2:
-; CHECK-O2-NEXT: mov w9, #2
-; CHECK-O2-NEXT: b .LBB0_4
-; CHECK-O2-NEXT: .LBB0_3:
-; CHECK-O2-NEXT: mov w9, #1
-; CHECK-O2-NEXT: .LBB0_4: // %lor.end.sink.split
-; CHECK-O2-NEXT: cmp w8, w9
-; CHECK-O2-NEXT: cset w0, eq
-; CHECK-O2-NEXT: ret
-; CHECK-O2-NEXT: .LBB0_5:
-; CHECK-O2-NEXT: mov w0, wzr
+; CHECK-O2-NEXT: ldrsh w9, [x0, #4]
+; CHECK-O2-NEXT: mov w8, #1 // =0x1
+; CHECK-O2-NEXT: cmp w9, #0
+; CHECK-O2-NEXT: cinc w8, w8, ge
+; CHECK-O2-NEXT: cmp w8, w9, uxth
+; CHECK-O2-NEXT: cset w8, eq
+; CHECK-O2-NEXT: cmp x1, #0
+; CHECK-O2-NEXT: ccmn w9, #1, #4, eq
+; CHECK-O2-NEXT: csel w0, wzr, w8, gt
; CHECK-O2-NEXT: ret
;
; CHECK-O3-LABEL: needless_promotion:
; CHECK-O3: // %bb.0: // %entry
-; CHECK-O3-NEXT: ldrsh w8, [x0, #4]
-; CHECK-O3-NEXT: tbnz w8, #31, .LBB0_3
-; CHECK-O3-NEXT: // %bb.1: // %lor.rhs
-; CHECK-O3-NEXT: cbz x1, .LBB0_4
-; CHECK-O3-NEXT: // %bb.2:
-; CHECK-O3-NEXT: mov w9, #2
-; CHECK-O3-NEXT: cmp w8, w9
-; CHECK-O3-NEXT: cset w0, eq
-; CHECK-O3-NEXT: ret
-; CHECK-O3-NEXT: .LBB0_3:
-; CHECK-O3-NEXT: mov w9, #1
-; CHECK-O3-NEXT: cmp w8, w9
-; CHECK-O3-NEXT: cset w0, eq
-; CHECK-O3-NEXT: ret
-; CHECK-O3-NEXT: .LBB0_4:
-; CHECK-O3-NEXT: mov w0, wzr
+; CHECK-O3-NEXT: ldrsh w9, [x0, #4]
+; CHECK-O3-NEXT: mov w8, #1 // =0x1
+; CHECK-O3-NEXT: cmp w9, #0
+; CHECK-O3-NEXT: cinc w8, w8, ge
+; CHECK-O3-NEXT: cmp w8, w9, uxth
+; CHECK-O3-NEXT: cset w8, eq
+; CHECK-O3-NEXT: cmp x1, #0
+; CHECK-O3-NEXT: ccmn w9, #1, #4, eq
+; CHECK-O3-NEXT: csel w0, wzr, w8, gt
; CHECK-O3-NEXT: ret
entry:
%ident = getelementptr inbounds %struct.S, ptr %S, i64 0, i32 1
diff --git a/llvm/test/CodeGen/PowerPC/ppc-ctr-dead-code.ll b/llvm/test/CodeGen/PowerPC/ppc-ctr-dead-code.ll
index bb8337d237f51..aceac8bb69b82 100644
--- a/llvm/test/CodeGen/PowerPC/ppc-ctr-dead-code.ll
+++ b/llvm/test/CodeGen/PowerPC/ppc-ctr-dead-code.ll
@@ -8,25 +8,27 @@
define signext i32 @limit_loop(i32 signext %iters, ptr nocapture readonly %vec, i32 signext %limit) local_unnamed_addr {
; CHECK-LABEL: limit_loop:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: mr 6, 3
-; CHECK-NEXT: li 3, 0
-; CHECK-NEXT: cmpwi 6, 0
-; CHECK-NEXT: blelr 0
+; CHECK-NEXT: cmpwi 3, 0
+; CHECK-NEXT: ble 0, .LBB0_4
; CHECK-NEXT: # %bb.1: # %for.body.preheader
-; CHECK-NEXT: mtctr 6
; CHECK-NEXT: addi 4, 4, -4
-; CHECK-NEXT: b .LBB0_3
-; CHECK-NEXT: .p2align 4
-; CHECK-NEXT: .LBB0_2: # %for.cond
-; CHECK-NEXT: #
-; CHECK-NEXT: bdzlr
-; CHECK-NEXT: .LBB0_3: # %for.body
+; CHECK-NEXT: li 6, 1
+; CHECK-NEXT: .p2align 5
+; CHECK-NEXT: .LBB0_2: # %for.body
; CHECK-NEXT: #
-; CHECK-NEXT: lwzu 6, 4(4)
-; CHECK-NEXT: cmpw 6, 5
-; CHECK-NEXT: blt 0, .LBB0_2
-; CHECK-NEXT: # %bb.4:
+; CHECK-NEXT: lwzu 7, 4(4)
+; CHECK-NEXT: cmpd 1, 6, 3
+; CHECK-NEXT: addi 6, 6, 1
+; CHECK-NEXT: cmpw 7, 5
+; CHECK-NEXT: crand 20, 0, 4
+; CHECK-NEXT: bc 12, 20, .LBB0_2
+; CHECK-NEXT: # %bb.3: # %cleanup.loopexit
; CHECK-NEXT: li 3, 1
+; CHECK-NEXT: isellt 3, 0, 3
+; CHECK-NEXT: clrldi 3, 3, 32
+; CHECK-NEXT: blr
+; CHECK-NEXT: .LBB0_4:
+; CHECK-NEXT: li 3, 0
; CHECK-NEXT: blr
entry:
%cmp5 = icmp sgt i32 %iters, 0
diff --git a/llvm/test/CodeGen/X86/loop-search.ll b/llvm/test/CodeGen/X86/loop-search.ll
index 0d5f97d21fb3a..4fe5c48fcd26a 100644
--- a/llvm/test/CodeGen/X86/loop-search.ll
+++ b/llvm/test/CodeGen/X86/loop-search.ll
@@ -10,24 +10,29 @@ define zeroext i1 @search(i32 %needle, ptr nocapture readonly %haystack, i32 %co
; CHECK-NEXT: testl %edx, %edx
; CHECK-NEXT: jle LBB0_5
; CHECK-NEXT: ## %bb.1: ## %for.body.preheader
-; CHECK-NEXT: movslq %edx, %rax
-; CHECK-NEXT: xorl %ecx, %ecx
+; CHECK-NEXT: movslq %edx, %rcx
+; CHECK-NEXT: movl $1, %edx
; CHECK-NEXT: .p2align 4
; CHECK-NEXT: LBB0_2: ## %for.body
; CHECK-NEXT: ## =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: cmpl %edi, (%rsi,%rcx,4)
-; CHECK-NEXT: je LBB0_6
-; CHECK-NEXT: ## %bb.3: ## %for.cond
+; CHECK-NEXT: movl -4(%rsi,%rdx,4), %r8d
+; CHECK-NEXT: cmpl %edi, %r8d
+; CHECK-NEXT: sete %al
+; CHECK-NEXT: negb %al
+; CHECK-NEXT: cmpl %edi, %r8d
+; CHECK-NEXT: je LBB0_4
+; CHECK-NEXT: ## %bb.3: ## %for.body
; CHECK-NEXT: ## in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT: incq %rcx
-; CHECK-NEXT: cmpq %rax, %rcx
+; CHECK-NEXT: cmpq %rcx, %rdx
+; CHECK-NEXT: leaq 1(%rdx), %rdx
; CHECK-NEXT: jl LBB0_2
-; CHECK-NEXT: LBB0_5:
-; CHECK-NEXT: xorl %eax, %eax
+; CHECK-NEXT: LBB0_4: ## %cleanup
+; CHECK-NEXT: andb $1, %al
; CHECK-NEXT: ## kill: def $al killed $al killed $eax
; CHECK-NEXT: retq
-; CHECK-NEXT: LBB0_6:
-; CHECK-NEXT: movb $1, %al
+; CHECK-NEXT: LBB0_5:
+; CHECK-NEXT: xorl %eax, %eax
+; CHECK-NEXT: andb $1, %al
; CHECK-NEXT: ## kill: def $al killed $al killed $eax
; CHECK-NEXT: retq
entry:
diff --git a/llvm/test/Transforms/LICM/hoist-phi.ll b/llvm/test/Transforms/LICM/hoist-phi.ll
index bf999b98a1dac..3cedb14edfa30 100644
--- a/llvm/test/Transforms/LICM/hoist-phi.ll
+++ b/llvm/test/Transforms/LICM/hoist-phi.ll
@@ -629,16 +629,16 @@ define void @triangle_phi_loopexit(i32 %x, ptr %p) {
; CHECK-DISABLED-NEXT: entry:
; CHECK-DISABLED-NEXT: [[ADD:%.*]] = add i32 [[X:%.*]], 1
; CHECK-DISABLED-NEXT: [[CMP1:%.*]] = icmp sgt i32 [[X]], 0
+; CHECK-DISABLED-NEXT: [[CMP1_NOT:%.*]] = xor i1 [[CMP1]], true
; CHECK-DISABLED-NEXT: [[CMP2:%.*]] = icmp sgt i32 10, [[ADD]]
+; CHECK-DISABLED-NEXT: [[TMP0:%.*]] = select i1 [[CMP1]], i32 [[ADD]], i32 [[X]]
+; CHECK-DISABLED-NEXT: [[OR_COND:%.*]] = or i1 [[CMP1_NOT]], [[CMP2]]
+; CHECK-DISABLED-NEXT: [[CMP3:%.*]] = icmp ne i32 [[TMP0]], 0
; CHECK-DISABLED-NEXT: br label [[LOOP:%.*]]
; CHECK-DISABLED: loop:
-; CHECK-DISABLED-NEXT: br i1 [[CMP1]], label [[IF:%.*]], label [[THEN:%.*]]
-; CHECK-DISABLED: if:
-; CHECK-DISABLED-NEXT: br i1 [[CMP2]], label [[THEN]], label [[END:%.*]]
+; CHECK-DISABLED-NEXT: br i1 [[OR_COND]], label [[THEN:%.*]], label [[END:%.*]]
; CHECK-DISABLED: then:
-; CHECK-DISABLED-NEXT: [[PHI:%.*]] = phi i32 [ [[ADD]], [[IF]] ], [ [[X]], [[LOOP]] ]
-; CHECK-DISABLED-NEXT: store i32 [[PHI]], ptr [[P:%.*]], align 4
-; CHECK-DISABLED-NEXT: [[CMP3:%.*]] = icmp ne i32 [[PHI]], 0
+; CHECK-DISABLED-NEXT: store i32 [[TMP0]], ptr [[P:%.*]], align 4
; CHECK-DISABLED-NEXT: br i1 [[CMP3]], label [[LOOP]], label [[END]]
; CHECK-DISABLED: end:
; CHECK-DISABLED-NEXT: ret void
@@ -647,20 +647,16 @@ define void @triangle_phi_loopexit(i32 %x, ptr %p) {
; CHECK-ENABLED-NEXT: entry:
; CHECK-ENABLED-NEXT: [[ADD:%.*]] = add i32 [[X:%.*]], 1
; CHECK-ENABLED-NEXT: [[CMP1:%.*]] = icmp sgt i32 [[X]], 0
+; CHECK-ENABLED-NEXT: [[CMP1_NOT:%.*]] = xor i1 [[CMP1]], true
; CHECK-ENABLED-NEXT: [[CMP2:%.*]] = icmp sgt i32 10, [[ADD]]
-; CHECK-ENABLED-NEXT: br i1 [[CMP1]], label [[IF_LICM:%.*]], label [[THEN_LICM:%.*]]
-; CHECK-ENABLED: if.licm:
-; CHECK-ENABLED-NEXT: br label [[THEN_LICM]]
-; CHECK-ENABLED: then.licm:
-; CHECK-ENABLED-NEXT: [[PHI:%.*]] = phi i32 [ [[ADD]], [[IF_LICM]] ], [ [[X]], [[ENTRY:%.*]] ]
-; CHECK-ENABLED-NEXT: [[CMP3:%.*]] = icmp ne i32 [[PHI]], 0
+; CHECK-ENABLED-NEXT: [[TMP0:%.*]] = select i1 [[CMP1]], i32 [[ADD]], i32 [[X]]
+; CHECK-ENABLED-NEXT: [[OR_COND:%.*]] = or i1 [[CMP1_NOT]], [[CMP2]]
+; CHECK-ENABLED-NEXT: [[CMP3:%.*]] = icmp ne i32 [[TMP0]], 0
; CHECK-ENABLED-NEXT: br label [[LOOP:%.*]]
; CHECK-ENABLED: loop:
-; CHECK-ENABLED-NEXT: br i1 [[CMP1]], label [[IF:%.*]], label [[THEN:%.*]]
-; CHECK-ENABLED: if:
-; CHECK-ENABLED-NEXT: br i1 [[CMP2]], label [[THEN]], label [[END:%.*]]
+; CHECK-ENABLED-NEXT: br i1 [[OR_COND]], label [[THEN:%.*]], label [[END:%.*]]
; CHECK-ENABLED: then:
-; CHECK-ENABLED-NEXT: store i32 [[PHI]], ptr [[P:%.*]], align 4
+; CHECK-ENABLED-NEXT: store i32 [[TMP0]], ptr [[P:%.*]], align 4
; CHECK-ENABLED-NEXT: br i1 [[CMP3]], label [[LOOP]], label [[END]]
; CHECK-ENABLED: end:
; CHECK-ENABLED-NEXT: ret void
diff --git a/llvm/test/Transforms/LICM/sinking.ll b/llvm/test/Transforms/LICM/sinking.ll
index e7ac07b50625a..5d369afcd2578 100644
--- a/llvm/test/Transforms/LICM/sinking.ll
+++ b/llvm/test/Transforms/LICM/sinking.ll
@@ -80,7 +80,7 @@ define double @test2c(ptr %P) {
; CHECK-NEXT: call void @foo()
; CHECK-NEXT: br i1 true, label [[LOOP]], label [[OUT:%.*]]
; CHECK: Out:
-; CHECK-NEXT: [[A_LE:%.*]] = load double, ptr [[P:%.*]], align 8, !invariant.load !0
+; CHECK-NEXT: [[A_LE:%.*]] = load double, ptr [[P:%.*]], align 8, !invariant.load [[META0:![0-9]+]]
; CHECK-NEXT: ret double [[A_LE]]
;
br label %Loop
@@ -535,23 +535,16 @@ define i32 @test14(i32 %N, i32 %N2, i1 %C) {
; CHECK-NEXT: Entry:
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: Loop:
-; CHECK-NEXT: [[N_ADDR_0_PN:%.*]] = phi i32 [ [[DEC:%.*]], [[CONTLOOP:%.*]] ], [ [[N:%.*]], [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[N_ADDR_0_PN:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[DEC:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[DEC]] = add i32 [[N_ADDR_0_PN]], -1
-; CHECK-NEXT: br i1 [[C:%.*]], label [[CONTLOOP]], label [[OUT12_SPLIT_LOOP_EXIT1:%.*]]
-; CHECK: ContLoop:
; CHECK-NEXT: [[TMP_1:%.*]] = icmp ne i32 [[N_ADDR_0_PN]], 1
-; CHECK-NEXT: br i1 [[TMP_1]], label [[LOOP]], label [[OUT12_SPLIT_LOOP_EXIT:%.*]]
-; CHECK: Out12.split.loop.exit:
-; CHECK-NEXT: [[N_ADDR_0_PN_LCSSA4:%.*]] = phi i32 [ [[N_ADDR_0_PN]], [[CONTLOOP]] ]
-; CHECK-NEXT: [[SINK_MUL_LE3:%.*]] = mul i32 [[N]], [[N_ADDR_0_PN_LCSSA4]]
-; CHECK-NEXT: br label [[OUT12:%.*]]
-; CHECK: Out12.split.loop.exit1:
+; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[C:%.*]], i1 [[TMP_1]], i1 false
+; CHECK-NEXT: br i1 [[OR_COND]], label [[LOOP]], label [[OUT12:%.*]]
+; CHECK: Out12:
; CHECK-NEXT: [[N_ADDR_0_PN_LCSSA:%.*]] = phi i32 [ [[N_ADDR_0_PN]], [[LOOP]] ]
; CHECK-NEXT: [[SINK_MUL_LE:%.*]] = mul i32 [[N]], [[N_ADDR_0_PN_LCSSA]]
; CHECK-NEXT: [[SINK_SUB_LE:%.*]] = sub i32 [[SINK_MUL_LE]], [[N]]
-; CHECK-NEXT: br label [[OUT12]]
-; CHECK: Out12:
-; CHECK-NEXT: [[TMP:%.*]] = phi i32 [ [[SINK_MUL_LE3]], [[OUT12_SPLIT_LOOP_EXIT]] ], [ [[SINK_SUB_LE]], [[OUT12_SPLIT_LOOP_EXIT1]] ]
+; CHECK-NEXT: [[TMP:%.*]] = select i1 [[C]], i32 [[SINK_MUL_LE]], i32 [[SINK_SUB_LE]]
; CHECK-NEXT: ret i32 [[TMP]]
;
Entry:
@@ -581,24 +574,17 @@ define i32 @test15(i32 %N, i32 %N2, i1 %C) {
; CHECK-NEXT: Entry:
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: Loop:
-; CHECK-NEXT: [[N_ADDR_0_PN:%.*]] = phi i32 [ [[DEC:%.*]], [[CONTLOOP:%.*]] ], [ [[N:%.*]], [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[N_ADDR_0_PN:%.*]] = phi i32 [ [[N:%.*]], [[ENTRY:%.*]] ], [ [[DEC:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[DEC]] = add i32 [[N_ADDR_0_PN]], -1
-; CHECK-NEXT: br i1 [[C:%.*]], label [[CONTLOOP]], label [[OUT12_SPLIT_LOOP_EXIT1:%.*]]
-; CHECK: ContLoop:
; CHECK-NEXT: [[TMP_1:%.*]] = icmp ne i32 [[N_ADDR_0_PN]], 1
-; CHECK-NEXT: br i1 [[TMP_1]], label [[LOOP]], label [[OUT12_SPLIT_LOOP_EXIT:%.*]]
-; CHECK: Out12.split.loop.exit:
-; CHECK-NEXT: [[N_ADDR_0_PN_LCSSA5:%.*]] = phi i32 [ [[N_ADDR_0_PN]], [[CONTLOOP]] ]
+; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[C:%.*]], i1 [[TMP_1]], i1 false
+; CHECK-NEXT: br i1 [[OR_COND]], label [[LOOP]], label [[OUT12:%.*]]
+; CHECK: Out12:
+; CHECK-NEXT: [[N_ADDR_0_PN_LCSSA5:%.*]] = phi i32 [ [[N_ADDR_0_PN]], [[LOOP]] ]
; CHECK-NEXT: [[SINK_MUL_LE4:%.*]] = mul i32 [[N]], [[N_ADDR_0_PN_LCSSA5]]
+; CHECK-NEXT: [[SINK_SUB_LE:%.*]] = sub i32 [[SINK_MUL_LE4]], [[N]]
; CHECK-NEXT: [[SINK_SUB2_LE:%.*]] = sub i32 [[SINK_MUL_LE4]], [[N2:%.*]]
-; CHECK-NEXT: br label [[OUT12:%.*]]
-; CHECK: Out12.split.loop.exit1:
-; CHECK-NEXT: [[N_ADDR_0_PN_LCSSA:%.*]] = phi i32 [ [[N_ADDR_0_PN]], [[LOOP]] ]
-; CHECK-NEXT: [[SINK_MUL_LE:%.*]] = mul i32 [[N]], [[N_ADDR_0_PN_LCSSA]]
-; CHECK-NEXT: [[SINK_SUB_LE:%.*]] = sub i32 [[SINK_MUL_LE]], [[N]]
-; CHECK-NEXT: br label [[OUT12]]
-; CHECK: Out12:
-; CHECK-NEXT: [[TMP:%.*]] = phi i32 [ [[SINK_SUB2_LE]], [[OUT12_SPLIT_LOOP_EXIT]] ], [ [[SINK_SUB_LE]], [[OUT12_SPLIT_LOOP_EXIT1]] ]
+; CHECK-NEXT: [[TMP:%.*]] = select i1 [[C]], i32 [[SINK_SUB2_LE]], i32 [[SINK_SUB_LE]]
; CHECK-NEXT: ret i32 [[TMP]]
;
Entry:
@@ -630,9 +616,9 @@ define i32 @test16(i1 %c, ptr %P, ptr %P2, i64 %V) {
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[LOOP_PH]] ], [ [[NEXT:%.*]], [[CONTLOOP:%.*]] ]
; CHECK-NEXT: [[L2:%.*]] = call i32 @getv()
; CHECK-NEXT: switch i32 [[L2]], label [[CONTLOOP]] [
-; CHECK-NEXT: i32 32, label [[OUT_SPLIT_LOOP_EXIT1:%.*]]
-; CHECK-NEXT: i32 46, label [[OUT_SPLIT_LOOP_EXIT1]]
-; CHECK-NEXT: i32 95, label [[OUT_SPLIT_LOOP_EXIT1]]
+; CHECK-NEXT: i32 32, label [[OUT_SPLIT_LOOP_EXIT1:%.*]]
+; CHECK-NEXT: i32 46, label [[OUT_SPLIT_LOOP_EXIT1]]
+; CHECK-NEXT: i32 95, label [[OUT_SPLIT_LOOP_EXIT1]]
; CHECK-NEXT: ]
; CHECK: ContLoop:
; CHECK-NEXT: [[NEXT]] = add nuw i64 [[IV]], 1
@@ -849,10 +835,10 @@ define void @test20(ptr %s, i1 %b, i32 %v1, i32 %v2) personality ptr @__CxxFrame
; CHECK-NEXT: br i1 [[B:%.*]], label [[TRY_CONT:%.*]], label [[WHILE_BODY:%.*]]
; CHECK: while.body:
; CHECK-NEXT: invoke void @may_throw()
-; CHECK-NEXT: to label [[WHILE_BODY2:%.*]] unwind label [[CATCH_DISPATCH:%.*]]
+; CHECK-NEXT: to label [[WHILE_BODY2:%.*]] unwind label [[CATCH_DISPATCH:%.*]]
; CHECK: while.body2:
; CHECK-NEXT: invoke void @may_throw2()
-; CHECK-NEXT: to label [[WHILE_COND]] unwind label [[CATCH_DISPATCH]]
+; CHECK-NEXT: to label [[WHILE_COND]] unwind label [[CATCH_DISPATCH]]
; CHECK: catch.dispatch:
; CHECK-NEXT: [[DOTLCSSA1:%.*]] = phi i32 [ [[SINKABLE]], [[WHILE_BODY]] ], [ [[SINKABLE2]], [[WHILE_BODY2]] ]
; CHECK-NEXT: [[CP:%.*]] = cleanuppad within none []
@@ -933,16 +919,16 @@ define void @test22(i1 %b, i32 %v1, i32 %v2) personality ptr @__CxxFrameHandler3
; CHECK-NEXT: br i1 [[B:%.*]], label [[TRY_CONT:%.*]], label [[WHILE_BODY:%.*]]
; CHECK: while.body:
; CHECK-NEXT: invoke void @may_throw()
-; CHECK-NEXT: to label [[WHILE_BODY2:%.*]] unwind label [[LPADBB:%.*]]
+; CHECK-NEXT: to label [[WHILE_BODY2:%.*]] unwind label [[LPADBB:%.*]]
; CHECK: while.body2:
; CHECK-NEXT: [[V:%.*]] = call i32 @getv()
; CHECK-NEXT: [[MUL:%.*]] = mul i32 [[V]], [[V2:%.*]]
; CHECK-NEXT: invoke void @may_throw2()
-; CHECK-NEXT: to label [[WHILE_COND]] unwind label [[LPADBB]]
+; CHECK-NEXT: to label [[WHILE_COND]] unwind label [[LPADBB]]
; CHECK: lpadBB:
; CHECK-NEXT: [[DOTLCSSA1:%.*]] = phi i32 [ 0, [[WHILE_BODY]] ], [ [[MUL]], [[WHILE_BODY2]] ]
; CHECK-NEXT: [[TMP0:%.*]] = landingpad { ptr, i32 }
-; CHECK-NEXT: catch ptr null
+; CHECK-NEXT: catch ptr null
; CHECK-NEXT: br label [[LPADBBSUCC1:%.*]]
; CHECK: lpadBBSucc1:
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/LoopIdiom/AArch64/byte-compare-index.ll b/llvm/test/Transforms/LoopIdiom/AArch64/byte-compare-index.ll
index 39037761c81bb..00cea91a67709 100644
--- a/llvm/test/Transforms/LoopIdiom/AArch64/byte-compare-index.ll
+++ b/llvm/test/Transforms/LoopIdiom/AArch64/byte-compare-index.ll
@@ -160,20 +160,21 @@ define i32 @compare_bytes_simple(ptr %a, ptr %b, i32 %len, i32 %extra, i32 %n) {
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP37:%.*]] = zext i32 [[MISMATCH_INDEX]] to i64
; LOOP-DEL-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP39:%.*]] = load i8, ptr [[TMP38]], align 1
; LOOP-DEL-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP41:%.*]] = load i8, ptr [[TMP40]], align 1
; LOOP-DEL-NEXT: [[TMP42:%.*]] = icmp eq i8 [[TMP39]], [[TMP41]]
-; LOOP-DEL-NEXT: br i1 [[TMP42]], label [[MISMATCH_LOOP_INC]], label [[WHILE_END]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP42]], true
; LOOP-DEL-NEXT: [[TMP43]] = add i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP44:%.*]] = icmp eq i32 [[TMP43]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP44]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP45:%.*]] = select i1 [[TMP42]], i32 [[N]], i32 [[MISMATCH_INDEX]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[DOTNOT]], [[TMP44]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
; LOOP-DEL: while.end:
-; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ], [ [[TMP45]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[RES:%.*]] = add i32 [[MISMATCH_RESULT]], [[EXTRA]]
; LOOP-DEL-NEXT: ret i32 [[RES]]
;
@@ -381,20 +382,21 @@ define i32 @compare_bytes_signed_wrap(ptr %a, ptr %b, i32 %len, i32 %n) {
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP37:%.*]] = zext i32 [[MISMATCH_INDEX]] to i64
; LOOP-DEL-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP39:%.*]] = load i8, ptr [[TMP38]], align 1
; LOOP-DEL-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP41:%.*]] = load i8, ptr [[TMP40]], align 1
; LOOP-DEL-NEXT: [[TMP42:%.*]] = icmp eq i8 [[TMP39]], [[TMP41]]
-; LOOP-DEL-NEXT: br i1 [[TMP42]], label [[MISMATCH_LOOP_INC]], label [[WHILE_END]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP42]], true
; LOOP-DEL-NEXT: [[TMP43]] = add nsw i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP44:%.*]] = icmp eq i32 [[TMP43]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP44]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP45:%.*]] = select i1 [[TMP42]], i32 [[N]], i32 [[MISMATCH_INDEX]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = select i1 [[DOTNOT]], i1 true, i1 [[TMP44]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
; LOOP-DEL: while.end:
-; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ], [ [[TMP45]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: ret i32 [[MISMATCH_RESULT]]
;
; NO-TRANSFORM-LABEL: define i32 @compare_bytes_signed_wrap(
@@ -608,20 +610,21 @@ define i32 @compare_bytes_simple_end_ne_found(ptr %a, ptr %b, ptr %c, ptr %d, i3
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX3:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX3:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP37:%.*]] = zext i32 [[MISMATCH_INDEX3]] to i64
; LOOP-DEL-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP39:%.*]] = load i8, ptr [[TMP38]], align 1
; LOOP-DEL-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP41:%.*]] = load i8, ptr [[TMP40]], align 1
; LOOP-DEL-NEXT: [[TMP42:%.*]] = icmp eq i8 [[TMP39]], [[TMP41]]
-; LOOP-DEL-NEXT: br i1 [[TMP42]], label [[MISMATCH_LOOP_INC]], label [[BYTE_COMPARE]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP42]], true
; LOOP-DEL-NEXT: [[TMP43]] = add i32 [[MISMATCH_INDEX3]], 1
; LOOP-DEL-NEXT: [[TMP44:%.*]] = icmp eq i32 [[TMP43]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP44]], label [[BYTE_COMPARE]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP46:%.*]] = select i1 [[TMP42]], i32 [[N]], i32 [[MISMATCH_INDEX3]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[DOTNOT]], [[TMP44]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[BYTE_COMPARE]], label [[MISMATCH_LOOP]]
; LOOP-DEL: byte.compare:
-; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX3]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ], [ [[TMP46]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP45:%.*]] = icmp eq i32 [[MISMATCH_RESULT]], [[N]]
; LOOP-DEL-NEXT: [[SPEC_SELECT:%.*]] = select i1 [[TMP45]], i32 [[N]], i32 [[MISMATCH_RESULT]]
; LOOP-DEL-NEXT: [[SPEC_SELECT4:%.*]] = select i1 [[TMP45]], ptr [[D]], ptr [[C]]
@@ -860,20 +863,21 @@ define i32 @compare_bytes_extra_cmp(ptr %a, ptr %b, i32 %len, i32 %n, i32 %x) {
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP43:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP37:%.*]] = zext i32 [[MISMATCH_INDEX]] to i64
; LOOP-DEL-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP39:%.*]] = load i8, ptr [[TMP38]], align 1
; LOOP-DEL-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP37]]
; LOOP-DEL-NEXT: [[TMP41:%.*]] = load i8, ptr [[TMP40]], align 1
; LOOP-DEL-NEXT: [[TMP42:%.*]] = icmp eq i8 [[TMP39]], [[TMP41]]
-; LOOP-DEL-NEXT: br i1 [[TMP42]], label [[MISMATCH_LOOP_INC]], label [[WHILE_END]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP42]], true
; LOOP-DEL-NEXT: [[TMP43]] = add i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP44:%.*]] = icmp eq i32 [[TMP43]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP44]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP45:%.*]] = select i1 [[TMP42]], i32 [[N]], i32 [[MISMATCH_INDEX]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[DOTNOT]], [[TMP44]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
; LOOP-DEL: while.end:
-; LOOP-DEL-NEXT: [[INC_LCSSA:%.*]] = phi i32 [ [[X]], [[ENTRY:%.*]] ], [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[INC_LCSSA:%.*]] = phi i32 [ [[X]], [[ENTRY:%.*]] ], [ [[N]], [[MISMATCH_VEC_LOOP_INC]] ], [ [[TMP36]], [[MISMATCH_VEC_LOOP_FOUND]] ], [ [[TMP45]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: ret i32 [[INC_LCSSA]]
;
; NO-TRANSFORM-LABEL: define i32 @compare_bytes_extra_cmp(
@@ -1040,7 +1044,8 @@ define void @compare_bytes_cleanup_block(ptr %src1, ptr %src2) {
; LOOP-DEL-NEXT: [[TMP2:%.*]] = load i8, ptr [[TMP1]], align 1
; LOOP-DEL-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP0]]
; LOOP-DEL-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP3]], align 1
-; LOOP-DEL-NEXT: [[TMP5:%.*]] = icmp ne i8 [[TMP2]], [[TMP4]]
+; LOOP-DEL-NEXT: [[TMP8:%.*]] = icmp eq i8 [[TMP2]], [[TMP4]]
+; LOOP-DEL-NEXT: [[TMP5:%.*]] = xor i1 [[TMP8]], true
; LOOP-DEL-NEXT: [[TMP6]] = add i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP7:%.*]] = icmp eq i32 [[TMP6]], 0
; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[TMP5]], [[TMP7]]
diff --git a/llvm/test/Transforms/LoopIdiom/RISCV/byte-compare-index.ll b/llvm/test/Transforms/LoopIdiom/RISCV/byte-compare-index.ll
index afc28cfda45aa..7060aded5012b 100644
--- a/llvm/test/Transforms/LoopIdiom/RISCV/byte-compare-index.ll
+++ b/llvm/test/Transforms/LoopIdiom/RISCV/byte-compare-index.ll
@@ -237,20 +237,21 @@ define i32 @compare_bytes_simple(ptr %a, ptr %b, i32 %len, i32 %n) {
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP29:%.*]] = zext i32 [[MISMATCH_INDEX]] to i64
; LOOP-DEL-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP31:%.*]] = load i8, ptr [[TMP30]], align 1
; LOOP-DEL-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP33:%.*]] = load i8, ptr [[TMP32]], align 1
; LOOP-DEL-NEXT: [[TMP34:%.*]] = icmp eq i8 [[TMP31]], [[TMP33]]
-; LOOP-DEL-NEXT: br i1 [[TMP34]], label [[MISMATCH_LOOP_INC]], label [[WHILE_END]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP34]], true
; LOOP-DEL-NEXT: [[TMP35]] = add i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP36:%.*]] = icmp eq i32 [[TMP35]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP36]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP38:%.*]] = select i1 [[TMP34]], i32 [[N]], i32 [[MISMATCH_INDEX]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[DOTNOT]], [[TMP36]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
; LOOP-DEL: while.end:
-; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ], [ [[TMP38]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: ret i32 [[MISMATCH_RESULT]]
;
; MASKED-LABEL: define i32 @compare_bytes_simple(
@@ -604,20 +605,21 @@ define i32 @compare_bytes_signed_wrap(ptr %a, ptr %b, i32 %len, i32 %n) {
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP29:%.*]] = zext i32 [[MISMATCH_INDEX]] to i64
; LOOP-DEL-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP31:%.*]] = load i8, ptr [[TMP30]], align 1
; LOOP-DEL-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP33:%.*]] = load i8, ptr [[TMP32]], align 1
; LOOP-DEL-NEXT: [[TMP34:%.*]] = icmp eq i8 [[TMP31]], [[TMP33]]
-; LOOP-DEL-NEXT: br i1 [[TMP34]], label [[MISMATCH_LOOP_INC]], label [[WHILE_END]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP34]], true
; LOOP-DEL-NEXT: [[TMP35]] = add nsw i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP36:%.*]] = icmp eq i32 [[TMP35]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP36]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP38:%.*]] = select i1 [[TMP34]], i32 [[N]], i32 [[MISMATCH_INDEX]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = select i1 [[DOTNOT]], i1 true, i1 [[TMP36]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
; LOOP-DEL: while.end:
-; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ], [ [[TMP38]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: ret i32 [[MISMATCH_RESULT]]
;
; MASKED-LABEL: define i32 @compare_bytes_signed_wrap(
@@ -1014,20 +1016,21 @@ define i32 @compare_bytes_simple_end_ne_found(ptr %a, ptr %b, ptr %c, ptr %d, i3
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX3:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX3:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP29:%.*]] = zext i32 [[MISMATCH_INDEX3]] to i64
; LOOP-DEL-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP31:%.*]] = load i8, ptr [[TMP30]], align 1
; LOOP-DEL-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP33:%.*]] = load i8, ptr [[TMP32]], align 1
; LOOP-DEL-NEXT: [[TMP34:%.*]] = icmp eq i8 [[TMP31]], [[TMP33]]
-; LOOP-DEL-NEXT: br i1 [[TMP34]], label [[MISMATCH_LOOP_INC]], label [[BYTE_COMPARE]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP34]], true
; LOOP-DEL-NEXT: [[TMP35]] = add i32 [[MISMATCH_INDEX3]], 1
; LOOP-DEL-NEXT: [[TMP36:%.*]] = icmp eq i32 [[TMP35]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP36]], label [[BYTE_COMPARE]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP38:%.*]] = select i1 [[TMP34]], i32 [[N]], i32 [[MISMATCH_INDEX3]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[DOTNOT]], [[TMP36]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[BYTE_COMPARE]], label [[MISMATCH_LOOP]]
; LOOP-DEL: byte.compare:
-; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX3]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_RESULT:%.*]] = phi i32 [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ], [ [[TMP38]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP37:%.*]] = icmp eq i32 [[MISMATCH_RESULT]], [[N]]
; LOOP-DEL-NEXT: [[SPEC_SELECT:%.*]] = select i1 [[TMP37]], i32 [[N]], i32 [[MISMATCH_RESULT]]
; LOOP-DEL-NEXT: [[SPEC_SELECT4:%.*]] = select i1 [[TMP37]], ptr [[D]], ptr [[C]]
@@ -1455,20 +1458,21 @@ define i32 @compare_bytes_extra_cmp(ptr %a, ptr %b, i32 %len, i32 %n, i32 %x) {
; LOOP-DEL: mismatch_loop_pre:
; LOOP-DEL-NEXT: br label [[MISMATCH_LOOP:%.*]]
; LOOP-DEL: mismatch_loop:
-; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP_INC:%.*]] ]
+; LOOP-DEL-NEXT: [[MISMATCH_INDEX:%.*]] = phi i32 [ [[TMP0]], [[MISMATCH_LOOP_PRE]] ], [ [[TMP35:%.*]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: [[TMP29:%.*]] = zext i32 [[MISMATCH_INDEX]] to i64
; LOOP-DEL-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP31:%.*]] = load i8, ptr [[TMP30]], align 1
; LOOP-DEL-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP29]]
; LOOP-DEL-NEXT: [[TMP33:%.*]] = load i8, ptr [[TMP32]], align 1
; LOOP-DEL-NEXT: [[TMP34:%.*]] = icmp eq i8 [[TMP31]], [[TMP33]]
-; LOOP-DEL-NEXT: br i1 [[TMP34]], label [[MISMATCH_LOOP_INC]], label [[WHILE_END]]
-; LOOP-DEL: mismatch_loop_inc:
+; LOOP-DEL-NEXT: [[DOTNOT:%.*]] = xor i1 [[TMP34]], true
; LOOP-DEL-NEXT: [[TMP35]] = add i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP36:%.*]] = icmp eq i32 [[TMP35]], [[N]]
-; LOOP-DEL-NEXT: br i1 [[TMP36]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
+; LOOP-DEL-NEXT: [[TMP38:%.*]] = select i1 [[TMP34]], i32 [[N]], i32 [[MISMATCH_INDEX]]
+; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[DOTNOT]], [[TMP36]]
+; LOOP-DEL-NEXT: br i1 [[OR_COND]], label [[WHILE_END]], label [[MISMATCH_LOOP]]
; LOOP-DEL: while.end:
-; LOOP-DEL-NEXT: [[INC_LCSSA:%.*]] = phi i32 [ [[X]], [[ENTRY:%.*]] ], [ [[N]], [[MISMATCH_LOOP_INC]] ], [ [[MISMATCH_INDEX]], [[MISMATCH_LOOP]] ], [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ]
+; LOOP-DEL-NEXT: [[INC_LCSSA:%.*]] = phi i32 [ [[X]], [[ENTRY:%.*]] ], [ [[N]], [[MISMATCH_VECTOR_LOOP_INC]] ], [ [[TMP28]], [[MISMATCH_VECTOR_LOOP_FOUND]] ], [ [[TMP38]], [[MISMATCH_LOOP]] ]
; LOOP-DEL-NEXT: ret i32 [[INC_LCSSA]]
;
; MASKED-LABEL: define i32 @compare_bytes_extra_cmp(
@@ -1812,7 +1816,8 @@ define void @compare_bytes_cleanup_block(ptr %src1, ptr %src2) {
; LOOP-DEL-NEXT: [[TMP2:%.*]] = load i8, ptr [[TMP1]], align 1
; LOOP-DEL-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP0]]
; LOOP-DEL-NEXT: [[TMP4:%.*]] = load i8, ptr [[TMP3]], align 1
-; LOOP-DEL-NEXT: [[TMP5:%.*]] = icmp ne i8 [[TMP2]], [[TMP4]]
+; LOOP-DEL-NEXT: [[TMP8:%.*]] = icmp eq i8 [[TMP2]], [[TMP4]]
+; LOOP-DEL-NEXT: [[TMP5:%.*]] = xor i1 [[TMP8]], true
; LOOP-DEL-NEXT: [[TMP6]] = add i32 [[MISMATCH_INDEX]], 1
; LOOP-DEL-NEXT: [[TMP7:%.*]] = icmp eq i32 [[TMP6]], 0
; LOOP-DEL-NEXT: [[OR_COND:%.*]] = or i1 [[TMP5]], [[TMP7]]
diff --git a/llvm/test/Transforms/LoopStrengthReduce/AArch64/pr53625.ll b/llvm/test/Transforms/LoopStrengthReduce/AArch64/pr53625.ll
index 3c1094f2ee31d..5787f6c041b63 100644
--- a/llvm/test/Transforms/LoopStrengthReduce/AArch64/pr53625.ll
+++ b/llvm/test/Transforms/LoopStrengthReduce/AArch64/pr53625.ll
@@ -8,22 +8,25 @@ define i32 @test(i32 %c, ptr %a, ptr %b) {
; CHECK-LABEL: test:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: cmp w0, #1
-; CHECK-NEXT: b.lt .LBB0_4
+; CHECK-NEXT: b.lt .LBB0_5
; CHECK-NEXT: // %bb.1: // %for.body.preheader
; CHECK-NEXT: mov w8, w0
+; CHECK-NEXT: sub x8, x8, #1
; CHECK-NEXT: .LBB0_2: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: ldr w9, [x1], #4
-; CHECK-NEXT: cbnz w9, .LBB0_5
-; CHECK-NEXT: // %bb.3: // %for.cond
+; CHECK-NEXT: ldr w10, [x1], #4
+; CHECK-NEXT: cmp w10, #0
+; CHECK-NEXT: cset w0, ne
+; CHECK-NEXT: cbnz w10, .LBB0_4
+; CHECK-NEXT: // %bb.3: // %for.body
; CHECK-NEXT: // in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT: subs x8, x8, #1
-; CHECK-NEXT: b.ne .LBB0_2
-; CHECK-NEXT: .LBB0_4:
-; CHECK-NEXT: mov w0, wzr
+; CHECK-NEXT: mov x9, x8
+; CHECK-NEXT: sub x8, x8, #1
+; CHECK-NEXT: cbnz x9, .LBB0_2
+; CHECK-NEXT: .LBB0_4: // %return
; CHECK-NEXT: ret
; CHECK-NEXT: .LBB0_5:
-; CHECK-NEXT: mov w0, #1 // =0x1
+; CHECK-NEXT: mov w0, wzr
; CHECK-NEXT: ret
entry:
%cmp13 = icmp sgt i32 %c, 0
@@ -56,13 +59,13 @@ define i64 @IVIncHoist_not_all_user_in_header(i32 %c, ptr %a, ptr %b) {
; CHECK-LABEL: IVIncHoist_not_all_user_in_header:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: cmp w0, #1
-; CHECK-NEXT: b.lt .LBB1_5
+; CHECK-NEXT: b.lt .LBB1_6
; CHECK-NEXT: // %bb.1: // %for.body.preheader
-; CHECK-NEXT: mov x8, xzr
; CHECK-NEXT: mov w9, w0
+; CHECK-NEXT: mov x8, xzr
; CHECK-NEXT: add x10, x1, #4
+; CHECK-NEXT: sub x9, x9, #1
; CHECK-NEXT: add x11, x2, #8
-; CHECK-NEXT: mov w0, #1 // =0x1
; CHECK-NEXT: .LBB1_2: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
; CHECK-NEXT: ldr w12, [x10, x8, lsl #2]
@@ -70,18 +73,22 @@ define i64 @IVIncHoist_not_all_user_in_header(i32 %c, ptr %a, ptr %b) {
; CHECK-NEXT: // %bb.3: // %if.then
; CHECK-NEXT: // in Loop: Header=BB1_2 Depth=1
; CHECK-NEXT: ldr w12, [x11, x8, lsl #2]
-; CHECK-NEXT: cbnz w12, .LBB1_6
-; CHECK-NEXT: // %bb.4: // %for.cond
+; CHECK-NEXT: add x13, x8, #3
+; CHECK-NEXT: cmp w12, #0
+; CHECK-NEXT: csel x0, xzr, x13, eq
+; CHECK-NEXT: cbnz w12, .LBB1_5
+; CHECK-NEXT: // %bb.4: // %if.then
; CHECK-NEXT: // in Loop: Header=BB1_2 Depth=1
-; CHECK-NEXT: add x8, x8, #1
; CHECK-NEXT: cmp x9, x8
+; CHECK-NEXT: add x8, x8, #1
; CHECK-NEXT: b.ne .LBB1_2
-; CHECK-NEXT: .LBB1_5:
+; CHECK-NEXT: .LBB1_5: // %return
+; CHECK-NEXT: ret
+; CHECK-NEXT: .LBB1_6:
; CHECK-NEXT: mov x0, xzr
; CHECK-NEXT: ret
-; CHECK-NEXT: .LBB1_6: // %if.then.return.loopexit_crit_edge
-; CHECK-NEXT: add x0, x8, #3
-; CHECK-NEXT: .LBB1_7: // %return
+; CHECK-NEXT: .LBB1_7:
+; CHECK-NEXT: mov w0, #1 // =0x1
; CHECK-NEXT: ret
entry:
%cmp13 = icmp sgt i32 %c, 0
@@ -126,23 +133,25 @@ define i32 @negative_test_type_is_struct(i32 %c, ptr %a, ptr %b) {
; CHECK-LABEL: negative_test_type_is_struct:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: cmp w0, #1
-; CHECK-NEXT: b.lt .LBB2_4
+; CHECK-NEXT: b.lt .LBB2_5
; CHECK-NEXT: // %bb.1: // %for.body.preheader
; CHECK-NEXT: mov w8, w0
+; CHECK-NEXT: sub x8, x8, #1
; CHECK-NEXT: .LBB2_2: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: ldr w9, [x1]
-; CHECK-NEXT: cbnz w9, .LBB2_5
-; CHECK-NEXT: // %bb.3: // %for.cond
+; CHECK-NEXT: ldr w10, [x1], #4
+; CHECK-NEXT: cmp w10, #0
+; CHECK-NEXT: cset w0, ne
+; CHECK-NEXT: cbnz w10, .LBB2_4
+; CHECK-NEXT: // %bb.3: // %for.body
; CHECK-NEXT: // in Loop: Header=BB2_2 Depth=1
-; CHECK-NEXT: subs x8, x8, #1
-; CHECK-NEXT: add x1, x1, #4
-; CHECK-NEXT: b.ne .LBB2_2
-; CHECK-NEXT: .LBB2_4:
-; CHECK-NEXT: mov w0, wzr
+; CHECK-NEXT: mov x9, x8
+; CHECK-NEXT: sub x8, x8, #1
+; CHECK-NEXT: cbnz x9, .LBB2_2
+; CHECK-NEXT: .LBB2_4: // %return
; CHECK-NEXT: ret
; CHECK-NEXT: .LBB2_5:
-; CHECK-NEXT: mov w0, #1 // =0x1
+; CHECK-NEXT: mov w0, wzr
; CHECK-NEXT: ret
entry:
%cmp13 = icmp sgt i32 %c, 0
diff --git a/llvm/test/Transforms/LoopStrengthReduce/AArch64/small-constant.ll b/llvm/test/Transforms/LoopStrengthReduce/AArch64/small-constant.ll
index cd6b410b67aa3..85670b9f39961 100644
--- a/llvm/test/Transforms/LoopStrengthReduce/AArch64/small-constant.ll
+++ b/llvm/test/Transforms/LoopStrengthReduce/AArch64/small-constant.ll
@@ -19,22 +19,23 @@ define float @test1(ptr nocapture readonly %arr, i64 %start, float %threshold) {
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: cbz x1, .LBB0_4
; CHECK-NEXT: // %bb.1: // %for.body.preheader
+; CHECK-NEXT: fmov s2, #-7.00000000
; CHECK-NEXT: add x8, x0, #28
; CHECK-NEXT: .LBB0_2: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
; CHECK-NEXT: ldr s1, [x8, x1, lsl #2]
+; CHECK-NEXT: adds x1, x1, #1
+; CHECK-NEXT: cset w9, hs
; CHECK-NEXT: fcmp s1, s0
-; CHECK-NEXT: b.gt .LBB0_5
-; CHECK-NEXT: // %bb.3: // %for.cond
-; CHECK-NEXT: // in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT: add x1, x1, #1
-; CHECK-NEXT: cbnz x1, .LBB0_2
+; CHECK-NEXT: fcsel s1, s1, s2, gt
+; CHECK-NEXT: ccmp w9, #0, #0, le
+; CHECK-NEXT: b.eq .LBB0_2
+; CHECK-NEXT: // %bb.3: // %cleanup2
+; CHECK-NEXT: fmov s0, s1
+; CHECK-NEXT: ret
; CHECK-NEXT: .LBB0_4:
; CHECK-NEXT: fmov s0, #-7.00000000
; CHECK-NEXT: ret
-; CHECK-NEXT: .LBB0_5: // %cleanup2
-; CHECK-NEXT: fmov s0, s1
-; CHECK-NEXT: ret
entry:
%cmp11 = icmp eq i64 %start, 0
br i1 %cmp11, label %cleanup2, label %for.body
@@ -64,24 +65,25 @@ define float @test2(ptr nocapture readonly %arr, i64 %start, float %threshold) {
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: cbz x1, .LBB1_4
; CHECK-NEXT: // %bb.1: // %for.body.preheader
+; CHECK-NEXT: fmov s2, #-7.00000000
; CHECK-NEXT: add x8, x0, #28
; CHECK-NEXT: .LBB1_2: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
; CHECK-NEXT: scvtf s1, x1
-; CHECK-NEXT: fadd s2, s1, s0
-; CHECK-NEXT: ldr s1, [x8, x1, lsl #2]
-; CHECK-NEXT: fcmp s1, s2
-; CHECK-NEXT: b.gt .LBB1_5
-; CHECK-NEXT: // %bb.3: // %for.cond
-; CHECK-NEXT: // in Loop: Header=BB1_2 Depth=1
-; CHECK-NEXT: add x1, x1, #1
-; CHECK-NEXT: cbnz x1, .LBB1_2
+; CHECK-NEXT: ldr s3, [x8, x1, lsl #2]
+; CHECK-NEXT: adds x1, x1, #1
+; CHECK-NEXT: cset w9, hs
+; CHECK-NEXT: fadd s1, s1, s0
+; CHECK-NEXT: fcmp s3, s1
+; CHECK-NEXT: fcsel s1, s3, s2, gt
+; CHECK-NEXT: ccmp w9, #0, #0, le
+; CHECK-NEXT: b.eq .LBB1_2
+; CHECK-NEXT: // %bb.3: // %cleanup4
+; CHECK-NEXT: fmov s0, s1
+; CHECK-NEXT: ret
; CHECK-NEXT: .LBB1_4:
; CHECK-NEXT: fmov s0, #-7.00000000
; CHECK-NEXT: ret
-; CHECK-NEXT: .LBB1_5: // %cleanup4
-; CHECK-NEXT: fmov s0, s1
-; CHECK-NEXT: ret
entry:
%cmp14 = icmp eq i64 %start, 0
br i1 %cmp14, label %cleanup4, label %for.body
diff --git a/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll b/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll
index 5f6e66e344625..227c45fa80a1d 100644
--- a/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll
+++ b/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll
@@ -2967,14 +2967,9 @@ define void @non_loop(i32 %arg) {
; CHECK: preheader:
; CHECK-NEXT: br label %header
; CHECK: header:
-; CHECK-NEXT: br i1 true, label %latchExit, label %latch
-; CHECK: latch:
-; CHECK-NEXT: br label %latchExit
-; CHECK: latchExit:
-; CHECK-NEXT: %i2.ph = phi i32 [ %arg, %header ], [ -1, %latch ]
; CHECK-NEXT: br label %returnblock
; CHECK: returnblock:
-; CHECK-NEXT: %i2 = phi i32 [ -1, %entry ], [ %i2.ph, %latchExit ]
+; CHECK-NEXT: %i2 = phi i32 [ -1, %entry ], [ %arg, %header ]
; CHECK-NEXT: ret void
;
@@ -3007,267 +3002,102 @@ returnblock: ; preds = %latchExit, %entr
define void @unique_exit(i32 %N, i32 %M) {
; EPILOG-LABEL: @unique_exit(
; EPILOG-NEXT: preheader:
-; EPILOG-NEXT: %M.shifted = shl i32 %M, 3
-; EPILOG-NEXT: %umax = call i32 @llvm.umax.i32(i32 %M.shifted, i32 1)
-; EPILOG-NEXT: %0 = freeze i32 %umax
-; EPILOG-NEXT: %1 = add i32 %0, -1
-; EPILOG-NEXT: %xtraiter = and i32 %0, 7
-; EPILOG-NEXT: %2 = icmp ult i32 %1, 7
-; EPILOG-NEXT: br i1 %2, label %latchExit.unr-lcssa, label %preheader.new
-; EPILOG: preheader.new:
-; EPILOG-NEXT: %unroll_iter = sub i32 %0, %xtraiter
+; EPILOG-NEXT: %M.shifted = shl nuw i32 %M, 3
; EPILOG-NEXT: br label %header
; EPILOG: header:
-; EPILOG-NEXT: %i4 = phi i32 [ 0, %preheader.new ], [ %inc.7, %latch.7 ]
-; EPILOG-NEXT: %niter = phi i32 [ 0, %preheader.new ], [ %niter.next.7, %latch.7 ]
-; EPILOG-NEXT: %inc = add nuw nsw i32 %i4, 1
+; EPILOG-NEXT: %i4 = phi i32 [ 0, %preheader ], [ %inc, %header ]
+; EPILOG-NEXT: %inc = add nuw i32 %i4, 1
; EPILOG-NEXT: %cmp1 = icmp ult i32 %inc, %N
-; EPILOG-NEXT: br i1 %cmp1, label %latch, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch:
-; EPILOG-NEXT: %inc.1 = add nuw nsw i32 %i4, 2
-; EPILOG-NEXT: %cmp1.1 = icmp ult i32 %inc.1, %N
-; EPILOG-NEXT: br i1 %cmp1.1, label %latch.1, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch.1:
-; EPILOG-NEXT: %inc.2 = add nuw nsw i32 %i4, 3
-; EPILOG-NEXT: %cmp1.2 = icmp ult i32 %inc.2, %N
-; EPILOG-NEXT: br i1 %cmp1.2, label %latch.2, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch.2:
-; EPILOG-NEXT: %inc.3 = add nuw nsw i32 %i4, 4
-; EPILOG-NEXT: %cmp1.3 = icmp ult i32 %inc.3, %N
-; EPILOG-NEXT: br i1 %cmp1.3, label %latch.3, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch.3:
-; EPILOG-NEXT: %inc.4 = add nuw nsw i32 %i4, 5
-; EPILOG-NEXT: %cmp1.4 = icmp ult i32 %inc.4, %N
-; EPILOG-NEXT: br i1 %cmp1.4, label %latch.4, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch.4:
-; EPILOG-NEXT: %inc.5 = add nuw nsw i32 %i4, 6
-; EPILOG-NEXT: %cmp1.5 = icmp ult i32 %inc.5, %N
-; EPILOG-NEXT: br i1 %cmp1.5, label %latch.5, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch.5:
-; EPILOG-NEXT: %inc.6 = add nuw nsw i32 %i4, 7
-; EPILOG-NEXT: %cmp1.6 = icmp ult i32 %inc.6, %N
-; EPILOG-NEXT: br i1 %cmp1.6, label %latch.6, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch.6:
-; EPILOG-NEXT: %inc.7 = add nuw i32 %i4, 8
-; EPILOG-NEXT: %cmp1.7 = icmp ult i32 %inc.7, %N
-; EPILOG-NEXT: br i1 %cmp1.7, label %latch.7, label %latchExit.epilog-lcssa.loopexit
-; EPILOG: latch.7:
-; EPILOG-NEXT: %niter.next.7 = add nuw i32 %niter, 8
-; EPILOG-NEXT: %niter.ncmp.7 = icmp ne i32 %niter.next.7, %unroll_iter
-; EPILOG-NEXT: br i1 %niter.ncmp.7, label %header, label %latchExit.unr-lcssa.loopexit
-; EPILOG: latchExit.unr-lcssa.loopexit:
-; EPILOG-NEXT: %i2.ph.ph.ph = phi i32 [ -1, %latch.7 ]
-; EPILOG-NEXT: %i4.unr.ph = phi i32 [ %inc.7, %latch.7 ]
-; EPILOG-NEXT: br label %latchExit.unr-lcssa
-; EPILOG: latchExit.unr-lcssa:
-; EPILOG-NEXT: %i2.ph.ph = phi i32 [ poison, %preheader ], [ %i2.ph.ph.ph, %latchExit.unr-lcssa.loopexit ]
-; EPILOG-NEXT: %i4.unr = phi i32 [ 0, %preheader ], [ %i4.unr.ph, %latchExit.unr-lcssa.loopexit ]
-; EPILOG-NEXT: %lcmp.mod = icmp ne i32 %xtraiter, 0
-; EPILOG-NEXT: br i1 %lcmp.mod, label %header.epil.preheader, label %latchExit
-; EPILOG: header.epil.preheader:
-; EPILOG-NEXT: br label %header.epil
-; EPILOG: header.epil:
-; EPILOG-NEXT: %i4.epil = phi i32 [ %inc.epil, %latch.epil ], [ %i4.unr, %header.epil.preheader ]
-; EPILOG-NEXT: %epil.iter = phi i32 [ 0, %header.epil.preheader ], [ %epil.iter.next, %latch.epil ]
-; EPILOG-NEXT: %inc.epil = add nuw i32 %i4.epil, 1
-; EPILOG-NEXT: %cmp1.epil = icmp ult i32 %inc.epil, %N
-; EPILOG-NEXT: br i1 %cmp1.epil, label %latch.epil, label %latchExit.epilog-lcssa.loopexit2
-; EPILOG: latch.epil:
-; EPILOG-NEXT: %cmp.epil = icmp ult i32 %inc.epil, %M.shifted
-; EPILOG-NEXT: %epil.iter.next = add i32 %epil.iter, 1
-; EPILOG-NEXT: %epil.iter.cmp = icmp ne i32 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %header.epil, label %latchExit.epilog-lcssa.loopexit2, !llvm.loop !8
-; EPILOG: latchExit.epilog-lcssa.loopexit:
-; EPILOG-NEXT: %i2.ph.ph1.ph = phi i32 [ %i4, %header ], [ %inc, %latch ], [ %inc.1, %latch.1 ], [ %inc.2, %latch.2 ], [ %inc.3, %latch.3 ], [ %inc.4, %latch.4 ], [ %inc.5, %latch.5 ], [ %inc.6, %latch.6 ]
-; EPILOG-NEXT: br label %latchExit.epilog-lcssa
-; EPILOG: latchExit.epilog-lcssa.loopexit2:
-; EPILOG-NEXT: %i2.ph.ph1.ph3 = phi i32 [ %i4.epil, %header.epil ], [ -1, %latch.epil ]
-; EPILOG-NEXT: br label %latchExit.epilog-lcssa
-; EPILOG: latchExit.epilog-lcssa:
-; EPILOG-NEXT: %i2.ph.ph1 = phi i32 [ %i2.ph.ph1.ph, %latchExit.epilog-lcssa.loopexit ], [ %i2.ph.ph1.ph3, %latchExit.epilog-lcssa.loopexit2 ]
-; EPILOG-NEXT: br label %latchExit
+; EPILOG-NEXT: %cmp = icmp ult i32 %inc, %M.shifted
+; EPILOG-NEXT: %0 = select i1 %cmp1, i32 -1, i32 %i4
+; EPILOG-NEXT: %or.cond = select i1 %cmp1, i1 %cmp, i1 false
+; EPILOG-NEXT: br i1 %or.cond, label %header, label %latchExit
; EPILOG: latchExit:
-; EPILOG-NEXT: %i2.ph = phi i32 [ %i2.ph.ph, %latchExit.unr-lcssa ], [ %i2.ph.ph1, %latchExit.epilog-lcssa ]
; EPILOG-NEXT: ret void
;
; EPILOG-BLOCK-LABEL: @unique_exit(
; EPILOG-BLOCK-NEXT: preheader:
; EPILOG-BLOCK-NEXT: %M.shifted = shl i32 %M, 3
; EPILOG-BLOCK-NEXT: %umax = call i32 @llvm.umax.i32(i32 %M.shifted, i32 1)
-; EPILOG-BLOCK-NEXT: %0 = freeze i32 %umax
-; EPILOG-BLOCK-NEXT: %1 = add i32 %0, -1
-; EPILOG-BLOCK-NEXT: %xtraiter = and i32 %0, 1
-; EPILOG-BLOCK-NEXT: %2 = icmp ult i32 %1, 1
-; EPILOG-BLOCK-NEXT: br i1 %2, label %latchExit.unr-lcssa, label %preheader.new
+; EPILOG-BLOCK-NEXT: %0 = add i32 %umax, -1
+; EPILOG-BLOCK-NEXT: %1 = freeze i32 %0
+; EPILOG-BLOCK-NEXT: %umax1 = call i32 @llvm.umax.i32(i32 %N, i32 1)
+; EPILOG-BLOCK-NEXT: %2 = add i32 %umax1, -1
+; EPILOG-BLOCK-NEXT: %umin = call i32 @llvm.umin.i32(i32 %1, i32 %2)
+; EPILOG-BLOCK-NEXT: %3 = add nuw i32 %umin, 1
+; EPILOG-BLOCK-NEXT: %xtraiter = and i32 %3, 1
+; EPILOG-BLOCK-NEXT: %4 = icmp ult i32 %umin, 1
+; EPILOG-BLOCK-NEXT: br i1 %4, label %latchExit.unr-lcssa, label %preheader.new
; EPILOG-BLOCK: preheader.new:
-; EPILOG-BLOCK-NEXT: %unroll_iter = sub i32 %0, %xtraiter
+; EPILOG-BLOCK-NEXT: %unroll_iter = sub i32 %3, %xtraiter
; EPILOG-BLOCK-NEXT: br label %header
; EPILOG-BLOCK: header:
-; EPILOG-BLOCK-NEXT: %i4 = phi i32 [ 0, %preheader.new ], [ %inc.1, %latch.1 ]
-; EPILOG-BLOCK-NEXT: %niter = phi i32 [ 0, %preheader.new ], [ %niter.next.1, %latch.1 ]
-; EPILOG-BLOCK-NEXT: %inc = add nuw nsw i32 %i4, 1
-; EPILOG-BLOCK-NEXT: %cmp1 = icmp ult i32 %inc, %N
-; EPILOG-BLOCK-NEXT: br i1 %cmp1, label %latch, label %latchExit.epilog-lcssa.loopexit
-; EPILOG-BLOCK: latch:
+; EPILOG-BLOCK-NEXT: %i4 = phi i32 [ 0, %preheader.new ], [ %inc.1, %header ]
+; EPILOG-BLOCK-NEXT: %niter = phi i32 [ 0, %preheader.new ], [ %niter.next.1, %header ]
; EPILOG-BLOCK-NEXT: %inc.1 = add nuw i32 %i4, 2
-; EPILOG-BLOCK-NEXT: %cmp1.1 = icmp ult i32 %inc.1, %N
-; EPILOG-BLOCK-NEXT: br i1 %cmp1.1, label %latch.1, label %latchExit.epilog-lcssa.loopexit
-; EPILOG-BLOCK: latch.1:
-; EPILOG-BLOCK-NEXT: %niter.next.1 = add nuw i32 %niter, 2
+; EPILOG-BLOCK-NEXT: %niter.next.1 = add i32 %niter, 2
; EPILOG-BLOCK-NEXT: %niter.ncmp.1 = icmp ne i32 %niter.next.1, %unroll_iter
; EPILOG-BLOCK-NEXT: br i1 %niter.ncmp.1, label %header, label %latchExit.unr-lcssa.loopexit, !llvm.loop !8
; EPILOG-BLOCK: latchExit.unr-lcssa.loopexit:
-; EPILOG-BLOCK-NEXT: %i2.ph.ph.ph = phi i32 [ -1, %latch.1 ]
-; EPILOG-BLOCK-NEXT: %i4.unr.ph = phi i32 [ %inc.1, %latch.1 ]
; EPILOG-BLOCK-NEXT: br label %latchExit.unr-lcssa
; EPILOG-BLOCK: latchExit.unr-lcssa:
-; EPILOG-BLOCK-NEXT: %i2.ph.ph = phi i32 [ poison, %preheader ], [ %i2.ph.ph.ph, %latchExit.unr-lcssa.loopexit ]
-; EPILOG-BLOCK-NEXT: %i4.unr = phi i32 [ 0, %preheader ], [ %i4.unr.ph, %latchExit.unr-lcssa.loopexit ]
; EPILOG-BLOCK-NEXT: %lcmp.mod = icmp ne i32 %xtraiter, 0
; EPILOG-BLOCK-NEXT: br i1 %lcmp.mod, label %header.epil.preheader, label %latchExit
; EPILOG-BLOCK: header.epil.preheader:
; EPILOG-BLOCK-NEXT: br label %header.epil
; EPILOG-BLOCK: header.epil:
-; EPILOG-BLOCK-NEXT: %inc.epil = add nuw i32 %i4.unr, 1
-; EPILOG-BLOCK-NEXT: %cmp1.epil = icmp ult i32 %inc.epil, %N
-; EPILOG-BLOCK-NEXT: br i1 %cmp1.epil, label %latch.epil, label %latchExit.epilog-lcssa
-; EPILOG-BLOCK: latch.epil:
-; EPILOG-BLOCK-NEXT: br label %latchExit.epilog-lcssa
-; EPILOG-BLOCK: latchExit.epilog-lcssa.loopexit:
-; EPILOG-BLOCK-NEXT: %i2.ph.ph1.ph = phi i32 [ %i4, %header ], [ %inc, %latch ]
-; EPILOG-BLOCK-NEXT: br label %latchExit.epilog-lcssa
-; EPILOG-BLOCK: latchExit.epilog-lcssa:
-; EPILOG-BLOCK-NEXT: %i2.ph.ph1 = phi i32 [ -1, %latch.epil ], [ %i4.unr, %header.epil ], [ %i2.ph.ph1.ph, %latchExit.epilog-lcssa.loopexit ]
; EPILOG-BLOCK-NEXT: br label %latchExit
; EPILOG-BLOCK: latchExit:
-; EPILOG-BLOCK-NEXT: %i2.ph = phi i32 [ %i2.ph.ph, %latchExit.unr-lcssa ], [ %i2.ph.ph1, %latchExit.epilog-lcssa ]
; EPILOG-BLOCK-NEXT: ret void
;
; PROLOG-LABEL: @unique_exit(
; PROLOG-NEXT: preheader:
-; PROLOG-NEXT: %M.shifted = shl i32 %M, 3
-; PROLOG-NEXT: %umax = call i32 @llvm.umax.i32(i32 %M.shifted, i32 1)
-; PROLOG-NEXT: %0 = freeze i32 %umax
-; PROLOG-NEXT: %1 = add i32 %0, -1
-; PROLOG-NEXT: %xtraiter = and i32 %0, 7
-; PROLOG-NEXT: %lcmp.mod = icmp ne i32 %xtraiter, 0
-; PROLOG-NEXT: br i1 %lcmp.mod, label %header.prol.preheader, label %header.prol.loopexit
-; PROLOG: header.prol.preheader:
-; PROLOG-NEXT: br label %header.prol
-; PROLOG: header.prol:
-; PROLOG-NEXT: %i4.prol = phi i32 [ %inc.prol, %latch.prol ], [ 0, %header.prol.preheader ]
-; PROLOG-NEXT: %prol.iter = phi i32 [ 0, %header.prol.preheader ], [ %prol.iter.next, %latch.prol ]
-; PROLOG-NEXT: %inc.prol = add nuw i32 %i4.prol, 1
-; PROLOG-NEXT: %cmp1.prol = icmp ult i32 %inc.prol, %N
-; PROLOG-NEXT: br i1 %cmp1.prol, label %latch.prol, label %latchExit.unr-lcssa.loopexit1
-; PROLOG: latch.prol:
-; PROLOG-NEXT: %cmp.prol = icmp ult i32 %inc.prol, %M.shifted
-; PROLOG-NEXT: %prol.iter.next = add i32 %prol.iter, 1
-; PROLOG-NEXT: %prol.iter.cmp = icmp ne i32 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %header.prol, label %header.prol.loopexit.unr-lcssa, !llvm.loop !8
-; PROLOG: header.prol.loopexit.unr-lcssa:
-; PROLOG-NEXT: %i4.unr.ph = phi i32 [ %inc.prol, %latch.prol ]
-; PROLOG-NEXT: %i2.ph.unr.ph = phi i32 [ -1, %latch.prol ]
-; PROLOG-NEXT: br label %header.prol.loopexit
-; PROLOG: header.prol.loopexit:
-; PROLOG-NEXT: %i4.unr = phi i32 [ 0, %preheader ], [ %i4.unr.ph, %header.prol.loopexit.unr-lcssa ]
-; PROLOG-NEXT: %i2.ph.unr = phi i32 [ poison, %preheader ], [ %i2.ph.unr.ph, %header.prol.loopexit.unr-lcssa ]
-; PROLOG-NEXT: %2 = icmp ult i32 %1, 7
-; PROLOG-NEXT: br i1 %2, label %latchExit, label %preheader.new
-; PROLOG: preheader.new:
+; PROLOG-NEXT: %M.shifted = shl nuw i32 %M, 3
; PROLOG-NEXT: br label %header
; PROLOG: header:
-; PROLOG-NEXT: %i4 = phi i32 [ %i4.unr, %preheader.new ], [ %inc.7, %latch.7 ]
+; PROLOG-NEXT: %i4 = phi i32 [ 0, %preheader ], [ %inc, %header ]
; PROLOG-NEXT: %inc = add nuw i32 %i4, 1
; PROLOG-NEXT: %cmp1 = icmp ult i32 %inc, %N
-; PROLOG-NEXT: br i1 %cmp1, label %latch, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch:
-; PROLOG-NEXT: %inc.1 = add nuw i32 %i4, 2
-; PROLOG-NEXT: %cmp1.1 = icmp ult i32 %inc.1, %N
-; PROLOG-NEXT: br i1 %cmp1.1, label %latch.1, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch.1:
-; PROLOG-NEXT: %inc.2 = add nuw i32 %i4, 3
-; PROLOG-NEXT: %cmp1.2 = icmp ult i32 %inc.2, %N
-; PROLOG-NEXT: br i1 %cmp1.2, label %latch.2, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch.2:
-; PROLOG-NEXT: %inc.3 = add nuw i32 %i4, 4
-; PROLOG-NEXT: %cmp1.3 = icmp ult i32 %inc.3, %N
-; PROLOG-NEXT: br i1 %cmp1.3, label %latch.3, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch.3:
-; PROLOG-NEXT: %inc.4 = add nuw i32 %i4, 5
-; PROLOG-NEXT: %cmp1.4 = icmp ult i32 %inc.4, %N
-; PROLOG-NEXT: br i1 %cmp1.4, label %latch.4, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch.4:
-; PROLOG-NEXT: %inc.5 = add nuw i32 %i4, 6
-; PROLOG-NEXT: %cmp1.5 = icmp ult i32 %inc.5, %N
-; PROLOG-NEXT: br i1 %cmp1.5, label %latch.5, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch.5:
-; PROLOG-NEXT: %inc.6 = add nuw i32 %i4, 7
-; PROLOG-NEXT: %cmp1.6 = icmp ult i32 %inc.6, %N
-; PROLOG-NEXT: br i1 %cmp1.6, label %latch.6, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch.6:
-; PROLOG-NEXT: %inc.7 = add nuw i32 %i4, 8
-; PROLOG-NEXT: %cmp1.7 = icmp ult i32 %inc.7, %N
-; PROLOG-NEXT: br i1 %cmp1.7, label %latch.7, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latch.7:
-; PROLOG-NEXT: %cmp.7 = icmp ult i32 %inc.7, %M.shifted
-; PROLOG-NEXT: br i1 %cmp.7, label %header, label %latchExit.unr-lcssa.loopexit
-; PROLOG: latchExit.unr-lcssa.loopexit:
-; PROLOG-NEXT: %i2.ph.ph.ph = phi i32 [ %i4, %header ], [ %inc, %latch ], [ %inc.1, %latch.1 ], [ %inc.2, %latch.2 ], [ %inc.3, %latch.3 ], [ %inc.4, %latch.4 ], [ %inc.5, %latch.5 ], [ %inc.6, %latch.6 ], [ -1, %latch.7 ]
-; PROLOG-NEXT: br label %latchExit.unr-lcssa
-; PROLOG: latchExit.unr-lcssa.loopexit1:
-; PROLOG-NEXT: %i2.ph.ph.ph2 = phi i32 [ %i4.prol, %header.prol ]
-; PROLOG-NEXT: br label %latchExit.unr-lcssa
-; PROLOG: latchExit.unr-lcssa:
-; PROLOG-NEXT: %i2.ph.ph = phi i32 [ %i2.ph.ph.ph, %latchExit.unr-lcssa.loopexit ], [ %i2.ph.ph.ph2, %latchExit.unr-lcssa.loopexit1 ]
-; PROLOG-NEXT: br label %latchExit
+; PROLOG-NEXT: %cmp = icmp ult i32 %inc, %M.shifted
+; PROLOG-NEXT: %0 = select i1 %cmp1, i32 -1, i32 %i4
+; PROLOG-NEXT: %or.cond = select i1 %cmp1, i1 %cmp, i1 false
+; PROLOG-NEXT: br i1 %or.cond, label %header, label %latchExit
; PROLOG: latchExit:
-; PROLOG-NEXT: %i2.ph = phi i32 [ %i2.ph.unr, %header.prol.loopexit ], [ %i2.ph.ph, %latchExit.unr-lcssa ]
; PROLOG-NEXT: ret void
;
; PROLOG-BLOCK-LABEL: @unique_exit(
; PROLOG-BLOCK-NEXT: preheader:
; PROLOG-BLOCK-NEXT: %M.shifted = shl i32 %M, 3
; PROLOG-BLOCK-NEXT: %umax = call i32 @llvm.umax.i32(i32 %M.shifted, i32 1)
-; PROLOG-BLOCK-NEXT: %0 = freeze i32 %umax
-; PROLOG-BLOCK-NEXT: %1 = add i32 %0, -1
-; PROLOG-BLOCK-NEXT: %xtraiter = and i32 %0, 1
+; PROLOG-BLOCK-NEXT: %0 = add i32 %umax, -1
+; PROLOG-BLOCK-NEXT: %1 = freeze i32 %0
+; PROLOG-BLOCK-NEXT: %umax1 = call i32 @llvm.umax.i32(i32 %N, i32 1)
+; PROLOG-BLOCK-NEXT: %2 = add i32 %umax1, -1
+; PROLOG-BLOCK-NEXT: %umin = call i32 @llvm.umin.i32(i32 %1, i32 %2)
+; PROLOG-BLOCK-NEXT: %3 = add nuw i32 %umin, 1
+; PROLOG-BLOCK-NEXT: %xtraiter = and i32 %3, 1
; PROLOG-BLOCK-NEXT: %lcmp.mod = icmp ne i32 %xtraiter, 0
; PROLOG-BLOCK-NEXT: br i1 %lcmp.mod, label %header.prol.preheader, label %header.prol.loopexit
; PROLOG-BLOCK: header.prol.preheader:
; PROLOG-BLOCK-NEXT: br label %header.prol
; PROLOG-BLOCK: header.prol:
-; PROLOG-BLOCK-NEXT: %cmp1.prol = icmp ult i32 1, %N
-; PROLOG-BLOCK-NEXT: br i1 %cmp1.prol, label %latch.prol, label %latchExit.unr-lcssa
-; PROLOG-BLOCK: latch.prol:
; PROLOG-BLOCK-NEXT: br label %header.prol.loopexit
; PROLOG-BLOCK: header.prol.loopexit:
-; PROLOG-BLOCK-NEXT: %i4.unr = phi i32 [ 0, %preheader ], [ 1, %latch.prol ]
-; PROLOG-BLOCK-NEXT: %i2.ph.unr = phi i32 [ poison, %preheader ], [ -1, %latch.prol ]
-; PROLOG-BLOCK-NEXT: %2 = icmp ult i32 %1, 1
-; PROLOG-BLOCK-NEXT: br i1 %2, label %latchExit, label %preheader.new
+; PROLOG-BLOCK-NEXT: %i4.unr = phi i32 [ 0, %preheader ], [ 1, %header.prol ]
+; PROLOG-BLOCK-NEXT: %4 = icmp ult i32 %umin, 1
+; PROLOG-BLOCK-NEXT: br i1 %4, label %latchExit, label %preheader.new
; PROLOG-BLOCK: preheader.new:
; PROLOG-BLOCK-NEXT: br label %header
; PROLOG-BLOCK: header:
-; PROLOG-BLOCK-NEXT: %i4 = phi i32 [ %i4.unr, %preheader.new ], [ %inc.1, %latch.1 ]
-; PROLOG-BLOCK-NEXT: %inc = add nuw i32 %i4, 1
-; PROLOG-BLOCK-NEXT: %cmp1 = icmp ult i32 %inc, %N
-; PROLOG-BLOCK-NEXT: br i1 %cmp1, label %latch, label %latchExit.unr-lcssa.loopexit
-; PROLOG-BLOCK: latch:
+; PROLOG-BLOCK-NEXT: %i4 = phi i32 [ %i4.unr, %preheader.new ], [ %inc.1, %header ]
; PROLOG-BLOCK-NEXT: %inc.1 = add nuw i32 %i4, 2
; PROLOG-BLOCK-NEXT: %cmp1.1 = icmp ult i32 %inc.1, %N
-; PROLOG-BLOCK-NEXT: br i1 %cmp1.1, label %latch.1, label %latchExit.unr-lcssa.loopexit
-; PROLOG-BLOCK: latch.1:
; PROLOG-BLOCK-NEXT: %cmp.1 = icmp ult i32 %inc.1, %M.shifted
-; PROLOG-BLOCK-NEXT: br i1 %cmp.1, label %header, label %latchExit.unr-lcssa.loopexit, !llvm.loop !8
-; PROLOG-BLOCK: latchExit.unr-lcssa.loopexit:
-; PROLOG-BLOCK-NEXT: %i2.ph.ph.ph = phi i32 [ %i4, %header ], [ %inc, %latch ], [ -1, %latch.1 ]
-; PROLOG-BLOCK-NEXT: br label %latchExit.unr-lcssa
+; PROLOG-BLOCK-NEXT: %or.cond.1 = select i1 %cmp1.1, i1 %cmp.1, i1 false
+; PROLOG-BLOCK-NEXT: br i1 %or.cond.1, label %header, label %latchExit.unr-lcssa, !llvm.loop !8
; PROLOG-BLOCK: latchExit.unr-lcssa:
-; PROLOG-BLOCK-NEXT: %i2.ph.ph = phi i32 [ 0, %header.prol ], [ %i2.ph.ph.ph, %latchExit.unr-lcssa.loopexit ]
; PROLOG-BLOCK-NEXT: br label %latchExit
; PROLOG-BLOCK: latchExit:
-; PROLOG-BLOCK-NEXT: %i2.ph = phi i32 [ %i2.ph.unr, %header.prol.loopexit ], [ %i2.ph.ph, %latchExit.unr-lcssa ]
; PROLOG-BLOCK-NEXT: ret void
;
@@ -3430,7 +3260,7 @@ define i64 @test5(i64 %trip, i64 %add, i1 %cond) {
; EPILOG-NEXT: %cmp.epil = icmp ne i64 %iv_next.epil, %trip
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %latchexit.epilog-lcssa, !llvm.loop !9
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %latchexit.epilog-lcssa, !llvm.loop !8
; EPILOG: latchexit.epilog-lcssa:
; EPILOG-NEXT: %sum.next.lcssa.ph1 = phi i64 [ %sum.next.epil, %loop_latch.epil ]
; EPILOG-NEXT: br label %latchexit
@@ -3535,7 +3365,7 @@ define i64 @test5(i64 %trip, i64 %add, i1 %cond) {
; PROLOG-NEXT: %cmp.prol = icmp ne i64 %iv_next.prol, %trip
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !9
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !8
; PROLOG: loop_header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %iv.unr.ph = phi i64 [ %iv_next.prol, %loop_latch.prol ]
; PROLOG-NEXT: %sum.unr.ph = phi i64 [ %sum.next.prol, %loop_latch.prol ]
@@ -3864,7 +3694,7 @@ define i32 @test6(ptr nocapture %a, i64 %n, i1 %cond, i32 %x) {
; EPILOG-NEXT: %exitcond.epil = icmp eq i64 %indvars.iv.next.epil, %n
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %header.epil, label %latch_exit.epilog-lcssa, !llvm.loop !10
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %header.epil, label %latch_exit.epilog-lcssa, !llvm.loop !9
; EPILOG: latch_exit.epilog-lcssa:
; EPILOG-NEXT: %sum.0.lcssa.ph1 = phi i32 [ %add.epil, %latch.epil ]
; EPILOG-NEXT: br label %latch_exit
@@ -3984,7 +3814,7 @@ define i32 @test6(ptr nocapture %a, i64 %n, i1 %cond, i32 %x) {
; PROLOG-NEXT: %exitcond.prol = icmp eq i64 %indvars.iv.next.prol, %n
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %header.prol, label %header.prol.loopexit.unr-lcssa, !llvm.loop !10
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %header.prol, label %header.prol.loopexit.unr-lcssa, !llvm.loop !9
; PROLOG: header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %sum.0.lcssa.unr.ph = phi i32 [ %add.prol, %latch.prol ]
; PROLOG-NEXT: %indvars.iv.unr.ph = phi i64 [ %indvars.iv.next.prol, %latch.prol ]
@@ -4258,7 +4088,7 @@ define i32 @test7(i32 %arg, i32 %arg1, i32 %arg2) {
; EPILOG-NEXT: %i9.epil = icmp slt i64 %add.epil, %sext
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %header.epil, label %latchexit.epilog-lcssa, !llvm.loop !11
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %header.epil, label %latchexit.epilog-lcssa, !llvm.loop !10
; EPILOG: latchexit.epilog-lcssa:
; EPILOG-NEXT: br label %latchexit
; EPILOG: latchexit:
@@ -4348,7 +4178,7 @@ define i32 @test7(i32 %arg, i32 %arg1, i32 %arg2) {
; PROLOG-NEXT: %i9.prol = icmp slt i64 %add.prol, %sext
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %header.prol, label %header.prol.loopexit.unr-lcssa, !llvm.loop !11
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %header.prol, label %header.prol.loopexit.unr-lcssa, !llvm.loop !10
; PROLOG: header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %i6.unr.ph = phi i64 [ %add.prol, %latch.prol ]
; PROLOG-NEXT: br label %header.prol.loopexit
@@ -4535,7 +4365,7 @@ define void @test8() {
; EPILOG-NEXT: %i6.epil = icmp ult i64 %i4.epil, 100
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %innerH.epil, label %exit.epilog-lcssa, !llvm.loop !12
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %innerH.epil, label %exit.epilog-lcssa, !llvm.loop !11
; EPILOG: exit.epilog-lcssa:
; EPILOG-NEXT: br label %exit
; EPILOG: exit.loopexit:
@@ -4639,7 +4469,7 @@ define void @test8() {
; PROLOG-NEXT: %i6.prol = icmp ult i64 %i4.prol, 100
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %innerH.prol, label %innerH.prol.loopexit.unr-lcssa, !llvm.loop !12
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %innerH.prol, label %innerH.prol.loopexit.unr-lcssa, !llvm.loop !11
; PROLOG: innerH.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %i3.unr.ph = phi i64 [ %i4.prol, %latch.prol ]
; PROLOG-NEXT: br label %innerH.prol.loopexit
@@ -4845,7 +4675,7 @@ define ptr addrspace(1) @test9(ptr nocapture readonly %arg, i32 %n) {
; EPILOG-NEXT: %iv.next.epil = add nuw nsw i64 %phi.epil, 1
; EPILOG-NEXT: %epil.iter.next = add i32 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i32 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %header.epil, label %outerLatch.loopexit.epilog-lcssa, !llvm.loop !13
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %header.epil, label %outerLatch.loopexit.epilog-lcssa, !llvm.loop !12
; EPILOG: outerLatch.loopexit.epilog-lcssa:
; EPILOG-NEXT: br label %outerLatch.loopexit
; EPILOG: outerLatch.loopexit:
@@ -4978,7 +4808,7 @@ define ptr addrspace(1) @test9(ptr nocapture readonly %arg, i32 %n) {
; PROLOG-NEXT: %iv.next.prol = add nuw nsw i64 %phi.prol, 1
; PROLOG-NEXT: %prol.iter.next = add i32 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i32 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %header.prol, label %header.prol.loopexit.unr-lcssa, !llvm.loop !13
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %header.prol, label %header.prol.loopexit.unr-lcssa, !llvm.loop !12
; PROLOG: header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %phi.unr.ph = phi i64 [ %iv.next.prol, %latch.prol ]
; PROLOG-NEXT: br label %header.prol.loopexit
@@ -5247,7 +5077,7 @@ define void @test10(i64 %trip, i64 %trip2) {
; EPILOG-NEXT: %cmp.epil = icmp ne i64 %iv_next.epil, %trip
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit2.epilog-lcssa, !llvm.loop !14
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit2.epilog-lcssa, !llvm.loop !13
; EPILOG: exit2.epilog-lcssa:
; EPILOG-NEXT: br label %exit2
; EPILOG: exit2:
@@ -5321,7 +5151,7 @@ define void @test10(i64 %trip, i64 %trip2) {
; PROLOG-NEXT: %cmp.prol = icmp ne i64 %iv_next.prol, %trip
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !14
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !13
; PROLOG: loop_header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %iv.unr.ph = phi i64 [ %iv_next.prol, %loop_latch.prol ]
; PROLOG-NEXT: br label %loop_header.prol.loopexit
@@ -5520,7 +5350,7 @@ define void @test11(i64 %trip, i1 %cond) {
; EPILOG-NEXT: %cmp.epil = icmp ne i64 %iv_next.epil, %trip
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit2.epilog-lcssa, !llvm.loop !15
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit2.epilog-lcssa, !llvm.loop !14
; EPILOG: exit2.epilog-lcssa:
; EPILOG-NEXT: br label %exit2
; EPILOG: exit2:
@@ -5587,7 +5417,7 @@ define void @test11(i64 %trip, i1 %cond) {
; PROLOG-NEXT: %cmp.prol = icmp ne i64 %iv_next.prol, %trip
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !15
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !14
; PROLOG: loop_header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %iv.unr.ph = phi i64 [ %iv_next.prol, %loop_latch.prol ]
; PROLOG-NEXT: br label %loop_header.prol.loopexit
@@ -5794,7 +5624,7 @@ define void @test12(i64 %trip, i64 %trip2, i1 %cond) {
; EPILOG-NEXT: %cmp.epil = icmp ne i64 %iv_next.epil, %trip
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit1.epilog-lcssa.loopexit1, !llvm.loop !16
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit1.epilog-lcssa.loopexit1, !llvm.loop !15
; EPILOG: exit1.epilog-lcssa.loopexit:
; EPILOG-NEXT: br label %exit1.epilog-lcssa
; EPILOG: exit1.epilog-lcssa.loopexit1:
@@ -5880,7 +5710,7 @@ define void @test12(i64 %trip, i64 %trip2, i1 %cond) {
; PROLOG-NEXT: %cmp.prol = icmp ne i64 %iv_next.prol, %trip
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !16
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !15
; PROLOG: loop_header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %iv.unr.ph = phi i64 [ %iv_next.prol, %loop_latch.prol ]
; PROLOG-NEXT: br label %loop_header.prol.loopexit
@@ -6135,7 +5965,7 @@ define void @test13(i64 %trip, i64 %trip2) {
; EPILOG-NEXT: %cmp.epil = icmp ne i64 %iv_next.epil, %trip
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit1.epilog-lcssa.loopexit1, !llvm.loop !17
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit1.epilog-lcssa.loopexit1, !llvm.loop !16
; EPILOG: exit1.epilog-lcssa.loopexit:
; EPILOG-NEXT: br label %exit1.epilog-lcssa
; EPILOG: exit1.epilog-lcssa.loopexit1:
@@ -6225,7 +6055,7 @@ define void @test13(i64 %trip, i64 %trip2) {
; PROLOG-NEXT: %cmp.prol = icmp ne i64 %iv_next.prol, %trip
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !17
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !16
; PROLOG: loop_header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %iv.unr.ph = phi i64 [ %iv_next.prol, %loop_latch.prol ]
; PROLOG-NEXT: br label %loop_header.prol.loopexit
@@ -6474,7 +6304,7 @@ define void @test14(i64 %trip, i1 %cond) {
; EPILOG-NEXT: %cmp.epil = icmp ne i64 %iv_next.epil, %trip
; EPILOG-NEXT: %epil.iter.next = add i64 %epil.iter, 1
; EPILOG-NEXT: %epil.iter.cmp = icmp ne i64 %epil.iter.next, %xtraiter
-; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit1.epilog-lcssa.loopexit1, !llvm.loop !18
+; EPILOG-NEXT: br i1 %epil.iter.cmp, label %loop_header.epil, label %exit1.epilog-lcssa.loopexit1, !llvm.loop !17
; EPILOG: exit1.epilog-lcssa.loopexit:
; EPILOG-NEXT: br label %exit1.epilog-lcssa
; EPILOG: exit1.epilog-lcssa.loopexit1:
@@ -6557,7 +6387,7 @@ define void @test14(i64 %trip, i1 %cond) {
; PROLOG-NEXT: %cmp.prol = icmp ne i64 %iv_next.prol, %trip
; PROLOG-NEXT: %prol.iter.next = add i64 %prol.iter, 1
; PROLOG-NEXT: %prol.iter.cmp = icmp ne i64 %prol.iter.next, %xtraiter
-; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !18
+; PROLOG-NEXT: br i1 %prol.iter.cmp, label %loop_header.prol, label %loop_header.prol.loopexit.unr-lcssa, !llvm.loop !17
; PROLOG: loop_header.prol.loopexit.unr-lcssa:
; PROLOG-NEXT: %iv.unr.ph = phi i64 [ %iv_next.prol, %loop_latch.prol ]
; PROLOG-NEXT: br label %loop_header.prol.loopexit
diff --git a/llvm/test/Transforms/LoopUnroll/scevunroll.ll b/llvm/test/Transforms/LoopUnroll/scevunroll.ll
index b6b14e365cc1d..aa33b979bf17e 100644
--- a/llvm/test/Transforms/LoopUnroll/scevunroll.ll
+++ b/llvm/test/Transforms/LoopUnroll/scevunroll.ll
@@ -310,16 +310,16 @@ define void @nsw_latch(ptr %a) nounwind {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: br label [[FOR_COND:%.*]]
-; CHECK: for.cond:
-; CHECK-NEXT: br i1 false, label [[RETURN:%.*]], label [[FOR_BODY_1:%.*]]
-; CHECK: for.body.1:
-; CHECK-NEXT: br i1 false, label [[FOR_COND_1:%.*]], label [[RETURN]]
-; CHECK: for.cond.1:
-; CHECK-NEXT: br label [[RETURN]]
+; CHECK-NEXT: [[B_03:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
+; CHECK-NEXT: [[TOBOOL:%.*]] = icmp eq i32 [[B_03]], 0
+; CHECK-NEXT: [[ADD]] = add nuw nsw i32 [[B_03]], 8
+; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = xor i1 [[TOBOOL]], true
+; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[TOBOOL]], i32 [[B_03]], i32 [[B_03]]
+; CHECK-NEXT: [[TMP1:%.*]] = select i1 [[TOBOOL]], i32 0, i32 1
+; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[TOBOOL_NOT]], i1 true, i1 false
+; CHECK-NEXT: br i1 [[OR_COND]], label [[RETURN:%.*]], label [[FOR_BODY]]
; CHECK: return:
-; CHECK-NEXT: [[B_03_LCSSA:%.*]] = phi i32 [ 0, [[FOR_COND]] ], [ 8, [[FOR_BODY_1]] ], [ 0, [[FOR_COND_1]] ]
-; CHECK-NEXT: [[RETVAL_0:%.*]] = phi i32 [ 0, [[FOR_COND]] ], [ 1, [[FOR_BODY_1]] ], [ 0, [[FOR_COND_1]] ]
+; CHECK-NEXT: [[B_03_LCSSA:%.*]] = phi i32 [ [[TMP0]], [[FOR_BODY]] ]
; CHECK-NEXT: store i32 [[B_03_LCSSA]], ptr [[A:%.*]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LoopUnroll/unroll-after-peel.ll b/llvm/test/Transforms/LoopUnroll/unroll-after-peel.ll
index 504abafd23bf5..a1af3593e67b6 100644
--- a/llvm/test/Transforms/LoopUnroll/unroll-after-peel.ll
+++ b/llvm/test/Transforms/LoopUnroll/unroll-after-peel.ll
@@ -8,10 +8,10 @@ define i64 @hoge(i1 %c) {
; CHECK: bb1.peel.begin:
; CHECK-NEXT: br label [[BB1_PEEL:%.*]]
; CHECK: bb1.peel:
-; CHECK-NEXT: br i1 [[C:%.*]], label [[BB2_PEEL:%.*]], label [[BB4:%.*]]
-; CHECK: bb2.peel:
; CHECK-NEXT: [[TMP3_PEEL:%.*]] = icmp slt i32 0, 9
-; CHECK-NEXT: br i1 [[TMP3_PEEL]], label [[BB1_PEEL_NEXT:%.*]], label [[BB4]]
+; CHECK-NEXT: %0 = select i1 [[C:%.*]], i32 8, i32 0
+; CHECK-NEXT: [[OR_COND_PEEL:%.*]] = and i1 [[C]], [[TMP3_PEEL]]
+; CHECK-NEXT: br i1 [[OR_COND_PEEL]], label [[BB1_PEEL_NEXT:%.*]], label [[BB4:%.*]]
; CHECK: bb1.peel.next:
; CHECK-NEXT: br label [[BB1_PEEL_NEXT1:%.*]]
; CHECK: bb1.peel.next1:
@@ -21,10 +21,8 @@ define i64 @hoge(i1 %c) {
; CHECK: bb1:
; CHECK-NEXT: br i1 [[C]], label [[BB1]], label [[BB4_LOOPEXIT:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: bb4.loopexit:
-; CHECK-NEXT: [[TMP5_PH:%.*]] = phi i32 [ 8, [[BB1]] ]
; CHECK-NEXT: br label [[BB4]]
; CHECK: bb4:
-; CHECK-NEXT: [[TMP5:%.*]] = phi i32 [ 0, [[BB1_PEEL]] ], [ 8, [[BB2_PEEL]] ], [ [[TMP5_PH]], [[BB4_LOOPEXIT]] ]
; CHECK-NEXT: [[TMP6:%.*]] = call i64 (...) @llvm.experimental.deoptimize.i64(i32 10) [ "deopt"() ]
; CHECK-NEXT: ret i64 [[TMP6]]
;
diff --git a/llvm/test/Transforms/LoopUnroll/unroll-header-exiting-with-phis-multiple-exiting-blocks.ll b/llvm/test/Transforms/LoopUnroll/unroll-header-exiting-with-phis-multiple-exiting-blocks.ll
index 91f2e6a1f25ad..c1680dd274d00 100644
--- a/llvm/test/Transforms/LoopUnroll/unroll-header-exiting-with-phis-multiple-exiting-blocks.ll
+++ b/llvm/test/Transforms/LoopUnroll/unroll-header-exiting-with-phis-multiple-exiting-blocks.ll
@@ -8,52 +8,25 @@ define i16 @full_unroll_multiple_exiting_blocks(ptr %A, i16 %x, i16 %y) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[HEADER:%.*]]
; CHECK: header:
-; CHECK-NEXT: [[LV:%.*]] = load i16, ptr [[A:%.*]], align 2
-; CHECK-NEXT: [[RES_NEXT:%.*]] = add i16 123, [[LV]]
-; CHECK-NEXT: br label [[EXITING_1:%.*]]
-; CHECK: exiting.1:
-; CHECK-NEXT: [[EC_1:%.*]] = icmp eq i16 [[LV]], [[X:%.*]]
-; CHECK-NEXT: br i1 [[EC_1]], label [[EXIT:%.*]], label [[EXITING_2:%.*]]
-; CHECK: exiting.2:
-; CHECK-NEXT: [[EC_2:%.*]] = icmp eq i16 [[LV]], [[Y:%.*]]
-; CHECK-NEXT: br i1 [[EC_2]], label [[EXIT]], label [[LATCH:%.*]]
-; CHECK: latch:
-; CHECK-NEXT: [[PTR_1:%.*]] = getelementptr inbounds i16, ptr [[A]], i64 1
+; CHECK-NEXT: [[RES_NEXT:%.*]] = phi i16 [ 123, [[ENTRY:%.*]] ], [ [[RES_NEXT_1:%.*]], [[LATCH_3:%.*]] ]
+; CHECK-NEXT: [[I_0:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[INC9:%.*]], [[LATCH_3]] ]
+; CHECK-NEXT: [[PTR_1:%.*]] = getelementptr inbounds i16, ptr [[A:%.*]], i64 [[I_0]]
; CHECK-NEXT: [[LV_1:%.*]] = load i16, ptr [[PTR_1]], align 2
-; CHECK-NEXT: [[RES_NEXT_1:%.*]] = add i16 [[RES_NEXT]], [[LV_1]]
-; CHECK-NEXT: br label [[EXITING_1_1:%.*]]
-; CHECK: exiting.1.1:
-; CHECK-NEXT: [[EC_1_1:%.*]] = icmp eq i16 [[LV_1]], [[X]]
-; CHECK-NEXT: br i1 [[EC_1_1]], label [[EXIT]], label [[EXITING_2_1:%.*]]
-; CHECK: exiting.2.1:
-; CHECK-NEXT: [[EC_2_1:%.*]] = icmp eq i16 [[LV_1]], [[Y]]
-; CHECK-NEXT: br i1 [[EC_2_1]], label [[EXIT]], label [[LATCH_1:%.*]]
-; CHECK: latch.1:
-; CHECK-NEXT: [[PTR_2:%.*]] = getelementptr inbounds i16, ptr [[A]], i64 2
-; CHECK-NEXT: [[LV_2:%.*]] = load i16, ptr [[PTR_2]], align 2
-; CHECK-NEXT: [[RES_NEXT_2:%.*]] = add i16 [[RES_NEXT_1]], [[LV_2]]
-; CHECK-NEXT: br label [[EXITING_1_2:%.*]]
-; CHECK: exiting.1.2:
-; CHECK-NEXT: [[EC_1_2:%.*]] = icmp eq i16 [[LV_2]], [[X]]
-; CHECK-NEXT: br i1 [[EC_1_2]], label [[EXIT]], label [[EXITING_2_2:%.*]]
-; CHECK: exiting.2.2:
-; CHECK-NEXT: [[EC_2_2:%.*]] = icmp eq i16 [[LV_2]], [[Y]]
-; CHECK-NEXT: br i1 [[EC_2_2]], label [[EXIT]], label [[LATCH_2:%.*]]
-; CHECK: latch.2:
-; CHECK-NEXT: [[PTR_3:%.*]] = getelementptr inbounds i16, ptr [[A]], i64 3
-; CHECK-NEXT: [[LV_3:%.*]] = load i16, ptr [[PTR_3]], align 2
-; CHECK-NEXT: [[RES_NEXT_3:%.*]] = add i16 [[RES_NEXT_2]], [[LV_3]]
-; CHECK-NEXT: br i1 false, label [[EXITING_1_3:%.*]], label [[EXIT]]
-; CHECK: exiting.1.3:
-; CHECK-NEXT: [[EC_1_3:%.*]] = icmp eq i16 [[LV_3]], [[X]]
-; CHECK-NEXT: br i1 [[EC_1_3]], label [[EXIT]], label [[EXITING_2_3:%.*]]
-; CHECK: exiting.2.3:
-; CHECK-NEXT: [[EC_2_3:%.*]] = icmp eq i16 [[LV_3]], [[Y]]
-; CHECK-NEXT: br i1 [[EC_2_3]], label [[EXIT]], label [[LATCH_3:%.*]]
-; CHECK: latch.3:
-; CHECK-NEXT: unreachable
+; CHECK-NEXT: [[RES_NEXT_1]] = add i16 [[RES_NEXT]], [[LV_1]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ult i64 [[I_0]], 3
+; CHECK-NEXT: [[CMP_NOT:%.*]] = xor i1 [[CMP]], true
+; CHECK-NEXT: [[EC_1_1:%.*]] = icmp eq i16 [[LV_1]], [[X:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[CMP]], i16 0, i16 [[RES_NEXT_1]]
+; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[CMP_NOT]], i1 true, i1 [[EC_1_1]]
+; CHECK-NEXT: [[EC_2_1:%.*]] = icmp eq i16 [[LV_1]], [[Y:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = select i1 [[OR_COND]], i16 [[TMP0]], i16 1
+; CHECK-NEXT: [[EC_2_3:%.*]] = select i1 [[OR_COND]], i1 true, i1 [[EC_2_1]]
+; CHECK-NEXT: br i1 [[EC_2_3]], label [[EXIT:%.*]], label [[LATCH_3]]
+; CHECK: latch:
+; CHECK-NEXT: [[INC9]] = add i64 [[I_0]], 1
+; CHECK-NEXT: br label [[HEADER]]
; CHECK: exit:
-; CHECK-NEXT: [[RES_LCSSA:%.*]] = phi i16 [ 0, [[EXITING_1]] ], [ 1, [[EXITING_2]] ], [ 0, [[EXITING_1_1]] ], [ 1, [[EXITING_2_1]] ], [ 0, [[EXITING_1_2]] ], [ 1, [[EXITING_2_2]] ], [ [[RES_NEXT_3]], [[LATCH_2]] ], [ 0, [[EXITING_1_3]] ], [ 1, [[EXITING_2_3]] ]
+; CHECK-NEXT: [[RES_LCSSA:%.*]] = phi i16 [ [[TMP1]], [[HEADER]] ]
; CHECK-NEXT: ret i16 [[RES_LCSSA]]
;
entry:
diff --git a/llvm/test/Transforms/LoopVectorize/early_exit_legality.ll b/llvm/test/Transforms/LoopVectorize/early_exit_legality.ll
index e9b8e8cdda526..df7fd4e1b7d55 100644
--- a/llvm/test/Transforms/LoopVectorize/early_exit_legality.ll
+++ b/llvm/test/Transforms/LoopVectorize/early_exit_legality.ll
@@ -323,7 +323,9 @@ loop.end:
; We don't currently support multiple uncountable early exits.
define i64 @multiple_uncountable_exits() {
; CHECK-LABEL: LV: Checking a loop in 'multiple_uncountable_exits'
-; CHECK: LV: Not vectorizing: Loop has too many uncountable exits.
+; CHECK: LV: Found an early exit loop with symbolic max backedge taken count: 63
+; CHECK-NEXT: LV: We can vectorize this loop!
+; CHECK-NOT: LV: Not vectorizing:
entry:
%p1 = alloca [1024 x i8]
%p2 = alloca [1024 x i8]
diff --git a/llvm/test/Transforms/LoopVectorize/single_early_exit_live_outs.ll b/llvm/test/Transforms/LoopVectorize/single_early_exit_live_outs.ll
index a21666a31b6a2..579818d7b7a64 100644
--- a/llvm/test/Transforms/LoopVectorize/single_early_exit_live_outs.ll
+++ b/llvm/test/Transforms/LoopVectorize/single_early_exit_live_outs.ll
@@ -1057,46 +1057,19 @@ define ptr @same_exit_block_post_inc_use1_ivptr() {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = alloca [1024 x i8], align 1
; CHECK-NEXT: call void @init_mem(ptr [[P1]], i64 1024)
-; CHECK-NEXT: [[PTREND:%.*]] = getelementptr i8, ptr [[P1]], i64 1024
-; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK: vector.ph:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P1]], i64 1024
-; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[P1]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP5]], align 1
-; CHECK-NEXT: [[TMP15:%.*]] = icmp ne <4 x i8> [[WIDE_LOAD]], splat (i8 72)
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP16:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP15]])
-; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
-; CHECK-NEXT: [[TMP18:%.*]] = or i1 [[TMP16]], [[TMP17]]
-; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
-; CHECK: middle.split:
-; CHECK-NEXT: br i1 [[TMP16]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
-; CHECK: middle.block:
-; CHECK-NEXT: br i1 true, label [[LOOP_END:%.*]], label [[SCALAR_PH]]
-; CHECK: vector.early.exit:
-; CHECK-NEXT: [[FIRST_ACTIVE_LANE:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP15]], i1 true)
-; CHECK-NEXT: [[TMP8:%.*]] = add i64 [[INDEX]], [[FIRST_ACTIVE_LANE]]
-; CHECK-NEXT: [[TMP9:%.*]] = add i64 [[TMP8]], 1
-; CHECK-NEXT: [[EARLY_EXIT_VALUE:%.*]] = getelementptr i8, ptr [[P1]], i64 [[TMP9]]
-; CHECK-NEXT: br label [[LOOP_END]]
-; CHECK: scalar.ph:
-; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi ptr [ [[TMP0]], [[MIDDLE_BLOCK]] ], [ [[P1]], [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
-; CHECK-NEXT: [[PTR:%.*]] = phi ptr [ [[PTR_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
+; CHECK-NEXT: [[PTR:%.*]] = phi ptr [ [[P1]], [[ENTRY:%.*]] ], [ [[PTR_NEXT:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[LD1:%.*]] = load i8, ptr [[PTR]], align 1
; CHECK-NEXT: [[PTR_NEXT]] = getelementptr inbounds i8, ptr [[PTR]], i64 1
; CHECK-NEXT: [[CMP3:%.*]] = icmp eq i8 [[LD1]], 72
-; CHECK-NEXT: br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_END]]
-; CHECK: loop.inc:
-; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne ptr [[PTR_NEXT]], [[PTREND]]
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP29:![0-9]+]]
+; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne ptr [[PTR_NEXT]], [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = select i1 [[CMP3]], ptr [[TMP0]], ptr [[PTR_NEXT]]
+; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[CMP3]], i1 [[EXITCOND]], i1 false
+; CHECK-NEXT: br i1 [[OR_COND]], label [[LOOP]], label [[LOOP_END:%.*]]
; CHECK: loop.end:
-; CHECK-NEXT: [[RETVAL:%.*]] = phi ptr [ [[PTR_NEXT]], [[LOOP]] ], [ [[PTREND]], [[LOOP_INC]] ], [ [[PTREND]], [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
+; CHECK-NEXT: [[RETVAL:%.*]] = phi ptr [ [[TMP1]], [[LOOP]] ]
; CHECK-NEXT: ret ptr [[RETVAL]]
;
entry:
@@ -1129,51 +1102,21 @@ define i64 @same_exit_block_post_inc_use2() {
; CHECK-NEXT: [[P2:%.*]] = alloca [1024 x i8], align 1
; CHECK-NEXT: call void @init_mem(ptr [[P1]], i64 1024)
; CHECK-NEXT: call void @init_mem(ptr [[P2]], i64 1024)
-; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK: vector.ph:
-; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 3, [[INDEX1]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP5]], align 1
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP6]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i8>, ptr [[TMP7]], align 1
-; CHECK-NEXT: [[TMP17:%.*]] = icmp ne <4 x i8> [[WIDE_LOAD]], [[WIDE_LOAD2]]
-; CHECK-NEXT: [[INDEX_NEXT3]] = add nuw i64 [[INDEX1]], 4
-; CHECK-NEXT: [[TMP18:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP17]])
-; CHECK-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
-; CHECK-NEXT: [[TMP20:%.*]] = or i1 [[TMP18]], [[TMP19]]
-; CHECK-NEXT: br i1 [[TMP20]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
-; CHECK: middle.split:
-; CHECK-NEXT: br i1 [[TMP18]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
-; CHECK: middle.block:
-; CHECK-NEXT: br i1 true, label [[LOOP_END:%.*]], label [[SCALAR_PH]]
-; CHECK: vector.early.exit:
-; CHECK-NEXT: [[FIRST_ACTIVE_LANE:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP17]], i1 true)
-; CHECK-NEXT: [[TMP10:%.*]] = add i64 [[INDEX1]], [[FIRST_ACTIVE_LANE]]
-; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[TMP10]], 1
-; CHECK-NEXT: [[EARLY_EXIT_VALUE:%.*]] = add i64 3, [[TMP11]]
-; CHECK-NEXT: br label [[LOOP_END]]
-; CHECK: scalar.ph:
-; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 67, [[MIDDLE_BLOCK]] ], [ 3, [[ENTRY:%.*]] ]
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 3, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: [[LD1:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX]]
; CHECK-NEXT: [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[CMP3:%.*]] = icmp eq i8 [[LD1]], [[LD2]]
-; CHECK-NEXT: br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_END]]
-; CHECK: loop.inc:
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP31:![0-9]+]]
+; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[CMP3]], i64 [[INDEX]], i64 [[INDEX_NEXT]]
+; CHECK-NEXT: [[OR_COND:%.*]] = and i1 [[CMP3]], [[EXITCOND]]
+; CHECK-NEXT: br i1 [[OR_COND]], label [[LOOP]], label [[LOOP_END:%.*]]
; CHECK: loop.end:
-; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[INDEX_NEXT]], [[LOOP]] ], [ [[INDEX]], [[LOOP_INC]] ], [ 66, [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
+; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[TMP0]], [[LOOP]] ]
; CHECK-NEXT: ret i64 [[RETVAL]]
;
entry:
@@ -1227,7 +1170,7 @@ define i64 @diff_exit_block_pre_inc_use1() {
; CHECK-NEXT: [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP6]])
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
; CHECK-NEXT: [[TMP9:%.*]] = or i1 [[TMP7]], [[TMP8]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP7]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1251,7 +1194,7 @@ define i64 @diff_exit_block_pre_inc_use1() {
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP33:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP29:![0-9]+]]
; CHECK: loop.early.exit:
; CHECK-NEXT: [[RETVAL1:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL1]]
@@ -1314,7 +1257,7 @@ define i64 @diff_exit_block_pre_inc_use2() {
; CHECK-NEXT: [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP6]])
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
; CHECK-NEXT: [[TMP9:%.*]] = or i1 [[TMP7]], [[TMP8]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP7]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1335,7 +1278,7 @@ define i64 @diff_exit_block_pre_inc_use2() {
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP35:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP31:![0-9]+]]
; CHECK: loop.early.exit:
; CHECK-NEXT: [[RETVAL1:%.*]] = phi i64 [ 67, [[LOOP]] ], [ 67, [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL1]]
@@ -1398,7 +1341,7 @@ define i64 @diff_exit_block_pre_inc_use3() {
; CHECK-NEXT: [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP6]])
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 64
; CHECK-NEXT: [[TMP9:%.*]] = or i1 [[TMP7]], [[TMP8]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP7]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1422,7 +1365,7 @@ define i64 @diff_exit_block_pre_inc_use3() {
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP37:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP33:![0-9]+]]
; CHECK: loop.early.exit:
; CHECK-NEXT: [[INDEX_LCSSA:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[INDEX_LCSSA]]
@@ -1483,7 +1426,7 @@ define i64 @diff_exit_block_post_inc_use1() {
; CHECK-NEXT: [[TMP14:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP13]])
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
; CHECK-NEXT: [[TMP16:%.*]] = or i1 [[TMP14]], [[TMP15]]
-; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP38:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP14]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1507,7 +1450,7 @@ define i64 @diff_exit_block_post_inc_use1() {
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP39:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP35:![0-9]+]]
; CHECK: loop.early.exit:
; CHECK-NEXT: [[RETVAL1:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL1]]
@@ -1570,7 +1513,7 @@ define i64 @diff_exit_block_post_inc_use2() {
; CHECK-NEXT: [[TMP18:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP17]])
; CHECK-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
; CHECK-NEXT: [[TMP20:%.*]] = or i1 [[TMP18]], [[TMP19]]
-; CHECK-NEXT: br i1 [[TMP20]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP40:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP20]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP18]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1595,7 +1538,7 @@ define i64 @diff_exit_block_post_inc_use2() {
; CHECK-NEXT: br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_EARLY_EXIT]]
; CHECK: loop.inc:
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP41:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP37:![0-9]+]]
; CHECK: loop.early.exit:
; CHECK-NEXT: [[RETVAL1:%.*]] = phi i64 [ [[INDEX_NEXT]], [[LOOP]] ], [ [[TMP21]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL1]]
@@ -1659,7 +1602,7 @@ define i64 @diff_exit_block_post_inc_use3(i64 %start) {
; CHECK-NEXT: [[TMP20:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP19]])
; CHECK-NEXT: [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 64
; CHECK-NEXT: [[TMP22:%.*]] = or i1 [[TMP20]], [[TMP21]]
-; CHECK-NEXT: br i1 [[TMP22]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP42:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP22]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP38:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP20]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1688,7 +1631,7 @@ define i64 @diff_exit_block_post_inc_use3(i64 %start) {
; CHECK-NEXT: br i1 [[CMP3]], label [[LOOP_INC]], label [[LOOP_EARLY_EXIT]]
; CHECK: loop.inc:
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP43:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP39:![0-9]+]]
; CHECK: loop.early.exit:
; CHECK-NEXT: [[RETVAL1:%.*]] = phi i64 [ [[INDEX2_NEXT]], [[LOOP]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL1]]
@@ -1751,7 +1694,7 @@ define i64 @loop_contains_safe_call() {
; CHECK-NEXT: [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT2]], 64
; CHECK-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]]
-; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP44:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP40:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP6]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1774,7 +1717,7 @@ define i64 @loop_contains_safe_call() {
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP45:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP41:![0-9]+]]
; CHECK: loop.end:
; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ 67, [[LOOP_INC]] ], [ 67, [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL]]
@@ -1827,7 +1770,7 @@ define i64 @loop_contains_safe_div() {
; CHECK-NEXT: [[TMP6:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP5]])
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT2]], 64
; CHECK-NEXT: [[TMP8:%.*]] = or i1 [[TMP6]], [[TMP7]]
-; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP46:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP42:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP6]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -1850,7 +1793,7 @@ define i64 @loop_contains_safe_div() {
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP47:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP43:![0-9]+]]
; CHECK: loop.end:
; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ 67, [[LOOP_INC]] ], [ 67, [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL]]
@@ -1904,7 +1847,7 @@ define i64 @loop_contains_load_after_early_exit(ptr dereferenceable(1024) align(
; CHECK-NEXT: [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP6]])
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
; CHECK-NEXT: [[TMP9:%.*]] = or i1 [[TMP7]], [[TMP8]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP48:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP44:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x i64> [[WIDE_LOAD2]], i32 3
; CHECK-NEXT: br i1 [[TMP7]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
@@ -1929,7 +1872,7 @@ define i64 @loop_contains_load_after_early_exit(ptr dereferenceable(1024) align(
; CHECK-NEXT: [[LD2:%.*]] = load i64, ptr [[ARRAYIDX2]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP49:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP45:![0-9]+]]
; CHECK: loop.end:
; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ [[LD2]], [[LOOP_INC]] ], [ [[TMP10]], [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL]]
@@ -1987,7 +1930,7 @@ define i64 @same_exit_block_pre_inc_use1_reverse() {
; CHECK-NEXT: [[TMP9:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP8]])
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT4]], 1020
; CHECK-NEXT: [[TMP11:%.*]] = or i1 [[TMP9]], [[TMP10]]
-; CHECK-NEXT: br i1 [[TMP11]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP50:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP11]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP46:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP9]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -2011,7 +1954,7 @@ define i64 @same_exit_block_pre_inc_use1_reverse() {
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], -1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDEX_NEXT]], 0
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP_END]], label [[LOOP1]], !llvm.loop [[LOOP51:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP_END]], label [[LOOP1]], !llvm.loop [[LOOP47:![0-9]+]]
; CHECK: loop.end:
; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP1]] ], [ 1024, [[LOOP_INC]] ], [ 1024, [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL]]
@@ -2114,7 +2057,7 @@ define i64 @same_exit_block_pre_inc_use1_deref_ptrs(ptr dereferenceable(1024) %p
; CHECK-NEXT: [[TMP7:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP6]])
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
; CHECK-NEXT: [[TMP9:%.*]] = or i1 [[TMP7]], [[TMP8]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP52:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_SPLIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP48:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP7]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
; CHECK: middle.block:
@@ -2138,7 +2081,7 @@ define i64 @same_exit_block_pre_inc_use1_deref_ptrs(ptr dereferenceable(1024) %p
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP53:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[LOOP]], label [[LOOP_END]], !llvm.loop [[LOOP49:![0-9]+]]
; CHECK: loop.end:
; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[LOOP]] ], [ 67, [[LOOP_INC]] ], [ 67, [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL]]
@@ -2220,8 +2163,4 @@ attributes #0 = { "vector-function-abi-variant"="_ZGVsNxv_foo(foo_vec)" }
; CHECK: [[LOOP47]] = distinct !{[[LOOP47]], [[META2]], [[META1]]}
; CHECK: [[LOOP48]] = distinct !{[[LOOP48]], [[META1]], [[META2]]}
; CHECK: [[LOOP49]] = distinct !{[[LOOP49]], [[META2]], [[META1]]}
-; CHECK: [[LOOP50]] = distinct !{[[LOOP50]], [[META1]], [[META2]]}
-; CHECK: [[LOOP51]] = distinct !{[[LOOP51]], [[META2]], [[META1]]}
-; CHECK: [[LOOP52]] = distinct !{[[LOOP52]], [[META1]], [[META2]]}
-; CHECK: [[LOOP53]] = distinct !{[[LOOP53]], [[META2]], [[META1]]}
;.
diff --git a/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll b/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll
index 3cd60e8d7f592..1414c403d4aad 100644
--- a/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll
+++ b/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll
@@ -65,24 +65,57 @@ define i64 @multiple_uncountable_exits() {
; CHECK-NEXT: [[P2:%.*]] = alloca [1024 x i8], align 1
; CHECK-NEXT: call void @init_mem(ptr [[P1]], i64 1024)
; CHECK-NEXT: call void @init_mem(ptr [[P2]], i64 1024)
+; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK: vector.ph:
; CHECK-NEXT: br label [[SEARCH1:%.*]]
+; CHECK: vector.body:
+; CHECK-NEXT: [[INDEX1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT3:%.*]], [[SEARCH1]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 3, i64 4, i64 5, i64 6>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[SEARCH1]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 3, [[INDEX1]]
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i32 0
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP1]], align 1
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i32 0
+; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i8>, ptr [[TMP3]], align 1
+; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x i8> [[WIDE_LOAD]], [[WIDE_LOAD2]]
+; CHECK-NEXT: [[TMP5:%.*]] = icmp ult <4 x i8> [[WIDE_LOAD]], splat (i8 34)
+; CHECK-NEXT: [[TMP6:%.*]] = select <4 x i1> [[TMP4]], <4 x i64> [[VEC_IND]], <4 x i64> splat (i64 100)
+; CHECK-NEXT: [[TMP7:%.*]] = or <4 x i1> [[TMP4]], [[TMP5]]
+; CHECK-NEXT: [[INDEX_NEXT3]] = add nuw i64 [[INDEX1]], 4
+; CHECK-NEXT: [[TMP8:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP7]])
+; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[TMP10:%.*]] = or i1 [[TMP8]], [[TMP9]]
+; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_SPLIT:%.*]], label [[SEARCH1]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: middle.split:
+; CHECK-NEXT: br i1 [[TMP8]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
+; CHECK: middle.block:
+; CHECK-NEXT: br i1 true, label [[LOOP_END:%.*]], label [[SCALAR_PH]]
+; CHECK: vector.early.exit:
+; CHECK-NEXT: [[FIRST_ACTIVE_LANE:%.*]] = call i64 @llvm.experimental.cttz.elts.i64.v4i1(<4 x i1> [[TMP7]], i1 true)
+; CHECK-NEXT: [[EARLY_EXIT_VALUE:%.*]] = extractelement <4 x i64> [[TMP6]], i64 [[FIRST_ACTIVE_LANE]]
+; CHECK-NEXT: br label [[LOOP_END]]
+; CHECK: scalar.ph:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 67, [[MIDDLE_BLOCK]] ], [ 3, [[ENTRY:%.*]] ]
+; CHECK-NEXT: br label [[SEARCH2:%.*]]
; CHECK: search1:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ 3, [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: [[LD1:%.*]] = load i8, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[INDEX]]
; CHECK-NEXT: [[LD2:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1
; CHECK-NEXT: [[CMP1:%.*]] = icmp eq i8 [[LD1]], [[LD2]]
-; CHECK-NEXT: br i1 [[CMP1]], label [[LOOP_END:%.*]], label [[SEARCH2:%.*]]
-; CHECK: search2:
; CHECK-NEXT: [[CMP2:%.*]] = icmp ult i8 [[LD1]], 34
-; CHECK-NEXT: br i1 [[CMP2]], label [[LOOP_END]], label [[LOOP_INC]]
+; CHECK-NEXT: [[TMP11:%.*]] = select i1 [[CMP1]], i64 [[INDEX]], i64 100
+; CHECK-NEXT: [[OR_COND:%.*]] = or i1 [[CMP1]], [[CMP2]]
+; CHECK-NEXT: br i1 [[OR_COND]], label [[LOOP_END]], label [[LOOP_INC]]
; CHECK: loop.inc:
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDEX_NEXT]], 67
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[SEARCH1]], label [[LOOP_END]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[SEARCH2]], label [[LOOP_END]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: loop.end:
-; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ [[INDEX]], [[SEARCH1]] ], [ 100, [[SEARCH2]] ], [ 43, [[LOOP_INC]] ]
+; CHECK-NEXT: [[RETVAL:%.*]] = phi i64 [ 43, [[LOOP_INC]] ], [ [[TMP11]], [[SEARCH2]] ], [ 43, [[MIDDLE_BLOCK]] ], [ [[EARLY_EXIT_VALUE]], [[VECTOR_EARLY_EXIT]] ]
; CHECK-NEXT: ret i64 [[RETVAL]]
;
entry:
@@ -179,7 +212,7 @@ define i64 @loop_contains_unsafe_call() {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[LOOP_INC:%.*]] ], [ 3, [[ENTRY:%.*]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: [[LD1:%.*]] = load i32, ptr [[ARRAYIDX]], align 1
-; CHECK-NEXT: [[BAD_CALL:%.*]] = call i32 @foo(i32 [[LD1]]) #[[ATTR1:[0-9]+]]
+; CHECK-NEXT: [[BAD_CALL:%.*]] = call i32 @foo(i32 [[LD1]]) #[[ATTR3:[0-9]+]]
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[BAD_CALL]], 34
; CHECK-NEXT: br i1 [[CMP]], label [[LOOP_INC]], label [[LOOP_END:%.*]]
; CHECK: loop.inc:
@@ -492,3 +525,9 @@ declare i32 @foo(i32) readonly
declare <vscale x 4 x i32> @foo_vec(<vscale x 4 x i32>)
attributes #0 = { "vector-function-abi-variant"="_ZGVsNxv_foo(foo_vec)" }
+;.
+; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
+; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
+; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
+; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
+;.
diff --git a/llvm/test/Transforms/SimpleLoopUnswitch/inject-invariant-conditions.ll b/llvm/test/Transforms/SimpleLoopUnswitch/inject-invariant-conditions.ll
index 536e0c6a0e74a..c856efe3a2284 100644
--- a/llvm/test/Transforms/SimpleLoopUnswitch/inject-invariant-conditions.ll
+++ b/llvm/test/Transforms/SimpleLoopUnswitch/inject-invariant-conditions.ll
@@ -1,6 +1,6 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals
-; RUN: opt < %s -S -simple-loop-unswitch-inject-invariant-conditions=true -passes="loop(simple-loop-unswitch<nontrivial>),simplifycfg" | FileCheck %s
-; RUN: opt < %s -S -simple-loop-unswitch-inject-invariant-conditions=true -passes="loop-mssa(simple-loop-unswitch<nontrivial>),simplifycfg" -verify-memoryssa | FileCheck %s
+; RUN: opt < %s -S -simple-loop-unswitch-inject-invariant-conditions=true -simplifycfg-branch-fold-threshold=0 -passes="loop(simple-loop-unswitch<nontrivial>),simplifycfg" | FileCheck %s
+; RUN: opt < %s -S -simple-loop-unswitch-inject-invariant-conditions=true -simplifycfg-branch-fold-threshold=0 -passes="loop-mssa(simple-loop-unswitch<nontrivial>),simplifycfg" -verify-memoryssa | FileCheck %s
define i32 @test_01(ptr noundef %p, i32 noundef %n, i32 noundef %limit, ptr noundef %arr, ptr noundef %x_p) {
; CHECK-LABEL: @test_01(
diff --git a/llvm/test/Transforms/SimplifyCFG/2008-07-13-InfLoopMiscompile.ll b/llvm/test/Transforms/SimplifyCFG/2008-07-13-InfLoopMiscompile.ll
index 2e9e7b19c73e2..a22e28685a124 100644
--- a/llvm/test/Transforms/SimplifyCFG/2008-07-13-InfLoopMiscompile.ll
+++ b/llvm/test/Transforms/SimplifyCFG/2008-07-13-InfLoopMiscompile.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S | FileCheck %s
+; RUN: opt < %s -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -simplifycfg-branch-fold-threshold=0 -S | FileCheck %s
; PR2540
; Outval should end up with a select from 0/2, not all constants.
diff --git a/llvm/test/Transforms/SimplifyCFG/X86/SpeculativeExec.ll b/llvm/test/Transforms/SimplifyCFG/X86/SpeculativeExec.ll
index 2de1c72c4007a..79248eacb401a 100644
--- a/llvm/test/Transforms/SimplifyCFG/X86/SpeculativeExec.ll
+++ b/llvm/test/Transforms/SimplifyCFG/X86/SpeculativeExec.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -phi-node-folding-threshold=2 -S | FileCheck %s
+; RUN: opt < %s -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -phi-node-folding-threshold=2 -simplifycfg-branch-fold-threshold=0 -S | FileCheck %s
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
diff --git a/llvm/test/Transforms/SimplifyCFG/X86/switch_to_lookup_table.ll b/llvm/test/Transforms/SimplifyCFG/X86/switch_to_lookup_table.ll
index 9124b356a46bf..28d0aae05b84f 100644
--- a/llvm/test/Transforms/SimplifyCFG/X86/switch_to_lookup_table.ll
+++ b/llvm/test/Transforms/SimplifyCFG/X86/switch_to_lookup_table.ll
@@ -1464,9 +1464,10 @@ define i32 @no_reuse_cmp2(i32 %x, i32 %y) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[EC:%.*]] = icmp ne i32 [[Y:%.*]], 0
; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[X:%.*]], 4
+; CHECK-NEXT: [[TMP1:%.*]] = select i1 [[EC]], i32 0, i32 100
+; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[EC]], i1 [[TMP0]], i1 false
; CHECK-NEXT: [[SWITCH_OFFSET:%.*]] = add nsw i32 [[X]], 10
-; CHECK-NEXT: [[SPEC_SELECT:%.*]] = select i1 [[TMP0]], i32 [[SWITCH_OFFSET]], i32 0
-; CHECK-NEXT: [[R_0:%.*]] = select i1 [[EC]], i32 [[SPEC_SELECT]], i32 100
+; CHECK-NEXT: [[R_0:%.*]] = select i1 [[OR_COND]], i32 [[SWITCH_OFFSET]], i32 [[TMP1]]
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[R_0]], 0
; CHECK-NEXT: [[DOTR_0:%.*]] = select i1 [[CMP]], i32 100, i32 [[R_0]]
; CHECK-NEXT: ret i32 [[DOTR_0]]
diff --git a/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest-phis.ll b/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest-phis.ll
index 150bdfcff92a1..7f7274fa53430 100644
--- a/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest-phis.ll
+++ b/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest-phis.ll
@@ -12,18 +12,16 @@ define void @incompatible_ivs_of_single_phi.falsedest.falsedest(i8 %v0, i8 %v1,
; ALL-LABEL: @incompatible_ivs_of_single_phi.falsedest.falsedest(
; ALL-NEXT: pred:
; ALL-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
-; ALL-NEXT: br i1 [[C0]], label [[DISPATCH:%.*]], label [[FINAL_RIGHT:%.*]]
-; ALL: dispatch:
; ALL-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; ALL-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT]]
+; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = select i1 [[C0]], i8 [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], i8 [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]]
+; ALL-NEXT: [[OR_COND:%.*]] = select i1 [[C0]], i1 [[C1]], i1 false
+; ALL-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT:%.*]]
; ALL: common.ret:
; ALL-NEXT: ret void
; ALL: final_left:
-; ALL-NEXT: [[FINAL_LEFT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_LEFT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
-; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI]])
+; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI:%.*]])
; ALL-NEXT: br label [[COMMON_RET:%.*]]
; ALL: final_right:
-; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], [[PRED:%.*]] ], [ [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
; ALL-NEXT: call void @sideeffect1(i8 [[FINAL_RIGHT_PHI]])
; ALL-NEXT: br label [[COMMON_RET]]
;
@@ -48,18 +46,16 @@ define void @incompatible_ivs_of_single_phi.falsedest.falsedest.extrause(i8 %v0,
; ALL-NEXT: pred:
; ALL-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
; ALL-NEXT: call void @use1(i1 [[C0]])
-; ALL-NEXT: br i1 [[C0]], label [[DISPATCH:%.*]], label [[FINAL_RIGHT:%.*]]
-; ALL: dispatch:
; ALL-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; ALL-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT]]
+; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = select i1 [[C0]], i8 [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], i8 [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]]
+; ALL-NEXT: [[OR_COND:%.*]] = select i1 [[C0]], i1 [[C1]], i1 false
+; ALL-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT:%.*]]
; ALL: common.ret:
; ALL-NEXT: ret void
; ALL: final_left:
-; ALL-NEXT: [[FINAL_LEFT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_LEFT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
-; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI]])
+; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI:%.*]])
; ALL-NEXT: br label [[COMMON_RET:%.*]]
; ALL: final_right:
-; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], [[PRED:%.*]] ], [ [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
; ALL-NEXT: call void @sideeffect1(i8 [[FINAL_RIGHT_PHI]])
; ALL-NEXT: br label [[COMMON_RET]]
;
@@ -84,18 +80,17 @@ define void @incompatible_ivs_of_single_phi.falsedest.truedest(i8 %v0, i8 %v1, i
; ALL-LABEL: @incompatible_ivs_of_single_phi.falsedest.truedest(
; ALL-NEXT: pred:
; ALL-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
-; ALL-NEXT: br i1 [[C0]], label [[DISPATCH:%.*]], label [[FINAL_RIGHT:%.*]]
-; ALL: dispatch:
+; ALL-NEXT: [[C0_NOT:%.*]] = xor i1 [[C0]], true
; ALL-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; ALL-NEXT: br i1 [[C1]], label [[FINAL_RIGHT]], label [[FINAL_LEFT:%.*]]
+; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = select i1 [[C0]], i8 [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], i8 [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]]
+; ALL-NEXT: [[OR_COND:%.*]] = select i1 [[C0_NOT]], i1 true, i1 [[C1]]
+; ALL-NEXT: br i1 [[OR_COND]], label [[FINAL_RIGHT:%.*]], label [[FINAL_LEFT:%.*]]
; ALL: common.ret:
; ALL-NEXT: ret void
; ALL: final_left:
-; ALL-NEXT: [[FINAL_LEFT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_LEFT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
-; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI]])
+; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI:%.*]])
; ALL-NEXT: br label [[COMMON_RET:%.*]]
; ALL: final_right:
-; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], [[PRED:%.*]] ], [ [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
; ALL-NEXT: call void @sideeffect1(i8 [[FINAL_RIGHT_PHI]])
; ALL-NEXT: br label [[COMMON_RET]]
;
@@ -156,18 +151,17 @@ define void @incompatible_ivs_of_single_phi.truedest.falsedest(i8 %v0, i8 %v1, i
; ALL-LABEL: @incompatible_ivs_of_single_phi.truedest.falsedest(
; ALL-NEXT: pred:
; ALL-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
-; ALL-NEXT: br i1 [[C0]], label [[FINAL_RIGHT:%.*]], label [[DISPATCH:%.*]]
-; ALL: dispatch:
+; ALL-NEXT: [[C0_NOT:%.*]] = xor i1 [[C0]], true
; ALL-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; ALL-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT]]
+; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = select i1 [[C0]], i8 [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], i8 [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]]
+; ALL-NEXT: [[OR_COND:%.*]] = select i1 [[C0_NOT]], i1 [[C1]], i1 false
+; ALL-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT:%.*]]
; ALL: common.ret:
; ALL-NEXT: ret void
; ALL: final_left:
-; ALL-NEXT: [[FINAL_LEFT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_LEFT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
-; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI]])
+; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI:%.*]])
; ALL-NEXT: br label [[COMMON_RET:%.*]]
; ALL: final_right:
-; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], [[PRED:%.*]] ], [ [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
; ALL-NEXT: call void @sideeffect1(i8 [[FINAL_RIGHT_PHI]])
; ALL-NEXT: br label [[COMMON_RET]]
;
@@ -228,18 +222,16 @@ define void @incompatible_ivs_of_single_phi.truedest.truedest(i8 %v0, i8 %v1, i8
; ALL-LABEL: @incompatible_ivs_of_single_phi.truedest.truedest(
; ALL-NEXT: pred:
; ALL-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
-; ALL-NEXT: br i1 [[C0]], label [[FINAL_RIGHT:%.*]], label [[DISPATCH:%.*]]
-; ALL: dispatch:
; ALL-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; ALL-NEXT: br i1 [[C1]], label [[FINAL_RIGHT]], label [[FINAL_LEFT:%.*]]
+; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = select i1 [[C0]], i8 [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], i8 [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]]
+; ALL-NEXT: [[OR_COND:%.*]] = select i1 [[C0]], i1 true, i1 [[C1]]
+; ALL-NEXT: br i1 [[OR_COND]], label [[FINAL_RIGHT:%.*]], label [[FINAL_LEFT:%.*]]
; ALL: common.ret:
; ALL-NEXT: ret void
; ALL: final_left:
-; ALL-NEXT: [[FINAL_LEFT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_LEFT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
-; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI]])
+; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI:%.*]])
; ALL-NEXT: br label [[COMMON_RET:%.*]]
; ALL: final_right:
-; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], [[PRED:%.*]] ], [ [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
; ALL-NEXT: call void @sideeffect1(i8 [[FINAL_RIGHT_PHI]])
; ALL-NEXT: br label [[COMMON_RET]]
;
@@ -264,18 +256,16 @@ define void @incompatible_ivs_of_single_phi.truedest.truedest.extrause(i8 %v0, i
; ALL-NEXT: pred:
; ALL-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
; ALL-NEXT: call void @use1(i1 [[C0]])
-; ALL-NEXT: br i1 [[C0]], label [[FINAL_RIGHT:%.*]], label [[DISPATCH:%.*]]
-; ALL: dispatch:
; ALL-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; ALL-NEXT: br i1 [[C1]], label [[FINAL_RIGHT]], label [[FINAL_LEFT:%.*]]
+; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = select i1 [[C0]], i8 [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], i8 [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]]
+; ALL-NEXT: [[OR_COND:%.*]] = select i1 [[C0]], i1 true, i1 [[C1]]
+; ALL-NEXT: br i1 [[OR_COND]], label [[FINAL_RIGHT:%.*]], label [[FINAL_LEFT:%.*]]
; ALL: common.ret:
; ALL-NEXT: ret void
; ALL: final_left:
-; ALL-NEXT: [[FINAL_LEFT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_LEFT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
-; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI]])
+; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI:%.*]])
; ALL-NEXT: br label [[COMMON_RET:%.*]]
; ALL: final_right:
-; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]], [[PRED:%.*]] ], [ [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
; ALL-NEXT: call void @sideeffect1(i8 [[FINAL_RIGHT_PHI]])
; ALL-NEXT: br label [[COMMON_RET]]
;
@@ -303,19 +293,17 @@ define void @incompatible_ivs_of_single_phi.insertpos(i8 %v0, i8 %v1, i8 %iv.of.
; ALL-NEXT: pred:
; ALL-NEXT: [[IV_OF_FINAL_RIGHT_FROM_PRED:%.*]] = load i8, ptr [[SRC0:%.*]], align 1
; ALL-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
-; ALL-NEXT: br i1 [[C0]], label [[DISPATCH:%.*]], label [[FINAL_RIGHT:%.*]]
-; ALL: dispatch:
; ALL-NEXT: [[IV_OF_FINAL_RIGHT_FROM_DISPATCH:%.*]] = load i8, ptr [[SRC1:%.*]], align 1
; ALL-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; ALL-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT]]
+; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = select i1 [[C0]], i8 [[IV_OF_FINAL_RIGHT_FROM_DISPATCH]], i8 [[IV_OF_FINAL_RIGHT_FROM_PRED]]
+; ALL-NEXT: [[OR_COND:%.*]] = select i1 [[C0]], i1 [[C1]], i1 false
+; ALL-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT:%.*]]
; ALL: common.ret:
; ALL-NEXT: ret void
; ALL: final_left:
-; ALL-NEXT: [[FINAL_LEFT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_LEFT_FROM_DISPATCH:%.*]], [[DISPATCH]] ]
-; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI]])
+; ALL-NEXT: call void @sideeffect0(i8 [[FINAL_LEFT_PHI:%.*]])
; ALL-NEXT: br label [[COMMON_RET:%.*]]
; ALL: final_right:
-; ALL-NEXT: [[FINAL_RIGHT_PHI:%.*]] = phi i8 [ [[IV_OF_FINAL_RIGHT_FROM_PRED]], [[PRED:%.*]] ], [ [[IV_OF_FINAL_RIGHT_FROM_DISPATCH]], [[DISPATCH]] ]
; ALL-NEXT: call void @sideeffect1(i8 [[FINAL_RIGHT_PHI]])
; ALL-NEXT: br label [[COMMON_RET]]
;
@@ -459,10 +447,10 @@ define float @D139275_c4001580(float %val) {
; ALL-LABEL: @D139275_c4001580(
; ALL-NEXT: entry:
; ALL-NEXT: [[CMP:%.*]] = fcmp ugt float [[VAL:%.*]], 0.000000e+00
-; ALL-NEXT: br i1 [[CMP]], label [[IF_END:%.*]], label [[RETURN:%.*]]
-; ALL: if.end:
; ALL-NEXT: [[CMP1:%.*]] = fcmp ult float [[VAL]], 1.000000e+00
-; ALL-NEXT: br i1 [[CMP1]], label [[IF_END3:%.*]], label [[RETURN]]
+; ALL-NEXT: [[TMP0:%.*]] = select i1 [[CMP]], float 0x3FB99999A0000000, float 0.000000e+00
+; ALL-NEXT: [[OR_COND:%.*]] = and i1 [[CMP]], [[CMP1]]
+; ALL-NEXT: br i1 [[OR_COND]], label [[IF_END3:%.*]], label [[RETURN:%.*]]
; ALL: if.end3:
; ALL-NEXT: [[CMP4:%.*]] = fcmp olt float [[VAL]], 0x3FC99999A0000000
; ALL-NEXT: br i1 [[CMP4]], label [[RETURN]], label [[IF_END6:%.*]]
@@ -470,7 +458,7 @@ define float @D139275_c4001580(float %val) {
; ALL-NEXT: [[SUB:%.*]] = fadd float [[VAL]], 0xBFB99999A0000000
; ALL-NEXT: br label [[RETURN]]
; ALL: return:
-; ALL-NEXT: [[RETVAL_0:%.*]] = phi float [ [[SUB]], [[IF_END6]] ], [ 0.000000e+00, [[ENTRY:%.*]] ], [ 0x3FB99999A0000000, [[IF_END]] ], [ 0.000000e+00, [[IF_END3]] ]
+; ALL-NEXT: [[RETVAL_0:%.*]] = phi float [ [[SUB]], [[IF_END6]] ], [ 0.000000e+00, [[IF_END3]] ], [ [[TMP0]], [[ENTRY:%.*]] ]
; ALL-NEXT: ret float [[RETVAL_0]]
;
entry:
@@ -494,6 +482,51 @@ return:
ret float %retval.0
}
+define i1 @_Z7compareRK1SS1_(ptr %a, ptr %b) {
+; ALL-LABEL: @_Z7compareRK1SS1_(
+; ALL-NEXT: entry:
+; ALL-NEXT: [[TMP0:%.*]] = load i32, ptr [[A:%.*]], align 4
+; ALL-NEXT: [[TMP1:%.*]] = load i32, ptr [[B:%.*]], align 4
+; ALL-NEXT: [[CMP_I:%.*]] = icmp slt i32 [[TMP0]], [[TMP1]]
+; ALL-NEXT: [[CMP_I_NOT:%.*]] = xor i1 [[CMP_I]], true
+; ALL-NEXT: [[CMP_I19:%.*]] = icmp eq i32 [[TMP0]], [[TMP1]]
+; ALL-NEXT: [[TMP2:%.*]] = select i1 [[CMP_I]], i1 true, i1 false
+; ALL-NEXT: [[OR_COND:%.*]] = and i1 [[CMP_I_NOT]], [[CMP_I19]]
+; ALL-NEXT: br i1 [[OR_COND]], label [[LAND_RHS:%.*]], label [[LOR_END:%.*]]
+; ALL: land.rhs:
+; ALL-NEXT: [[Y:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 4
+; ALL-NEXT: [[TMP3:%.*]] = load i32, ptr [[Y]], align 4
+; ALL-NEXT: [[Y14:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 4
+; ALL-NEXT: [[TMP4:%.*]] = load i32, ptr [[Y14]], align 4
+; ALL-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP3]], [[TMP4]]
+; ALL-NEXT: br label [[LOR_END]]
+; ALL: lor.end:
+; ALL-NEXT: [[TMP5:%.*]] = phi i1 [ [[CMP]], [[LAND_RHS]] ], [ [[TMP2]], [[ENTRY:%.*]] ]
+; ALL-NEXT: ret i1 [[TMP5]]
+;
+entry:
+ %0 = load i32, ptr %a, align 4
+ %1 = load i32, ptr %b, align 4
+ %cmp.i = icmp slt i32 %0, %1
+ br i1 %cmp.i, label %lor.end, label %lor.rhs
+
+lor.rhs: ; preds = %entry
+ %cmp.i19 = icmp eq i32 %0, %1
+ br i1 %cmp.i19, label %land.rhs, label %lor.end
+
+land.rhs: ; preds = %lor.rhs
+ %y = getelementptr inbounds nuw i8, ptr %a, i64 4
+ %2 = load i32, ptr %y, align 4
+ %y14 = getelementptr inbounds nuw i8, ptr %b, i64 4
+ %3 = load i32, ptr %y14, align 4
+ %cmp = icmp slt i32 %2, %3
+ br label %lor.end
+
+lor.end: ; preds = %lor.rhs, %land.rhs, %entry
+ %4 = phi i1 [ true, %entry ], [ false, %lor.rhs ], [ %cmp, %land.rhs ]
+ ret i1 %4
+}
+
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; CHEAP: {{.*}}
; COSTLY: {{.*}}
diff --git a/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest.ll b/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest.ll
index 7b88ec338cf5e..dd18d5121c642 100644
--- a/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest.ll
+++ b/llvm/test/Transforms/SimplifyCFG/fold-branch-to-common-dest.ll
@@ -479,21 +479,21 @@ define void @two_preds_with_extra_op_liveout(i8 %v0, i8 %v1, i8 %v2, i8 %v3) {
; CHECK-NEXT: br i1 [[C0]], label [[PRED0:%.*]], label [[PRED1:%.*]]
; CHECK: pred0:
; CHECK-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; CHECK-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[DISPATCH:%.*]]
+; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V1]], [[V2:%.*]]
+; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3_ADJ_OLD]], 0
+; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[C1]], i8 0, i8 [[V3_ADJ_OLD]]
+; CHECK-NEXT: [[OR_COND1:%.*]] = select i1 [[C1]], i1 true, i1 [[C3_OLD]]
+; CHECK-NEXT: br i1 [[OR_COND1]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT:%.*]]
; CHECK: pred1:
-; CHECK-NEXT: [[C2:%.*]] = icmp eq i8 [[V2:%.*]], 0
+; CHECK-NEXT: [[C2:%.*]] = icmp eq i8 [[V2]], 0
; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V1]], [[V2]]
; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V3_ADJ]], 0
; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[C2]], i1 [[C3]], i1 false
-; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT:%.*]]
-; CHECK: dispatch:
-; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V1]], [[V2]]
-; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3_ADJ_OLD]], 0
-; CHECK-NEXT: br i1 [[C3_OLD]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
+; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
; CHECK: common.ret:
; CHECK-NEXT: ret void
; CHECK: final_left:
-; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ_OLD]], [[DISPATCH]] ], [ 0, [[PRED0]] ], [ [[V3_ADJ]], [[PRED1]] ]
+; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ]], [[PRED1]] ], [ [[TMP0]], [[PRED0]] ]
; CHECK-NEXT: call void @use8(i8 [[MERGE_LEFT]])
; CHECK-NEXT: call void @sideeffect0()
; CHECK-NEXT: br label [[COMMON_RET:%.*]]
@@ -530,23 +530,23 @@ define void @two_preds_with_extra_op_liveout_multiuse(i8 %v0, i8 %v1, i8 %v2, i8
; CHECK-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
; CHECK-NEXT: br i1 [[C0]], label [[PRED0:%.*]], label [[PRED1:%.*]]
; CHECK: pred0:
-; CHECK-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; CHECK-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[DISPATCH:%.*]]
+; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3_ADJ_OLD:%.*]], 0
+; CHECK-NEXT: br i1 [[C3_OLD]], label [[FINAL_LEFT:%.*]], label [[DISPATCH:%.*]]
; CHECK: pred1:
; CHECK-NEXT: [[C2:%.*]] = icmp eq i8 [[V2:%.*]], 0
-; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V1]], [[V2]]
+; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V3_ADJ_OLD]], [[V2]]
; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V3_ADJ]], 0
; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[C2]], i1 [[C3]], i1 false
; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT:%.*]]
; CHECK: dispatch:
-; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V1]], [[V2]]
-; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3_ADJ_OLD]], 0
-; CHECK-NEXT: br i1 [[C3_OLD]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
+; CHECK-NEXT: [[V3_ADJ_OLD1:%.*]] = add i8 [[V3_ADJ_OLD]], [[V2]]
+; CHECK-NEXT: [[C3_OLD1:%.*]] = icmp eq i8 [[V3_ADJ_OLD1]], 0
+; CHECK-NEXT: br i1 [[C3_OLD1]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
; CHECK: common.ret:
; CHECK-NEXT: ret void
; CHECK: final_left:
-; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ_OLD]], [[DISPATCH]] ], [ 0, [[PRED0]] ], [ [[V3_ADJ]], [[PRED1]] ]
-; CHECK-NEXT: [[MERGE_LEFT_2:%.*]] = phi i8 [ [[V3_ADJ_OLD]], [[DISPATCH]] ], [ 42, [[PRED0]] ], [ [[V3_ADJ]], [[PRED1]] ]
+; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ_OLD1]], [[DISPATCH]] ], [ 0, [[PRED0]] ], [ [[V3_ADJ]], [[PRED1]] ]
+; CHECK-NEXT: [[MERGE_LEFT_2:%.*]] = phi i8 [ [[V3_ADJ_OLD1]], [[DISPATCH]] ], [ 42, [[PRED0]] ], [ [[V3_ADJ]], [[PRED1]] ]
; CHECK-NEXT: call void @use8(i8 [[MERGE_LEFT]])
; CHECK-NEXT: call void @use8(i8 [[MERGE_LEFT_2]])
; CHECK-NEXT: call void @sideeffect0()
@@ -661,21 +661,21 @@ define void @two_preds_with_extra_op_externally_used_only(i8 %v0, i8 %v1, i8 %v2
; CHECK-NEXT: br i1 [[C0]], label [[PRED0:%.*]], label [[PRED1:%.*]]
; CHECK: pred0:
; CHECK-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; CHECK-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[DISPATCH:%.*]]
+; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V1]], [[V2:%.*]]
+; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3:%.*]], 0
+; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[C1]], i8 0, i8 [[V3_ADJ_OLD]]
+; CHECK-NEXT: [[OR_COND1:%.*]] = select i1 [[C1]], i1 true, i1 [[C3_OLD]]
+; CHECK-NEXT: br i1 [[OR_COND1]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT:%.*]]
; CHECK: pred1:
-; CHECK-NEXT: [[C2:%.*]] = icmp eq i8 [[V2:%.*]], 0
+; CHECK-NEXT: [[C2:%.*]] = icmp eq i8 [[V2]], 0
; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V1]], [[V2]]
-; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V3:%.*]], 0
+; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V3]], 0
; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[C2]], i1 [[C3]], i1 false
-; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT:%.*]]
-; CHECK: dispatch:
-; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V1]], [[V2]]
-; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3]], 0
-; CHECK-NEXT: br i1 [[C3_OLD]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
+; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
; CHECK: common.ret:
; CHECK-NEXT: ret void
; CHECK: final_left:
-; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ_OLD]], [[DISPATCH]] ], [ 0, [[PRED0]] ], [ [[V3_ADJ]], [[PRED1]] ]
+; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ]], [[PRED1]] ], [ [[TMP0]], [[PRED0]] ]
; CHECK-NEXT: call void @use8(i8 [[MERGE_LEFT]])
; CHECK-NEXT: call void @sideeffect0()
; CHECK-NEXT: br label [[COMMON_RET:%.*]]
@@ -712,18 +712,18 @@ define void @two_preds_with_extra_op_externally_used_only_multiuse(i8 %v0, i8 %v
; CHECK-NEXT: [[C0:%.*]] = icmp eq i8 [[V0:%.*]], 0
; CHECK-NEXT: br i1 [[C0]], label [[PRED0:%.*]], label [[PRED1:%.*]]
; CHECK: pred0:
-; CHECK-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; CHECK-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[DISPATCH:%.*]]
+; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3:%.*]], 0
+; CHECK-NEXT: br i1 [[C3_OLD]], label [[FINAL_LEFT:%.*]], label [[DISPATCH:%.*]]
; CHECK: pred1:
; CHECK-NEXT: [[C2:%.*]] = icmp eq i8 [[V2:%.*]], 0
-; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V1]], [[V2]]
-; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V3:%.*]], 0
+; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V3]], [[V2]]
+; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V4:%.*]], 0
; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[C2]], i1 [[C3]], i1 false
; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT:%.*]]
; CHECK: dispatch:
-; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V1]], [[V2]]
-; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3]], 0
-; CHECK-NEXT: br i1 [[C3_OLD]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
+; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V3]], [[V2]]
+; CHECK-NEXT: [[C3_OLD1:%.*]] = icmp eq i8 [[V4]], 0
+; CHECK-NEXT: br i1 [[C3_OLD1]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
; CHECK: common.ret:
; CHECK-NEXT: ret void
; CHECK: final_left:
@@ -820,21 +820,21 @@ define void @two_preds_with_extra_op_externally_used_only_after_cond(i8 %v0, i8
; CHECK-NEXT: br i1 [[C0]], label [[PRED0:%.*]], label [[PRED1:%.*]]
; CHECK: pred0:
; CHECK-NEXT: [[C1:%.*]] = icmp eq i8 [[V1:%.*]], 0
-; CHECK-NEXT: br i1 [[C1]], label [[FINAL_LEFT:%.*]], label [[DISPATCH:%.*]]
+; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3:%.*]], 0
+; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V4:%.*]], [[V5:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[C1]], i8 0, i8 [[V3_ADJ_OLD]]
+; CHECK-NEXT: [[OR_COND1:%.*]] = select i1 [[C1]], i1 true, i1 [[C3_OLD]]
+; CHECK-NEXT: br i1 [[OR_COND1]], label [[FINAL_LEFT:%.*]], label [[FINAL_RIGHT:%.*]]
; CHECK: pred1:
; CHECK-NEXT: [[C2:%.*]] = icmp eq i8 [[V2:%.*]], 0
-; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V3:%.*]], 0
-; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V4:%.*]], [[V5:%.*]]
+; CHECK-NEXT: [[C3:%.*]] = icmp eq i8 [[V3]], 0
+; CHECK-NEXT: [[V3_ADJ:%.*]] = add i8 [[V4]], [[V5]]
; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[C2]], i1 [[C3]], i1 false
-; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT:%.*]]
-; CHECK: dispatch:
-; CHECK-NEXT: [[C3_OLD:%.*]] = icmp eq i8 [[V3]], 0
-; CHECK-NEXT: [[V3_ADJ_OLD:%.*]] = add i8 [[V4]], [[V5]]
-; CHECK-NEXT: br i1 [[C3_OLD]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
+; CHECK-NEXT: br i1 [[OR_COND]], label [[FINAL_LEFT]], label [[FINAL_RIGHT]]
; CHECK: common.ret:
; CHECK-NEXT: ret void
; CHECK: final_left:
-; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ_OLD]], [[DISPATCH]] ], [ 0, [[PRED0]] ], [ [[V3_ADJ]], [[PRED1]] ]
+; CHECK-NEXT: [[MERGE_LEFT:%.*]] = phi i8 [ [[V3_ADJ]], [[PRED1]] ], [ [[TMP0]], [[PRED0]] ]
; CHECK-NEXT: call void @use8(i8 [[MERGE_LEFT]])
; CHECK-NEXT: call void @sideeffect0()
; CHECK-NEXT: br label [[COMMON_RET:%.*]]
@@ -1149,13 +1149,13 @@ define i32 @test_builtin_fpclassify(float %x) {
; CHECK-LABEL: @test_builtin_fpclassify(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[ISZERO:%.*]] = fcmp oeq float [[X:%.*]], 0.000000e+00
-; CHECK-NEXT: br i1 [[ISZERO]], label [[FPCLASSIFY_END:%.*]], label [[FPCLASSIFY_NOT_ZERO:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = fcmp uno float [[X]], 0.000000e+00
+; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[ISZERO]], i32 2, i32 0
+; CHECK-NEXT: [[OR_COND:%.*]] = or i1 [[ISZERO]], [[CMP]]
+; CHECK-NEXT: br i1 [[OR_COND]], label [[FPCLASSIFY_END:%.*]], label [[FPCLASSIFY_NOT_NAN:%.*]]
; CHECK: fpclassify_end:
-; CHECK-NEXT: [[FPCLASSIFY_RESULT:%.*]] = phi i32 [ 2, [[ENTRY:%.*]] ], [ 0, [[FPCLASSIFY_NOT_ZERO]] ], [ 1, [[FPCLASSIFY_NOT_NAN:%.*]] ], [ [[NORMAL_OR_SUBNORMAL:%.*]], [[FPCLASSIFY_NOT_INF:%.*]] ]
+; CHECK-NEXT: [[FPCLASSIFY_RESULT:%.*]] = phi i32 [ 1, [[FPCLASSIFY_NOT_NAN]] ], [ [[NORMAL_OR_SUBNORMAL:%.*]], [[FPCLASSIFY_NOT_INF:%.*]] ], [ [[TMP0]], [[ENTRY:%.*]] ]
; CHECK-NEXT: ret i32 [[FPCLASSIFY_RESULT]]
-; CHECK: fpclassify_not_zero:
-; CHECK-NEXT: [[CMP:%.*]] = fcmp uno float [[X]], 0.000000e+00
-; CHECK-NEXT: br i1 [[CMP]], label [[FPCLASSIFY_END]], label [[FPCLASSIFY_NOT_NAN]]
; CHECK: fpclassify_not_nan:
; CHECK-NEXT: [[X_ABS:%.*]] = tail call float @llvm.fabs.f32(float [[X]])
; CHECK-NEXT: [[ISINF:%.*]] = fcmp oeq float [[X_ABS]], 0x7FF0000000000000
diff --git a/llvm/test/Transforms/SimplifyCFG/hoist-dbgvalue.ll b/llvm/test/Transforms/SimplifyCFG/hoist-dbgvalue.ll
index 5a69a0c002fb8..65a5f5e8e138b 100644
--- a/llvm/test/Transforms/SimplifyCFG/hoist-dbgvalue.ll
+++ b/llvm/test/Transforms/SimplifyCFG/hoist-dbgvalue.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S < %s | FileCheck %s
+; RUN: opt -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -simplifycfg-branch-fold-threshold=0 -S < %s | FileCheck %s
define i32 @foo(i32 %i) nounwind ssp !dbg !0 {
; CHECK-LABEL: @foo(
diff --git a/llvm/test/Transforms/SimplifyCFG/speculate-derefable-load.ll b/llvm/test/Transforms/SimplifyCFG/speculate-derefable-load.ll
index b8c999d700aa7..1b7effa6ef1c1 100644
--- a/llvm/test/Transforms/SimplifyCFG/speculate-derefable-load.ll
+++ b/llvm/test/Transforms/SimplifyCFG/speculate-derefable-load.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
-; RUN: opt -S -passes=simplifycfg < %s | FileCheck %s
+; RUN: opt -S -passes=simplifycfg -simplifycfg-branch-fold-threshold=0 < %s | FileCheck %s
define i64 @align_deref_align(i1 %c, ptr %p) {
; CHECK-LABEL: define i64 @align_deref_align(
diff --git a/llvm/test/Transforms/SimplifyCFG/switch-on-const-select.ll b/llvm/test/Transforms/SimplifyCFG/switch-on-const-select.ll
index e8b58639c13dd..38bb84def8e00 100644
--- a/llvm/test/Transforms/SimplifyCFG/switch-on-const-select.ll
+++ b/llvm/test/Transforms/SimplifyCFG/switch-on-const-select.ll
@@ -1,5 +1,5 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S | FileCheck %s
+; RUN: opt < %s -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -simplifycfg-branch-fold-threshold=0 -S | FileCheck %s
; Test basic folding to a conditional branch.
define i32 @foo(i64 %x, i64 %y) nounwind {
>From e365fd291ac0a5bbf49b8f33694608f4e5fba3fe Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Mon, 16 Jun 2025 14:56:11 -0700
Subject: [PATCH 2/8] Formatting
---
llvm/lib/Transforms/Utils/SimplifyCFG.cpp | 36 ++++++++++++-----------
1 file changed, 19 insertions(+), 17 deletions(-)
diff --git a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
index 7304bdaab7f88..a08a1f1cd247c 100644
--- a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -3909,11 +3909,10 @@ shouldFoldCondBranchesToCommonDestination(BranchInst *BI, BranchInst *PBI,
return std::nullopt;
}
-static bool performBranchToCommonDestFolding(BranchInst *BI, BranchInst *PBI,
- DomTreeUpdater *DTU,
- MemorySSAUpdater *MSSAU,
- const TargetTransformInfo *TTI,
- SmallDenseMap<PHINode*, SelectInst*, 8> &InsertNewPHIs) {
+static bool performBranchToCommonDestFolding(
+ BranchInst *BI, BranchInst *PBI, DomTreeUpdater *DTU,
+ MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI,
+ SmallDenseMap<PHINode *, SelectInst *, 8> &InsertNewPHIs) {
BasicBlock *BB = BI->getParent();
BasicBlock *PredBlock = PBI->getParent();
@@ -4008,7 +4007,7 @@ static bool performBranchToCommonDestFolding(BranchInst *BI, BranchInst *PBI,
Instruction *SI = It->second;
// Oprands might have been promoted to bonous inst
RemapInstruction(SI, VMap,
- RF_NoModuleLevelChanges | RF_IgnoreMissingLocals);
+ RF_NoModuleLevelChanges | RF_IgnoreMissingLocals);
// Insert SelectInst as the new PHINode incoming value
SI->insertBefore(PredBlock->getTerminator()->getIterator());
// Fix PHINode
@@ -4081,11 +4080,11 @@ bool llvm::foldBranchToCommonDest(BranchInst *BI, DomTreeUpdater *DTU,
SmallVector<BasicBlock *, 8> Preds;
struct InsertPointTy {
InstructionCost Cost;
- Value *TValue; // True Value
- Value *FValue; // False Value
- PHINode *Phi;
+ Value *TValue; // True Value
+ Value *FValue; // False Value
+ PHINode *Phi;
};
- SmallDenseMap<BranchInst*, SmallVector<InsertPointTy, 8>, 8> InsertPts;
+ SmallDenseMap<BranchInst *, SmallVector<InsertPointTy, 8>, 8> InsertPts;
for (BasicBlock *PredBlock : predecessors(BB)) {
BranchInst *PBI = dyn_cast<BranchInst>(PredBlock->getTerminator());
@@ -4093,8 +4092,9 @@ bool llvm::foldBranchToCommonDest(BranchInst *BI, DomTreeUpdater *DTU,
if (!PBI || PBI->isUnconditional())
continue;
- // If there is a PHI node in the common successor, verify that the same value flows in from both
- // blocks. Otherwise, check whether we can create a SelectInst to combine the incoming values
+ // If there is a PHI node in the common successor, verify that the same
+ // value flows in from both blocks. Otherwise, check whether we can create a
+ // SelectInst to combine the incoming values
if (!safeToMergeTerminators(BI, PBI)) {
if (BI == PBI)
continue;
@@ -4105,9 +4105,10 @@ bool llvm::foldBranchToCommonDest(BranchInst *BI, DomTreeUpdater *DTU,
Value *IV1 = Phi.getIncomingValueForBlock(PredBlock);
InstructionCost PCost;
if (TTI) {
- PCost = TTI->getCmpSelInstrCost(Instruction::Select, Phi.getType(),
- CmpInst::makeCmpResultType(Phi.getType()),
- CmpInst::BAD_ICMP_PREDICATE, CostKind);
+ PCost = TTI->getCmpSelInstrCost(
+ Instruction::Select, Phi.getType(),
+ CmpInst::makeCmpResultType(Phi.getType()),
+ CmpInst::BAD_ICMP_PREDICATE, CostKind);
}
auto &IP = InsertPts[PBI];
if (PBI->getSuccessor(0) == BB)
@@ -4210,8 +4211,9 @@ bool llvm::foldBranchToCommonDest(BranchInst *BI, DomTreeUpdater *DTU,
SmallDenseMap<PHINode *, SelectInst *, 8> newPhis;
if (InsertPts.contains(PBI)) {
Value *PC = PBI->getCondition();
- for (auto const InsertInfo: InsertPts[PBI]) {
- SelectInst *newPhi = SelectInst::Create(PC, InsertInfo.TValue, InsertInfo.FValue);
+ for (auto const InsertInfo : InsertPts[PBI]) {
+ SelectInst *newPhi =
+ SelectInst::Create(PC, InsertInfo.TValue, InsertInfo.FValue);
newPhis.insert(std::make_pair(InsertInfo.Phi, newPhi));
}
}
>From f622d10bf58a6f28fd9d47f9d46a293d63d31888 Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Mon, 16 Jun 2025 16:12:22 -0700
Subject: [PATCH 3/8] Fix clang hip math tests
---
clang/test/Headers/__clang_hip_math.hip | 2000 ++++++++++++-----------
1 file changed, 1064 insertions(+), 936 deletions(-)
diff --git a/clang/test/Headers/__clang_hip_math.hip b/clang/test/Headers/__clang_hip_math.hip
index 11c9cd301abb7..684ceb0d395b3 100644
--- a/clang/test/Headers/__clang_hip_math.hip
+++ b/clang/test/Headers/__clang_hip_math.hip
@@ -166,32 +166,43 @@ extern "C" __device__ uint64_t test___make_mantissa_base10(const char *p) {
// CHECK-NEXT: [[__TAGP_ADDR_0_I:%.*]] = phi ptr [ [[P:%.*]], [[ENTRY:%.*]] ], [ [[__TAGP_ADDR_1_I:%.*]], [[CLEANUP_I:%.*]] ]
// CHECK-NEXT: [[__R_0_I:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[__R_2_I:%.*]], [[CLEANUP_I]] ]
// CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I]], align 1, !tbaa [[TBAA4]]
-// CHECK-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP0]], 0
+// CHECK-NEXT: [[DOTFR2:%.*]] = freeze i8 [[TMP0]]
+// CHECK-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[DOTFR2]], 0
// CHECK-NEXT: br i1 [[CMP_NOT_I]], label [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT:%.*]], label [[WHILE_BODY_I:%.*]]
// CHECK: while.body.i:
-// CHECK-NEXT: [[TMP1:%.*]] = add i8 [[TMP0]], -48
+// CHECK-NEXT: [[TMP1:%.*]] = add i8 [[DOTFR2]], -48
// CHECK-NEXT: [[OR_COND_I:%.*]] = icmp ult i8 [[TMP1]], 10
-// CHECK-NEXT: br i1 [[OR_COND_I]], label [[IF_END31_I:%.*]], label [[IF_ELSE_I:%.*]]
-// CHECK: if.else.i:
-// CHECK-NEXT: [[TMP2:%.*]] = add i8 [[TMP0]], -97
+// CHECK-NEXT: [[TMP2:%.*]] = add i8 [[DOTFR2]], -97
// CHECK-NEXT: [[OR_COND33_I:%.*]] = icmp ult i8 [[TMP2]], 6
-// CHECK-NEXT: br i1 [[OR_COND33_I]], label [[IF_END31_I]], label [[IF_ELSE17_I:%.*]]
-// CHECK: if.else17.i:
-// CHECK-NEXT: [[TMP3:%.*]] = add i8 [[TMP0]], -65
-// CHECK-NEXT: [[OR_COND34_I:%.*]] = icmp ult i8 [[TMP3]], 6
-// CHECK-NEXT: br i1 [[OR_COND34_I]], label [[IF_END31_I]], label [[CLEANUP_I]]
+// CHECK-NEXT: [[TMP3:%.*]] = select i1 [[OR_COND33_I]], i64 -87, i64 -55
+// CHECK-NEXT: [[TMP4:%.*]] = select i1 [[OR_COND_I]], i64 -48, i64 [[TMP3]]
+// CHECK-NEXT: br i1 [[OR_COND_I]], label [[IF_END31_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// CHECK: switch.early.test:
+// CHECK-NEXT: switch i8 [[DOTFR2]], label [[CLEANUP_I]] [
+// CHECK-NEXT: i8 102, label [[IF_END31_I]]
+// CHECK-NEXT: i8 101, label [[IF_END31_I]]
+// CHECK-NEXT: i8 100, label [[IF_END31_I]]
+// CHECK-NEXT: i8 99, label [[IF_END31_I]]
+// CHECK-NEXT: i8 98, label [[IF_END31_I]]
+// CHECK-NEXT: i8 97, label [[IF_END31_I]]
+// CHECK-NEXT: i8 70, label [[IF_END31_I]]
+// CHECK-NEXT: i8 69, label [[IF_END31_I]]
+// CHECK-NEXT: i8 68, label [[IF_END31_I]]
+// CHECK-NEXT: i8 67, label [[IF_END31_I]]
+// CHECK-NEXT: i8 66, label [[IF_END31_I]]
+// CHECK-NEXT: i8 65, label [[IF_END31_I]]
+// CHECK-NEXT: ]
// CHECK: if.end31.i:
-// CHECK-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I]] ], [ -87, [[IF_ELSE_I]] ], [ -55, [[IF_ELSE17_I]] ]
// CHECK-NEXT: [[MUL24_I:%.*]] = shl i64 [[__R_0_I]], 4
-// CHECK-NEXT: [[CONV25_I:%.*]] = zext nneg i8 [[TMP0]] to i64
-// CHECK-NEXT: [[ADD26_I:%.*]] = add i64 [[MUL24_I]], [[DOTSINK]]
+// CHECK-NEXT: [[CONV25_I:%.*]] = zext nneg i8 [[DOTFR2]] to i64
+// CHECK-NEXT: [[ADD26_I:%.*]] = add i64 [[MUL24_I]], [[TMP4]]
// CHECK-NEXT: [[ADD28_I:%.*]] = add i64 [[ADD26_I]], [[CONV25_I]]
// CHECK-NEXT: [[INCDEC_PTR_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I]], i64 1
// CHECK-NEXT: br label [[CLEANUP_I]]
// CHECK: cleanup.i:
-// CHECK-NEXT: [[__TAGP_ADDR_1_I]] = phi ptr [ [[INCDEC_PTR_I]], [[IF_END31_I]] ], [ [[__TAGP_ADDR_0_I]], [[IF_ELSE17_I]] ]
-// CHECK-NEXT: [[__R_2_I]] = phi i64 [ [[ADD28_I]], [[IF_END31_I]] ], [ [[__R_0_I]], [[IF_ELSE17_I]] ]
-// CHECK-NEXT: [[COND_I:%.*]] = phi i1 [ true, [[IF_END31_I]] ], [ false, [[IF_ELSE17_I]] ]
+// CHECK-NEXT: [[__TAGP_ADDR_1_I]] = phi ptr [ [[INCDEC_PTR_I]], [[IF_END31_I]] ], [ [[__TAGP_ADDR_0_I]], [[SWITCH_EARLY_TEST]] ]
+// CHECK-NEXT: [[__R_2_I]] = phi i64 [ [[ADD28_I]], [[IF_END31_I]] ], [ [[__R_0_I]], [[SWITCH_EARLY_TEST]] ]
+// CHECK-NEXT: [[COND_I:%.*]] = phi i1 [ true, [[IF_END31_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// CHECK-NEXT: br i1 [[COND_I]], label [[WHILE_COND_I]], label [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK: _ZL22__make_mantissa_base16PKc.exit:
// CHECK-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, [[CLEANUP_I]] ], [ [[__R_0_I]], [[WHILE_COND_I]] ]
@@ -201,38 +212,48 @@ extern "C" __device__ uint64_t test___make_mantissa_base10(const char *p) {
// AMDGCNSPIRV-NEXT: entry:
// AMDGCNSPIRV-NEXT: br label [[WHILE_COND_I:%.*]]
// AMDGCNSPIRV: while.cond.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I:%.*]] = phi ptr addrspace(4) [ [[P:%.*]], [[ENTRY:%.*]] ], [ [[__TAGP_ADDR_1_I:%.*]], [[CLEANUP_I:%.*]] ]
-// AMDGCNSPIRV-NEXT: [[__R_0_I:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[__R_2_I:%.*]], [[CLEANUP_I]] ]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I:%.*]] = phi ptr addrspace(4) [ [[P:%.*]], [[ENTRY:%.*]] ], [ [[__TAGP_ADDR_1_I:%.*]], [[SWITCH_EARLY_TEST:%.*]] ], [ [[__TAGP_ADDR_1_I]], [[WHILE_BODY_I:%.*]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I]], [[SWITCH_EARLY_TEST]] ]
+// AMDGCNSPIRV-NEXT: [[__R_0_I:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[__R_2_I:%.*]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[WHILE_BODY_I]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I]], [[SWITCH_EARLY_TEST]] ]
// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], align 1, !tbaa [[TBAA5]]
// AMDGCNSPIRV-NEXT: [[CMP_NOT_I:%.*]] = icmp eq i8 [[TMP0]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I]], label [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT:%.*]], label [[WHILE_BODY_I:%.*]]
+// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I]], label [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT:%.*]], label [[WHILE_BODY_I]]
// AMDGCNSPIRV: while.body.i:
// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = add i8 [[TMP0]], -48
// AMDGCNSPIRV-NEXT: [[OR_COND_I:%.*]] = icmp ult i8 [[TMP1]], 10
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I]], label [[IF_END31_I:%.*]], label [[IF_ELSE_I:%.*]]
-// AMDGCNSPIRV: if.else.i:
// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = add i8 [[TMP0]], -97
// AMDGCNSPIRV-NEXT: [[OR_COND33_I:%.*]] = icmp ult i8 [[TMP2]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND33_I]], label [[IF_END31_I]], label [[IF_ELSE17_I:%.*]]
-// AMDGCNSPIRV: if.else17.i:
// AMDGCNSPIRV-NEXT: [[TMP3:%.*]] = add i8 [[TMP0]], -65
// AMDGCNSPIRV-NEXT: [[OR_COND34_I:%.*]] = icmp ult i8 [[TMP3]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND34_I]], label [[IF_END31_I]], label [[CLEANUP_I]]
-// AMDGCNSPIRV: if.end31.i:
-// AMDGCNSPIRV-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I]] ], [ -87, [[IF_ELSE_I]] ], [ -55, [[IF_ELSE17_I]] ]
+// AMDGCNSPIRV-NEXT: [[TMP4:%.*]] = select i1 [[OR_COND33_I]], i64 -87, i64 -55
+// AMDGCNSPIRV-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND_I]], i64 -48, i64 [[TMP4]]
+// AMDGCNSPIRV-NEXT: [[TMP6:%.*]] = or i1 [[OR_COND33_I]], [[OR_COND34_I]]
+// AMDGCNSPIRV-NEXT: [[OR_COND1:%.*]] = or i1 [[TMP6]], [[OR_COND_I]]
// AMDGCNSPIRV-NEXT: [[MUL24_I:%.*]] = shl i64 [[__R_0_I]], 4
// AMDGCNSPIRV-NEXT: [[CONV25_I:%.*]] = zext nneg i8 [[TMP0]] to i64
-// AMDGCNSPIRV-NEXT: [[ADD26_I:%.*]] = add i64 [[MUL24_I]], [[DOTSINK]]
+// AMDGCNSPIRV-NEXT: [[ADD26_I:%.*]] = add i64 [[MUL24_I]], [[TMP5]]
// AMDGCNSPIRV-NEXT: [[ADD28_I:%.*]] = add i64 [[ADD26_I]], [[CONV25_I]]
-// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], i64 1
-// AMDGCNSPIRV-NEXT: br label [[CLEANUP_I]]
-// AMDGCNSPIRV: cleanup.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I]], [[IF_END31_I]] ], [ [[__TAGP_ADDR_0_I]], [[IF_ELSE17_I]] ]
-// AMDGCNSPIRV-NEXT: [[__R_2_I]] = phi i64 [ [[ADD28_I]], [[IF_END31_I]] ], [ [[__R_0_I]], [[IF_ELSE17_I]] ]
-// AMDGCNSPIRV-NEXT: [[COND_I:%.*]] = phi i1 [ true, [[IF_END31_I]] ], [ false, [[IF_ELSE17_I]] ]
-// AMDGCNSPIRV-NEXT: br i1 [[COND_I]], label [[WHILE_COND_I]], label [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]], !llvm.loop [[LOOP12:![0-9]+]]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_IDX:%.*]] = zext i1 [[OR_COND1]] to i64
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I]], i64 [[__TAGP_ADDR_1_I_IDX]]
+// AMDGCNSPIRV-NEXT: [[__R_2_I]] = select i1 [[OR_COND1]], i64 [[ADD28_I]], i64 [[__R_0_I]]
+// AMDGCNSPIRV-NEXT: [[TMP7:%.*]] = freeze i1 [[OR_COND_I]]
+// AMDGCNSPIRV-NEXT: br i1 [[TMP7]], label [[WHILE_COND_I]], label [[SWITCH_EARLY_TEST]]
+// AMDGCNSPIRV: switch.early.test:
+// AMDGCNSPIRV-NEXT: switch i8 [[TMP0]], label [[_ZL22__MAKE_MANTISSA_BASE16PKC_EXIT]] [
+// AMDGCNSPIRV-NEXT: i8 102, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 101, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 100, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 99, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 98, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 97, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 70, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 69, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 68, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 67, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 66, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: i8 65, label [[WHILE_COND_I]]
+// AMDGCNSPIRV-NEXT: ]
// AMDGCNSPIRV: _ZL22__make_mantissa_base16PKc.exit:
-// AMDGCNSPIRV-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, [[CLEANUP_I]] ], [ [[__R_0_I]], [[WHILE_COND_I]] ]
+// AMDGCNSPIRV-NEXT: [[RETVAL_2_I:%.*]] = phi i64 [ 0, [[SWITCH_EARLY_TEST]] ], [ [[__R_0_I]], [[WHILE_COND_I]] ]
// AMDGCNSPIRV-NEXT: ret i64 [[RETVAL_2_I]]
//
extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) {
@@ -257,46 +278,57 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) {
// CHECK-NEXT: [[__TAGP_ADDR_0_I31_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I37_I:%.*]], [[CLEANUP_I36_I:%.*]] ], [ [[INCDEC_PTR_I]], [[WHILE_COND_I30_I_PREHEADER]] ]
// CHECK-NEXT: [[__R_0_I32_I:%.*]] = phi i64 [ [[__R_2_I_I:%.*]], [[CLEANUP_I36_I]] ], [ 0, [[WHILE_COND_I30_I_PREHEADER]] ]
// CHECK-NEXT: [[TMP2:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I31_I]], align 1, !tbaa [[TBAA4]]
-// CHECK-NEXT: [[CMP_NOT_I33_I:%.*]] = icmp eq i8 [[TMP2]], 0
+// CHECK-NEXT: [[DOTFR14:%.*]] = freeze i8 [[TMP2]]
+// CHECK-NEXT: [[CMP_NOT_I33_I:%.*]] = icmp eq i8 [[DOTFR14]], 0
// CHECK-NEXT: br i1 [[CMP_NOT_I33_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT:%.*]], label [[WHILE_BODY_I34_I:%.*]]
// CHECK: while.body.i34.i:
-// CHECK-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
+// CHECK-NEXT: [[TMP3:%.*]] = add i8 [[DOTFR14]], -48
// CHECK-NEXT: [[OR_COND_I35_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// CHECK-NEXT: br i1 [[OR_COND_I35_I]], label [[IF_END31_I_I:%.*]], label [[IF_ELSE_I_I:%.*]]
-// CHECK: if.else.i.i:
-// CHECK-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
+// CHECK-NEXT: [[TMP4:%.*]] = add i8 [[DOTFR14]], -97
// CHECK-NEXT: [[OR_COND33_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// CHECK-NEXT: br i1 [[OR_COND33_I_I]], label [[IF_END31_I_I]], label [[IF_ELSE17_I_I:%.*]]
-// CHECK: if.else17.i.i:
-// CHECK-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
-// CHECK-NEXT: [[OR_COND34_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// CHECK-NEXT: br i1 [[OR_COND34_I_I]], label [[IF_END31_I_I]], label [[CLEANUP_I36_I]]
+// CHECK-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND33_I_I]], i64 -87, i64 -55
+// CHECK-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND_I35_I]], i64 -48, i64 [[TMP5]]
+// CHECK-NEXT: br i1 [[OR_COND_I35_I]], label [[IF_END31_I_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// CHECK: switch.early.test:
+// CHECK-NEXT: switch i8 [[DOTFR14]], label [[CLEANUP_I36_I]] [
+// CHECK-NEXT: i8 102, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 101, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 100, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 99, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 98, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 97, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 70, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 69, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 68, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 67, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 66, label [[IF_END31_I_I]]
+// CHECK-NEXT: i8 65, label [[IF_END31_I_I]]
+// CHECK-NEXT: ]
// CHECK: if.end31.i.i:
-// CHECK-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I34_I]] ], [ -87, [[IF_ELSE_I_I]] ], [ -55, [[IF_ELSE17_I_I]] ]
// CHECK-NEXT: [[MUL24_I_I:%.*]] = shl i64 [[__R_0_I32_I]], 4
-// CHECK-NEXT: [[CONV25_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// CHECK-NEXT: [[ADD26_I_I:%.*]] = add i64 [[MUL24_I_I]], [[DOTSINK]]
+// CHECK-NEXT: [[CONV25_I_I:%.*]] = zext nneg i8 [[DOTFR14]] to i64
+// CHECK-NEXT: [[ADD26_I_I:%.*]] = add i64 [[MUL24_I_I]], [[TMP6]]
// CHECK-NEXT: [[ADD28_I_I:%.*]] = add i64 [[ADD26_I_I]], [[CONV25_I_I]]
// CHECK-NEXT: [[INCDEC_PTR_I40_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I31_I]], i64 1
// CHECK-NEXT: br label [[CLEANUP_I36_I]]
// CHECK: cleanup.i36.i:
-// CHECK-NEXT: [[__TAGP_ADDR_1_I37_I]] = phi ptr [ [[INCDEC_PTR_I40_I]], [[IF_END31_I_I]] ], [ [[__TAGP_ADDR_0_I31_I]], [[IF_ELSE17_I_I]] ]
-// CHECK-NEXT: [[__R_2_I_I]] = phi i64 [ [[ADD28_I_I]], [[IF_END31_I_I]] ], [ [[__R_0_I32_I]], [[IF_ELSE17_I_I]] ]
-// CHECK-NEXT: [[COND_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I]] ], [ false, [[IF_ELSE17_I_I]] ]
+// CHECK-NEXT: [[__TAGP_ADDR_1_I37_I]] = phi ptr [ [[INCDEC_PTR_I40_I]], [[IF_END31_I_I]] ], [ [[__TAGP_ADDR_0_I31_I]], [[SWITCH_EARLY_TEST]] ]
+// CHECK-NEXT: [[__R_2_I_I]] = phi i64 [ [[ADD28_I_I]], [[IF_END31_I_I]] ], [ [[__R_0_I32_I]], [[SWITCH_EARLY_TEST]] ]
+// CHECK-NEXT: [[COND_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// CHECK-NEXT: br i1 [[COND_I_I]], label [[WHILE_COND_I30_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]], !llvm.loop [[LOOP11]]
// CHECK: while.cond.i.i:
// CHECK-NEXT: [[__TAGP_ADDR_0_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I_I:%.*]], [[CLEANUP_I_I:%.*]] ], [ [[INCDEC_PTR_I]], [[IF_THEN_I]] ]
// CHECK-NEXT: [[__R_0_I_I:%.*]] = phi i64 [ [[__R_1_I_I:%.*]], [[CLEANUP_I_I]] ], [ 0, [[IF_THEN_I]] ]
-// CHECK-NEXT: [[TMP6:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I]], align 1, !tbaa [[TBAA4]]
-// CHECK-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// CHECK-NEXT: [[TMP7:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I]], align 1, !tbaa [[TBAA4]]
+// CHECK-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq i8 [[TMP7]], 0
// CHECK-NEXT: br i1 [[CMP_NOT_I_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]], label [[WHILE_BODY_I_I:%.*]]
// CHECK: while.body.i.i:
-// CHECK-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// CHECK-NEXT: [[OR_COND_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// CHECK-NEXT: [[TMP8:%.*]] = and i8 [[TMP7]], -8
+// CHECK-NEXT: [[OR_COND_I_I:%.*]] = icmp eq i8 [[TMP8]], 48
// CHECK-NEXT: br i1 [[OR_COND_I_I]], label [[IF_THEN_I_I:%.*]], label [[CLEANUP_I_I]]
// CHECK: if.then.i.i:
// CHECK-NEXT: [[MUL_I_I:%.*]] = shl i64 [[__R_0_I_I]], 3
-// CHECK-NEXT: [[CONV5_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// CHECK-NEXT: [[CONV5_I_I:%.*]] = zext nneg i8 [[TMP7]] to i64
// CHECK-NEXT: [[ADD_I_I:%.*]] = add i64 [[MUL_I_I]], -48
// CHECK-NEXT: [[SUB_I_I:%.*]] = add i64 [[ADD_I_I]], [[CONV5_I_I]]
// CHECK-NEXT: [[INCDEC_PTR_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I]], i64 1
@@ -308,16 +340,16 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) {
// CHECK: while.cond.i14.i:
// CHECK-NEXT: [[__TAGP_ADDR_0_I15_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I21_I:%.*]], [[CLEANUP_I20_I:%.*]] ], [ [[P]], [[ENTRY:%.*]] ]
// CHECK-NEXT: [[__R_0_I16_I:%.*]] = phi i64 [ [[__R_1_I22_I:%.*]], [[CLEANUP_I20_I]] ], [ 0, [[ENTRY]] ]
-// CHECK-NEXT: [[TMP8:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I]], align 1, !tbaa [[TBAA4]]
-// CHECK-NEXT: [[CMP_NOT_I17_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// CHECK-NEXT: [[TMP9:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I]], align 1, !tbaa [[TBAA4]]
+// CHECK-NEXT: [[CMP_NOT_I17_I:%.*]] = icmp eq i8 [[TMP9]], 0
// CHECK-NEXT: br i1 [[CMP_NOT_I17_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]], label [[WHILE_BODY_I18_I:%.*]]
// CHECK: while.body.i18.i:
-// CHECK-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// CHECK-NEXT: [[OR_COND_I19_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// CHECK-NEXT: [[TMP10:%.*]] = add i8 [[TMP9]], -48
+// CHECK-NEXT: [[OR_COND_I19_I:%.*]] = icmp ult i8 [[TMP10]], 10
// CHECK-NEXT: br i1 [[OR_COND_I19_I]], label [[IF_THEN_I24_I:%.*]], label [[CLEANUP_I20_I]]
// CHECK: if.then.i24.i:
// CHECK-NEXT: [[MUL_I25_I:%.*]] = mul i64 [[__R_0_I16_I]], 10
-// CHECK-NEXT: [[CONV5_I26_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// CHECK-NEXT: [[CONV5_I26_I:%.*]] = zext nneg i8 [[TMP9]] to i64
// CHECK-NEXT: [[ADD_I27_I:%.*]] = add i64 [[MUL_I25_I]], -48
// CHECK-NEXT: [[SUB_I28_I:%.*]] = add i64 [[ADD_I27_I]], [[CONV5_I26_I]]
// CHECK-NEXT: [[INCDEC_PTR_I29_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I]], i64 1
@@ -345,47 +377,57 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) {
// AMDGCNSPIRV: while.cond.i28.i.preheader:
// AMDGCNSPIRV-NEXT: br label [[WHILE_COND_I28_I:%.*]]
// AMDGCNSPIRV: while.cond.i28.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I29_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I34_I:%.*]], [[CLEANUP_I_I:%.*]] ], [ [[INCDEC_PTR_I]], [[WHILE_COND_I28_I_PREHEADER]] ]
-// AMDGCNSPIRV-NEXT: [[__R_0_I30_I:%.*]] = phi i64 [ [[__R_2_I_I:%.*]], [[CLEANUP_I_I]] ], [ 0, [[WHILE_COND_I28_I_PREHEADER]] ]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I29_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I34_I:%.*]], [[SWITCH_EARLY_TEST:%.*]] ], [ [[INCDEC_PTR_I]], [[WHILE_COND_I28_I_PREHEADER]] ], [ [[__TAGP_ADDR_1_I34_I]], [[WHILE_BODY_I32_I:%.*]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I]], [[SWITCH_EARLY_TEST]] ]
+// AMDGCNSPIRV-NEXT: [[__R_0_I30_I:%.*]] = phi i64 [ [[__R_2_I_I:%.*]], [[SWITCH_EARLY_TEST]] ], [ 0, [[WHILE_COND_I28_I_PREHEADER]] ], [ [[__R_2_I_I]], [[WHILE_BODY_I32_I]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I]], [[SWITCH_EARLY_TEST]] ]
// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I]], align 1, !tbaa [[TBAA5]]
// AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I:%.*]] = icmp eq i8 [[TMP2]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT:%.*]], label [[WHILE_BODY_I32_I:%.*]]
+// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT:%.*]], label [[WHILE_BODY_I32_I]]
// AMDGCNSPIRV: while.body.i32.i:
// AMDGCNSPIRV-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
// AMDGCNSPIRV-NEXT: [[OR_COND_I33_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I33_I]], label [[IF_END31_I_I:%.*]], label [[IF_ELSE_I_I:%.*]]
-// AMDGCNSPIRV: if.else.i.i:
// AMDGCNSPIRV-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
// AMDGCNSPIRV-NEXT: [[OR_COND33_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND33_I_I]], label [[IF_END31_I_I]], label [[IF_ELSE17_I_I:%.*]]
-// AMDGCNSPIRV: if.else17.i.i:
// AMDGCNSPIRV-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
// AMDGCNSPIRV-NEXT: [[OR_COND34_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND34_I_I]], label [[IF_END31_I_I]], label [[CLEANUP_I_I]]
-// AMDGCNSPIRV: if.end31.i.i:
-// AMDGCNSPIRV-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I32_I]] ], [ -87, [[IF_ELSE_I_I]] ], [ -55, [[IF_ELSE17_I_I]] ]
+// AMDGCNSPIRV-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND33_I_I]], i64 -87, i64 -55
+// AMDGCNSPIRV-NEXT: [[TMP7:%.*]] = select i1 [[OR_COND_I33_I]], i64 -48, i64 [[TMP6]]
+// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = or i1 [[OR_COND33_I_I]], [[OR_COND34_I_I]]
+// AMDGCNSPIRV-NEXT: [[OR_COND13:%.*]] = or i1 [[TMP8]], [[OR_COND_I33_I]]
// AMDGCNSPIRV-NEXT: [[MUL24_I_I:%.*]] = shl i64 [[__R_0_I30_I]], 4
// AMDGCNSPIRV-NEXT: [[CONV25_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// AMDGCNSPIRV-NEXT: [[ADD26_I_I:%.*]] = add i64 [[MUL24_I_I]], [[DOTSINK]]
+// AMDGCNSPIRV-NEXT: [[ADD26_I_I:%.*]] = add i64 [[MUL24_I_I]], [[TMP7]]
// AMDGCNSPIRV-NEXT: [[ADD28_I_I:%.*]] = add i64 [[ADD26_I_I]], [[CONV25_I_I]]
-// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I37_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I]], i64 1
-// AMDGCNSPIRV-NEXT: br label [[CLEANUP_I_I]]
-// AMDGCNSPIRV: cleanup.i.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I37_I]], [[IF_END31_I_I]] ], [ [[__TAGP_ADDR_0_I29_I]], [[IF_ELSE17_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[__R_2_I_I]] = phi i64 [ [[ADD28_I_I]], [[IF_END31_I_I]] ], [ [[__R_0_I30_I]], [[IF_ELSE17_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[COND_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I]] ], [ false, [[IF_ELSE17_I_I]] ]
-// AMDGCNSPIRV-NEXT: br i1 [[COND_I_I]], label [[WHILE_COND_I28_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]], !llvm.loop [[LOOP12]]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I_IDX:%.*]] = zext i1 [[OR_COND13]] to i64
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I]], i64 [[__TAGP_ADDR_1_I34_I_IDX]]
+// AMDGCNSPIRV-NEXT: [[__R_2_I_I]] = select i1 [[OR_COND13]], i64 [[ADD28_I_I]], i64 [[__R_0_I30_I]]
+// AMDGCNSPIRV-NEXT: [[TMP9:%.*]] = freeze i1 [[OR_COND_I33_I]]
+// AMDGCNSPIRV-NEXT: br i1 [[TMP9]], label [[WHILE_COND_I28_I]], label [[SWITCH_EARLY_TEST]]
+// AMDGCNSPIRV: switch.early.test:
+// AMDGCNSPIRV-NEXT: switch i8 [[TMP2]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]] [
+// AMDGCNSPIRV-NEXT: i8 102, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 101, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 100, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 99, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 98, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 97, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 70, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 69, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 68, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 67, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 66, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: i8 65, label [[WHILE_COND_I28_I]]
+// AMDGCNSPIRV-NEXT: ]
// AMDGCNSPIRV: while.cond.i.i:
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I_I:%.*]], [[WHILE_BODY_I_I:%.*]] ], [ [[INCDEC_PTR_I]], [[IF_THEN_I]] ]
// AMDGCNSPIRV-NEXT: [[__R_0_I_I:%.*]] = phi i64 [ [[__R_1_I_I:%.*]], [[WHILE_BODY_I_I]] ], [ 0, [[IF_THEN_I]] ]
-// AMDGCNSPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I]], align 1, !tbaa [[TBAA5]]
-// AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I]], align 1, !tbaa [[TBAA5]]
+// AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq i8 [[TMP10]], 0
// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]], label [[WHILE_BODY_I_I]]
// AMDGCNSPIRV: while.body.i.i:
-// AMDGCNSPIRV-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// AMDGCNSPIRV-NEXT: [[OR_COND_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// AMDGCNSPIRV-NEXT: [[TMP11:%.*]] = and i8 [[TMP10]], -8
+// AMDGCNSPIRV-NEXT: [[OR_COND_I_I:%.*]] = icmp eq i8 [[TMP11]], 48
// AMDGCNSPIRV-NEXT: [[MUL_I_I:%.*]] = shl i64 [[__R_0_I_I]], 3
-// AMDGCNSPIRV-NEXT: [[CONV5_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// AMDGCNSPIRV-NEXT: [[CONV5_I_I:%.*]] = zext nneg i8 [[TMP10]] to i64
// AMDGCNSPIRV-NEXT: [[ADD_I_I:%.*]] = add i64 [[MUL_I_I]], -48
// AMDGCNSPIRV-NEXT: [[SUB_I_I:%.*]] = add i64 [[ADD_I_I]], [[CONV5_I_I]]
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_IDX:%.*]] = zext i1 [[OR_COND_I_I]] to i64
@@ -395,14 +437,14 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) {
// AMDGCNSPIRV: while.cond.i14.i:
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I15_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I25_I:%.*]], [[WHILE_BODY_I18_I:%.*]] ], [ [[P]], [[ENTRY:%.*]] ]
// AMDGCNSPIRV-NEXT: [[__R_0_I16_I:%.*]] = phi i64 [ [[__R_1_I26_I:%.*]], [[WHILE_BODY_I18_I]] ], [ 0, [[ENTRY]] ]
-// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I]], align 1, !tbaa [[TBAA5]]
-// AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// AMDGCNSPIRV-NEXT: [[TMP12:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I]], align 1, !tbaa [[TBAA5]]
+// AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I:%.*]] = icmp eq i8 [[TMP12]], 0
// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I17_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]], label [[WHILE_BODY_I18_I]]
// AMDGCNSPIRV: while.body.i18.i:
-// AMDGCNSPIRV-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// AMDGCNSPIRV-NEXT: [[OR_COND_I19_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// AMDGCNSPIRV-NEXT: [[TMP13:%.*]] = add i8 [[TMP12]], -48
+// AMDGCNSPIRV-NEXT: [[OR_COND_I19_I:%.*]] = icmp ult i8 [[TMP13]], 10
// AMDGCNSPIRV-NEXT: [[MUL_I20_I:%.*]] = mul i64 [[__R_0_I16_I]], 10
-// AMDGCNSPIRV-NEXT: [[CONV5_I21_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// AMDGCNSPIRV-NEXT: [[CONV5_I21_I:%.*]] = zext nneg i8 [[TMP12]] to i64
// AMDGCNSPIRV-NEXT: [[ADD_I22_I:%.*]] = add i64 [[MUL_I20_I]], -48
// AMDGCNSPIRV-NEXT: [[SUB_I23_I:%.*]] = add i64 [[ADD_I22_I]], [[CONV5_I21_I]]
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_IDX:%.*]] = zext i1 [[OR_COND_I19_I]] to i64
@@ -410,7 +452,7 @@ extern "C" __device__ uint64_t test___make_mantissa_base16(const char *p) {
// AMDGCNSPIRV-NEXT: [[__R_1_I26_I]] = select i1 [[OR_COND_I19_I]], i64 [[SUB_I23_I]], i64 [[__R_0_I16_I]]
// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I]], label [[WHILE_COND_I14_I]], label [[_ZL15__MAKE_MANTISSAPKC_EXIT]], !llvm.loop [[LOOP11]]
// AMDGCNSPIRV: _ZL15__make_mantissaPKc.exit:
-// AMDGCNSPIRV-NEXT: [[RETVAL_0_I:%.*]] = phi i64 [ 0, [[WHILE_BODY_I_I]] ], [ [[__R_0_I_I]], [[WHILE_COND_I_I]] ], [ 0, [[CLEANUP_I_I]] ], [ [[__R_0_I30_I]], [[WHILE_COND_I28_I]] ], [ 0, [[WHILE_BODY_I18_I]] ], [ [[__R_0_I16_I]], [[WHILE_COND_I14_I]] ]
+// AMDGCNSPIRV-NEXT: [[RETVAL_0_I:%.*]] = phi i64 [ 0, [[WHILE_BODY_I_I]] ], [ [[__R_0_I_I]], [[WHILE_COND_I_I]] ], [ 0, [[SWITCH_EARLY_TEST]] ], [ [[__R_0_I30_I]], [[WHILE_COND_I28_I]] ], [ 0, [[WHILE_BODY_I18_I]] ], [ [[__R_0_I16_I]], [[WHILE_COND_I14_I]] ]
// AMDGCNSPIRV-NEXT: ret i64 [[RETVAL_0_I]]
//
extern "C" __device__ uint64_t test___make_mantissa(const char *p) {
@@ -461,22 +503,22 @@ extern "C" __device__ long long test_llabs(long x) {
// DEFAULT-LABEL: @test_acosf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X:%.*]]) #[[ATTR12:[0-9]+]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X:%.*]]) #[[ATTR14:[0-9]+]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_acosf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acos_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12:[0-9]+]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acos_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14:[0-9]+]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_acosf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X:%.*]]) #[[ATTR12:[0-9]+]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X:%.*]]) #[[ATTR14:[0-9]+]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_acosf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X:%.*]]) #[[ATTR12:[0-9]+]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acos_f32(float noundef [[X:%.*]]) #[[ATTR14:[0-9]+]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_acosf(
@@ -490,22 +532,22 @@ extern "C" __device__ float test_acosf(float x) {
// DEFAULT-LABEL: @test_acos(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_acos(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acos_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acos_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_acos(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_acos(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acos_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_acos(
@@ -519,22 +561,22 @@ extern "C" __device__ double test_acos(double x) {
// DEFAULT-LABEL: @test_acoshf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X:%.*]]) #[[ATTR13:[0-9]+]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X:%.*]]) #[[ATTR15:[0-9]+]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_acoshf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acosh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13:[0-9]+]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_acosh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15:[0-9]+]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_acoshf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X:%.*]]) #[[ATTR13:[0-9]+]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X:%.*]]) #[[ATTR15:[0-9]+]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_acoshf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X:%.*]]) #[[ATTR13:[0-9]+]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_acosh_f32(float noundef [[X:%.*]]) #[[ATTR15:[0-9]+]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_acoshf(
@@ -548,22 +590,22 @@ extern "C" __device__ float test_acoshf(float x) {
// DEFAULT-LABEL: @test_acosh(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_acosh(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acosh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_acosh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_acosh(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_acosh(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_acosh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_acosh(
@@ -577,22 +619,22 @@ extern "C" __device__ double test_acosh(double x) {
// DEFAULT-LABEL: @test_asinf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_asinf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_asinf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_asinf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_asinf(
@@ -606,22 +648,22 @@ extern "C" __device__ float test_asinf(float x) {
// DEFAULT-LABEL: @test_asin(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_asin(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asin_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asin_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_asin(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_asin(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asin_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_asin(
@@ -636,22 +678,22 @@ extern "C" __device__ double test_asin(double x) {
// DEFAULT-LABEL: @test_asinhf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_asinhf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asinh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_asinh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_asinhf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_asinhf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_asinh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_asinhf(
@@ -665,22 +707,22 @@ extern "C" __device__ float test_asinhf(float x) {
// DEFAULT-LABEL: @test_asinh(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_asinh(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asinh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_asinh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_asinh(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_asinh(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_asinh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_asinh(
@@ -694,22 +736,22 @@ extern "C" __device__ double test_asinh(double x) {
// DEFAULT-LABEL: @test_atan2f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_atan2f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan2_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan2_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_atan2f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_atan2f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan2_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_atan2f(
@@ -723,22 +765,22 @@ extern "C" __device__ float test_atan2f(float x, float y) {
// DEFAULT-LABEL: @test_atan2(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_atan2(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan2_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan2_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_atan2(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_atan2(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan2_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_atan2(
@@ -752,22 +794,22 @@ extern "C" __device__ double test_atan2(double x, double y) {
// DEFAULT-LABEL: @test_atanf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_atanf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atan_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_atanf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_atanf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atan_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_atanf(
@@ -781,22 +823,22 @@ extern "C" __device__ float test_atanf(float x) {
// DEFAULT-LABEL: @test_atan(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_atan(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atan_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_atan(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_atan(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atan_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_atan(
@@ -810,22 +852,22 @@ extern "C" __device__ double test_atan(double x) {
// DEFAULT-LABEL: @test_atanhf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_atanhf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atanh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_atanh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_atanhf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_atanhf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_atanh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_atanhf(
@@ -839,22 +881,22 @@ extern "C" __device__ float test_atanhf(float x) {
// DEFAULT-LABEL: @test_atanh(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_atanh(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atanh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_atanh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_atanh(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_atanh(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_atanh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_atanh(
@@ -868,22 +910,22 @@ extern "C" __device__ double test_atanh(double x) {
// DEFAULT-LABEL: @test_cbrtf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cbrtf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cbrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cbrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_cbrtf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_cbrtf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cbrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cbrtf(
@@ -897,22 +939,22 @@ extern "C" __device__ float test_cbrtf(float x) {
// DEFAULT-LABEL: @test_cbrt(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cbrt(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cbrt_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cbrt_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_cbrt(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_cbrt(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cbrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cbrt(
@@ -1042,22 +1084,22 @@ extern "C" __device__ double test_copysign(double x, double y) {
// DEFAULT-LABEL: @test_cosf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X:%.*]]) #[[ATTR14:[0-9]+]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X:%.*]]) #[[ATTR16:[0-9]+]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cosf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cos_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14:[0-9]+]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cos_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16:[0-9]+]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_cosf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR14:[0-9]+]]
+// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR16:[0-9]+]]
// APPROX-NEXT: ret float [[CALL_I1]]
//
// NCRDIV-LABEL: @test_cosf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X:%.*]]) #[[ATTR14:[0-9]+]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cos_f32(float noundef [[X:%.*]]) #[[ATTR16:[0-9]+]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cosf(
@@ -1071,22 +1113,22 @@ extern "C" __device__ float test_cosf(float x) {
// DEFAULT-LABEL: @test_cos(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cos(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cos_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cos_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_cos(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_cos(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cos_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cos(
@@ -1100,22 +1142,22 @@ extern "C" __device__ double test_cos(double x) {
// DEFAULT-LABEL: @test_coshf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_coshf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cosh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cosh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_coshf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_coshf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cosh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_coshf(
@@ -1129,22 +1171,22 @@ extern "C" __device__ float test_coshf(float x) {
// DEFAULT-LABEL: @test_cosh(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cosh(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cosh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cosh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_cosh(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_cosh(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cosh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cosh(
@@ -1158,22 +1200,22 @@ extern "C" __device__ double test_cosh(double x) {
// DEFAULT-LABEL: @test_cospif(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cospif(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cospi_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_cospi_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_cospif(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_cospif(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_cospi_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cospif(
@@ -1187,22 +1229,22 @@ extern "C" __device__ float test_cospif(float x) {
// DEFAULT-LABEL: @test_cospi(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cospi(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cospi_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_cospi_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_cospi(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_cospi(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_cospi_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cospi(
@@ -1216,22 +1258,22 @@ extern "C" __device__ double test_cospi(double x) {
// DEFAULT-LABEL: @test_cyl_bessel_i0f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cyl_bessel_i0f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i0_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i0_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_cyl_bessel_i0f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_cyl_bessel_i0f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cyl_bessel_i0f(
@@ -1245,22 +1287,22 @@ extern "C" __device__ float test_cyl_bessel_i0f(float x) {
// DEFAULT-LABEL: @test_cyl_bessel_i0(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cyl_bessel_i0(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i0_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i0_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_cyl_bessel_i0(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_cyl_bessel_i0(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cyl_bessel_i0(
@@ -1274,22 +1316,22 @@ extern "C" __device__ double test_cyl_bessel_i0(double x) {
// DEFAULT-LABEL: @test_cyl_bessel_i1f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cyl_bessel_i1f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_i1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_cyl_bessel_i1f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_cyl_bessel_i1f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_i1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cyl_bessel_i1f(
@@ -1303,22 +1345,22 @@ extern "C" __device__ float test_cyl_bessel_i1f(float x) {
// DEFAULT-LABEL: @test_cyl_bessel_i1(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_cyl_bessel_i1(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_i1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_cyl_bessel_i1(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_cyl_bessel_i1(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_i1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_cyl_bessel_i1(
@@ -1332,22 +1374,22 @@ extern "C" __device__ double test_cyl_bessel_i1(double x) {
// DEFAULT-LABEL: @test_erfcf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_erfcf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfc_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfc_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_erfcf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_erfcf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfc_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_erfcf(
@@ -1361,22 +1403,22 @@ extern "C" __device__ float test_erfcf(float x) {
// DEFAULT-LABEL: @test_erfc(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_erfc(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfc_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfc_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_erfc(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_erfc(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfc_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_erfc(
@@ -1390,22 +1432,22 @@ extern "C" __device__ double test_erfc(double x) {
// DEFAULT-LABEL: @test_erfinvf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_erfinvf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfinv_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_erfinv_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_erfinvf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_erfinvf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_erfinv_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_erfinvf(
@@ -1419,22 +1461,22 @@ extern "C" __device__ float test_erfinvf(float x) {
// DEFAULT-LABEL: @test_erfinv(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_erfinv(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfinv_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_erfinv_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_erfinv(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_erfinv(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_erfinv_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_erfinv(
@@ -1477,22 +1519,22 @@ extern "C" __device__ float test_exp10f(float x) {
// DEFAULT-LABEL: @test_exp10(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_exp10(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp10_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp10_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_exp10(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_exp10(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp10_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_exp10(
@@ -1535,22 +1577,22 @@ extern "C" __device__ float test_exp2f(float x) {
// DEFAULT-LABEL: @test_exp2(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_exp2(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp2_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp2_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_exp2(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_exp2(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp2_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_exp2(
@@ -1593,22 +1635,22 @@ extern "C" __device__ float test_expf(float x) {
// DEFAULT-LABEL: @test_exp(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_exp(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_exp_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_exp(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_exp(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_exp_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_exp(
@@ -1622,22 +1664,22 @@ extern "C" __device__ double test_exp(double x) {
// DEFAULT-LABEL: @test_expm1f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_expm1f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_expm1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_expm1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_expm1f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_expm1f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_expm1_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_expm1f(
@@ -1651,22 +1693,22 @@ extern "C" __device__ float test_expm1f(float x) {
// DEFAULT-LABEL: @test_expm1(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_expm1(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_expm1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_expm1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_expm1(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_expm1(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_expm1_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_expm1(
@@ -1738,22 +1780,22 @@ extern "C" __device__ double test_fabs(double x) {
// DEFAULT-LABEL: @test_fdimf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_fdimf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fdim_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fdim_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_fdimf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_fdimf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fdim_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_fdimf(
@@ -1767,22 +1809,22 @@ extern "C" __device__ float test_fdimf(float x, float y) {
// DEFAULT-LABEL: @test_fdim(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_fdim(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fdim_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fdim_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_fdim(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_fdim(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fdim_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_fdim(
@@ -2086,22 +2128,22 @@ extern "C" __device__ double test_fmin(double x, double y) {
// DEFAULT-LABEL: @test_fmodf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_fmodf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fmod_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_fmod_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_fmodf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_fmodf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_fmod_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_fmodf(
@@ -2115,22 +2157,22 @@ extern "C" __device__ float test_fmodf(float x, float y) {
// DEFAULT-LABEL: @test_fmod(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_fmod(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fmod_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_fmod_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_fmod(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_fmod(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_fmod_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_fmod(
@@ -2178,7 +2220,7 @@ extern "C" __device__ double test_fmod(double x, double y) {
// AMDGCNSPIRV-NEXT: entry:
// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = tail call addrspace(4) { float, i32 } @llvm.frexp.f32.i32(float [[X:%.*]])
// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = extractvalue { float, i32 } [[TMP0]], 1
-// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA13:![0-9]+]]
+// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA12:![0-9]+]]
// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = extractvalue { float, i32 } [[TMP0]], 0
// AMDGCNSPIRV-NEXT: ret float [[TMP2]]
//
@@ -2222,7 +2264,7 @@ extern "C" __device__ float test_frexpf(float x, int* y) {
// AMDGCNSPIRV-NEXT: entry:
// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = tail call addrspace(4) { double, i32 } @llvm.frexp.f64.i32(double [[X:%.*]])
// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = extractvalue { double, i32 } [[TMP0]], 1
-// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA13]]
+// AMDGCNSPIRV-NEXT: store i32 [[TMP1]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA12]]
// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = extractvalue { double, i32 } [[TMP0]], 0
// AMDGCNSPIRV-NEXT: ret double [[TMP2]]
//
@@ -2232,22 +2274,22 @@ extern "C" __device__ double test_frexp(double x, int* y) {
// DEFAULT-LABEL: @test_hypotf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_hypotf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_hypot_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_hypot_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_hypotf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_hypotf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_hypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_hypotf(
@@ -2261,22 +2303,22 @@ extern "C" __device__ float test_hypotf(float x, float y) {
// DEFAULT-LABEL: @test_hypot(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_hypot(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_hypot_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_hypot_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_hypot(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_hypot(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_hypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_hypot(
@@ -2290,22 +2332,22 @@ extern "C" __device__ double test_hypot(double x, double y) {
// DEFAULT-LABEL: @test_ilogbf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret i32 [[CALL_I]]
//
// FINITEONLY-LABEL: @test_ilogbf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret i32 [[CALL_I]]
//
// APPROX-LABEL: @test_ilogbf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret i32 [[CALL_I]]
//
// NCRDIV-LABEL: @test_ilogbf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret i32 [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_ilogbf(
@@ -2319,22 +2361,22 @@ extern "C" __device__ int test_ilogbf(float x) {
// DEFAULT-LABEL: @test_ilogb(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret i32 [[CALL_I]]
//
// FINITEONLY-LABEL: @test_ilogb(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret i32 [[CALL_I]]
//
// APPROX-LABEL: @test_ilogb(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret i32 [[CALL_I]]
//
// NCRDIV-LABEL: @test_ilogb(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call noundef i32 @__ocml_ilogb_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret i32 [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_ilogb(
@@ -2556,22 +2598,22 @@ extern "C" __device__ BOOL_TYPE test___isnan(double x) {
// DEFAULT-LABEL: @test_j0f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_j0f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_j0f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_j0f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_j0f(
@@ -2585,22 +2627,22 @@ extern "C" __device__ float test_j0f(float x) {
// DEFAULT-LABEL: @test_j0(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_j0(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_j0(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_j0(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_j0(
@@ -2614,22 +2656,22 @@ extern "C" __device__ double test_j0(double x) {
// DEFAULT-LABEL: @test_j1f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_j1f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_j1f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_j1f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_j1f(
@@ -2643,22 +2685,22 @@ extern "C" __device__ float test_j1f(float x) {
// DEFAULT-LABEL: @test_j1(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_j1(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_j1(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_j1(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_j1(
@@ -2677,14 +2719,14 @@ extern "C" __device__ double test_j1(double x) {
// DEFAULT-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// DEFAULT-NEXT: ]
// DEFAULT: if.then.i:
-// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL3JNFIF_EXIT:%.*]]
// DEFAULT: if.then2.i:
-// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL3JNFIF_EXIT]]
// DEFAULT: if.end4.i:
-// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]]
-// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]]
+// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// DEFAULT-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3JNFIF_EXIT]]
// DEFAULT: for.body.i:
@@ -2710,14 +2752,14 @@ extern "C" __device__ double test_j1(double x) {
// FINITEONLY-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// FINITEONLY-NEXT: ]
// FINITEONLY: if.then.i:
-// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL3JNFIF_EXIT:%.*]]
// FINITEONLY: if.then2.i:
-// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL3JNFIF_EXIT]]
// FINITEONLY: if.end4.i:
-// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
-// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
+// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_j1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// FINITEONLY-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3JNFIF_EXIT]]
// FINITEONLY: for.body.i:
@@ -2743,14 +2785,14 @@ extern "C" __device__ double test_j1(double x) {
// APPROX-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// APPROX-NEXT: ]
// APPROX: if.then.i:
-// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y:%.*]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL3JNFIF_EXIT:%.*]]
// APPROX: if.then2.i:
-// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL3JNFIF_EXIT]]
// APPROX: if.end4.i:
-// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]]
-// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]]
+// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// APPROX-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3JNFIF_EXIT]]
// APPROX: for.body.i:
@@ -2776,14 +2818,14 @@ extern "C" __device__ double test_j1(double x) {
// NCRDIV-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// NCRDIV-NEXT: ]
// NCRDIV: if.then.i:
-// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL3JNFIF_EXIT:%.*]]
// NCRDIV: if.then2.i:
-// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL3JNFIF_EXIT]]
// NCRDIV: if.end4.i:
-// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR14]]
-// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_j0_f32(float noundef [[Y]]) #[[ATTR16]]
+// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_j1_f32(float noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// NCRDIV-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3JNFIF_EXIT]]
// NCRDIV: for.body.i:
@@ -2830,7 +2872,7 @@ extern "C" __device__ double test_j1(double x) {
// AMDGCNSPIRV-NEXT: [[SUB_I]] = fsub contract float [[MUL8_I]], [[__X0_0_I2]]
// AMDGCNSPIRV-NEXT: [[INC_I]] = add nuw nsw i32 [[__I_0_I4]], 1
// AMDGCNSPIRV-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC_I]], [[X]]
-// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL3JNFIF_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP15:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL3JNFIF_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP14:![0-9]+]]
// AMDGCNSPIRV: _ZL3jnfif.exit:
// AMDGCNSPIRV-NEXT: [[RETVAL_0_I:%.*]] = phi float [ [[CALL_I20_I]], [[IF_THEN_I]] ], [ [[CALL_I22_I]], [[IF_THEN2_I]] ], [ [[CALL_I21_I]], [[IF_END4_I]] ], [ [[SUB_I]], [[FOR_BODY_I]] ]
// AMDGCNSPIRV-NEXT: ret float [[RETVAL_0_I]]
@@ -2846,14 +2888,14 @@ extern "C" __device__ float test_jnf(int x, float y) {
// DEFAULT-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// DEFAULT-NEXT: ]
// DEFAULT: if.then.i:
-// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL2JNID_EXIT:%.*]]
// DEFAULT: if.then2.i:
-// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL2JNID_EXIT]]
// DEFAULT: if.end4.i:
-// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]]
-// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]]
+// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// DEFAULT-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2JNID_EXIT]]
// DEFAULT: for.body.i:
@@ -2879,14 +2921,14 @@ extern "C" __device__ float test_jnf(int x, float y) {
// FINITEONLY-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// FINITEONLY-NEXT: ]
// FINITEONLY: if.then.i:
-// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL2JNID_EXIT:%.*]]
// FINITEONLY: if.then2.i:
-// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL2JNID_EXIT]]
// FINITEONLY: if.end4.i:
-// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
-// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
+// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_j1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// FINITEONLY-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2JNID_EXIT]]
// FINITEONLY: for.body.i:
@@ -2912,14 +2954,14 @@ extern "C" __device__ float test_jnf(int x, float y) {
// APPROX-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// APPROX-NEXT: ]
// APPROX: if.then.i:
-// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y:%.*]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL2JNID_EXIT:%.*]]
// APPROX: if.then2.i:
-// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL2JNID_EXIT]]
// APPROX: if.end4.i:
-// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]]
-// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]]
+// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// APPROX-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2JNID_EXIT]]
// APPROX: for.body.i:
@@ -2945,14 +2987,14 @@ extern "C" __device__ float test_jnf(int x, float y) {
// NCRDIV-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// NCRDIV-NEXT: ]
// NCRDIV: if.then.i:
-// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL2JNID_EXIT:%.*]]
// NCRDIV: if.then2.i:
-// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL2JNID_EXIT]]
// NCRDIV: if.end4.i:
-// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR14]]
-// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_j0_f64(double noundef [[Y]]) #[[ATTR16]]
+// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_j1_f64(double noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// NCRDIV-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2JNID_EXIT]]
// NCRDIV: for.body.i:
@@ -2999,7 +3041,7 @@ extern "C" __device__ float test_jnf(int x, float y) {
// AMDGCNSPIRV-NEXT: [[SUB_I]] = fsub contract double [[MUL8_I]], [[__X0_0_I2]]
// AMDGCNSPIRV-NEXT: [[INC_I]] = add nuw nsw i32 [[__I_0_I4]], 1
// AMDGCNSPIRV-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC_I]], [[X]]
-// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL2JNID_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP16:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL2JNID_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP15:![0-9]+]]
// AMDGCNSPIRV: _ZL2jnid.exit:
// AMDGCNSPIRV-NEXT: [[RETVAL_0_I:%.*]] = phi double [ [[CALL_I20_I]], [[IF_THEN_I]] ], [ [[CALL_I22_I]], [[IF_THEN2_I]] ], [ [[CALL_I21_I]], [[IF_END4_I]] ], [ [[SUB_I]], [[FOR_BODY_I]] ]
// AMDGCNSPIRV-NEXT: ret double [[RETVAL_0_I]]
@@ -3068,22 +3110,22 @@ extern "C" __device__ double test_ldexp(double x, int y) {
// DEFAULT-LABEL: @test_lgammaf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_lgammaf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_lgamma_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_lgamma_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_lgammaf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_lgammaf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_lgamma_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_lgammaf(
@@ -3097,22 +3139,22 @@ extern "C" __device__ float test_lgammaf(float x) {
// DEFAULT-LABEL: @test_lgamma(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_lgamma(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_lgamma_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_lgamma_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_lgamma(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_lgamma(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_lgamma_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_lgamma(
@@ -3291,22 +3333,22 @@ extern "C" __device__ float test_log10f(float x) {
// DEFAULT-LABEL: @test_log10(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_log10(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log10_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log10_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_log10(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_log10(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log10_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_log10(
@@ -3320,22 +3362,22 @@ extern "C" __device__ double test_log10(double x) {
// DEFAULT-LABEL: @test_log1pf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_log1pf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_log1p_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_log1p_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_log1pf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_log1pf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_log1p_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_log1pf(
@@ -3349,22 +3391,22 @@ extern "C" __device__ float test_log1pf(float x) {
// DEFAULT-LABEL: @test_log1p(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_log1p(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log1p_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log1p_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_log1p(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_log1p(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log1p_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_log1p(
@@ -3407,22 +3449,22 @@ extern "C" __device__ float test_log2f(float x) {
// DEFAULT-LABEL: @test_log2(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_log2(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log2_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_log2_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_log2(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_log2(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_log2_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_log2(
@@ -3436,22 +3478,22 @@ extern "C" __device__ double test_log2(double x) {
// DEFAULT-LABEL: @test_logbf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_logbf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_logb_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_logb_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_logbf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_logbf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_logb_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_logbf(
@@ -3465,22 +3507,22 @@ extern "C" __device__ float test_logbf(float x) {
// DEFAULT-LABEL: @test_logb(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_logb(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_logb_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_logb_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_logb(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_logb(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_logb_f64(double noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_logb(
@@ -3660,41 +3702,41 @@ extern "C" __device__ long int test_lround(double x) {
// DEFAULT-LABEL: @test_modff(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16:![0-9]+]]
// DEFAULT-NEXT: store float [[TMP0]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_modff(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_modf_f32(float noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_modf_f32(float noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16:![0-9]+]]
// FINITEONLY-NEXT: store float [[TMP0]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_modff(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16:![0-9]+]]
// APPROX-NEXT: store float [[TMP0]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_modff(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15:[0-9]+]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17:[0-9]+]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_modf_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA17:![0-9]+]]
// NCRDIV-NEXT: store float [[TMP0]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA17]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_modff(
@@ -3703,8 +3745,8 @@ extern "C" __device__ long int test_lround(double x) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15:[0-9]+]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func noundef addrspace(4) float @__ocml_modf_f32(float noundef [[X:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA17:![0-9]+]]
-// AMDGCNSPIRV-NEXT: store float [[TMP0]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA17]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA16:![0-9]+]]
+// AMDGCNSPIRV-NEXT: store float [[TMP0]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret float [[CALL_I]]
//
@@ -3715,41 +3757,41 @@ extern "C" __device__ float test_modff(float x, float* y) {
// DEFAULT-LABEL: @test_modf(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18:![0-9]+]]
// DEFAULT-NEXT: store double [[TMP0]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_modf(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_modf_f64(double noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_modf_f64(double noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18:![0-9]+]]
// FINITEONLY-NEXT: store double [[TMP0]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_modf(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18:![0-9]+]]
// APPROX-NEXT: store double [[TMP0]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_modf(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_modf_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA19:![0-9]+]]
// NCRDIV-NEXT: store double [[TMP0]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA19]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_modf(
@@ -3758,8 +3800,8 @@ extern "C" __device__ float test_modff(float x, float* y) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 8, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func noundef addrspace(4) double @__ocml_modf_f64(double noundef [[X:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__TMP_ASCAST_I]], align 8, !tbaa [[TBAA19:![0-9]+]]
-// AMDGCNSPIRV-NEXT: store double [[TMP0]], ptr addrspace(4) [[Y:%.*]], align 8, !tbaa [[TBAA19]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__TMP_ASCAST_I]], align 8, !tbaa [[TBAA18:![0-9]+]]
+// AMDGCNSPIRV-NEXT: store double [[TMP0]], ptr addrspace(4) [[Y:%.*]], align 8, !tbaa [[TBAA18]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 8, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret double [[CALL_I]]
//
@@ -3785,46 +3827,57 @@ extern "C" __device__ double test_modf(double x, double* y) {
// DEFAULT-NEXT: [[__TAGP_ADDR_0_I31_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I37_I_I:%.*]], [[CLEANUP_I36_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I30_I_I_PREHEADER]] ]
// DEFAULT-NEXT: [[__R_0_I32_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I36_I_I]] ], [ 0, [[WHILE_COND_I30_I_I_PREHEADER]] ]
// DEFAULT-NEXT: [[TMP2:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I31_I_I]], align 1, !tbaa [[TBAA4]]
-// DEFAULT-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
+// DEFAULT-NEXT: [[DOTFR14:%.*]] = freeze i8 [[TMP2]]
+// DEFAULT-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[DOTFR14]], 0
// DEFAULT-NEXT: br i1 [[CMP_NOT_I33_I_I]], label [[_ZL4NANFPKC_EXIT:%.*]], label [[WHILE_BODY_I34_I_I:%.*]]
// DEFAULT: while.body.i34.i.i:
-// DEFAULT-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
+// DEFAULT-NEXT: [[TMP3:%.*]] = add i8 [[DOTFR14]], -48
// DEFAULT-NEXT: [[OR_COND_I35_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// DEFAULT-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// DEFAULT: if.else.i.i.i:
-// DEFAULT-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
+// DEFAULT-NEXT: [[TMP4:%.*]] = add i8 [[DOTFR14]], -97
// DEFAULT-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// DEFAULT-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// DEFAULT: if.else17.i.i.i:
-// DEFAULT-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
-// DEFAULT-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// DEFAULT-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I36_I_I]]
+// DEFAULT-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// DEFAULT-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND_I35_I_I]], i64 -48, i64 [[TMP5]]
+// DEFAULT-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// DEFAULT: switch.early.test:
+// DEFAULT-NEXT: switch i8 [[DOTFR14]], label [[CLEANUP_I36_I_I]] [
+// DEFAULT-NEXT: i8 102, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 101, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 100, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 99, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 98, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 97, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 70, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 69, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 68, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 67, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 66, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 65, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: ]
// DEFAULT: if.end31.i.i.i:
-// DEFAULT-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I34_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
// DEFAULT-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I32_I_I]], 4
-// DEFAULT-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// DEFAULT-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// DEFAULT-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[DOTFR14]] to i64
+// DEFAULT-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP6]]
// DEFAULT-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
// DEFAULT-NEXT: [[INCDEC_PTR_I40_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I31_I_I]], i64 1
// DEFAULT-NEXT: br label [[CLEANUP_I36_I_I]]
// DEFAULT: cleanup.i36.i.i:
-// DEFAULT-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[IF_ELSE17_I_I_I]] ]
-// DEFAULT-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[IF_ELSE17_I_I_I]] ]
-// DEFAULT-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
+// DEFAULT-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[SWITCH_EARLY_TEST]] ]
+// DEFAULT-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[SWITCH_EARLY_TEST]] ]
+// DEFAULT-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// DEFAULT-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I30_I_I]], label [[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP11]]
// DEFAULT: while.cond.i.i.i:
// DEFAULT-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// DEFAULT-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// DEFAULT-NEXT: [[TMP6:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
-// DEFAULT-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// DEFAULT-NEXT: [[TMP7:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
+// DEFAULT-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 0
// DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I_I_I:%.*]]
// DEFAULT: while.body.i.i.i:
-// DEFAULT-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// DEFAULT-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// DEFAULT-NEXT: [[TMP8:%.*]] = and i8 [[TMP7]], -8
+// DEFAULT-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 48
// DEFAULT-NEXT: br i1 [[OR_COND_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[CLEANUP_I_I_I]]
// DEFAULT: if.then.i.i.i:
// DEFAULT-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// DEFAULT-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// DEFAULT-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP7]] to i64
// DEFAULT-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// DEFAULT-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// DEFAULT-NEXT: [[INCDEC_PTR_I_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I]], i64 1
@@ -3836,16 +3889,16 @@ extern "C" __device__ double test_modf(double x, double* y) {
// DEFAULT: while.cond.i14.i.i:
// DEFAULT-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I21_I_I:%.*]], [[CLEANUP_I20_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// DEFAULT-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I22_I_I:%.*]], [[CLEANUP_I20_I_I]] ], [ 0, [[ENTRY]] ]
-// DEFAULT-NEXT: [[TMP8:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
-// DEFAULT-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// DEFAULT-NEXT: [[TMP9:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
+// DEFAULT-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP9]], 0
// DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I18_I_I:%.*]]
// DEFAULT: while.body.i18.i.i:
-// DEFAULT-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// DEFAULT-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// DEFAULT-NEXT: [[TMP10:%.*]] = add i8 [[TMP9]], -48
+// DEFAULT-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP10]], 10
// DEFAULT-NEXT: br i1 [[OR_COND_I19_I_I]], label [[IF_THEN_I24_I_I:%.*]], label [[CLEANUP_I20_I_I]]
// DEFAULT: if.then.i24.i.i:
// DEFAULT-NEXT: [[MUL_I25_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// DEFAULT-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// DEFAULT-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP9]] to i64
// DEFAULT-NEXT: [[ADD_I27_I_I:%.*]] = add i64 [[MUL_I25_I_I]], -48
// DEFAULT-NEXT: [[SUB_I28_I_I:%.*]] = add i64 [[ADD_I27_I_I]], [[CONV5_I26_I_I]]
// DEFAULT-NEXT: [[INCDEC_PTR_I29_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I]], i64 1
@@ -3859,8 +3912,8 @@ extern "C" __device__ double test_modf(double x, double* y) {
// DEFAULT-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32
// DEFAULT-NEXT: [[BF_VALUE_I:%.*]] = and i32 [[CONV_I]], 4194303
// DEFAULT-NEXT: [[BF_SET9_I:%.*]] = or disjoint i32 [[BF_VALUE_I]], 2143289344
-// DEFAULT-NEXT: [[TMP10:%.*]] = bitcast i32 [[BF_SET9_I]] to float
-// DEFAULT-NEXT: ret float [[TMP10]]
+// DEFAULT-NEXT: [[TMP11:%.*]] = bitcast i32 [[BF_SET9_I]] to float
+// DEFAULT-NEXT: ret float [[TMP11]]
//
// FINITEONLY-LABEL: @test_nanf(
// FINITEONLY-NEXT: entry:
@@ -3884,46 +3937,57 @@ extern "C" __device__ double test_modf(double x, double* y) {
// APPROX-NEXT: [[__TAGP_ADDR_0_I31_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I37_I_I:%.*]], [[CLEANUP_I36_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I30_I_I_PREHEADER]] ]
// APPROX-NEXT: [[__R_0_I32_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I36_I_I]] ], [ 0, [[WHILE_COND_I30_I_I_PREHEADER]] ]
// APPROX-NEXT: [[TMP2:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I31_I_I]], align 1, !tbaa [[TBAA4]]
-// APPROX-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
+// APPROX-NEXT: [[DOTFR14:%.*]] = freeze i8 [[TMP2]]
+// APPROX-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[DOTFR14]], 0
// APPROX-NEXT: br i1 [[CMP_NOT_I33_I_I]], label [[_ZL4NANFPKC_EXIT:%.*]], label [[WHILE_BODY_I34_I_I:%.*]]
// APPROX: while.body.i34.i.i:
-// APPROX-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
+// APPROX-NEXT: [[TMP3:%.*]] = add i8 [[DOTFR14]], -48
// APPROX-NEXT: [[OR_COND_I35_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// APPROX-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// APPROX: if.else.i.i.i:
-// APPROX-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
+// APPROX-NEXT: [[TMP4:%.*]] = add i8 [[DOTFR14]], -97
// APPROX-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// APPROX-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// APPROX: if.else17.i.i.i:
-// APPROX-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
-// APPROX-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// APPROX-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I36_I_I]]
+// APPROX-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// APPROX-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND_I35_I_I]], i64 -48, i64 [[TMP5]]
+// APPROX-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// APPROX: switch.early.test:
+// APPROX-NEXT: switch i8 [[DOTFR14]], label [[CLEANUP_I36_I_I]] [
+// APPROX-NEXT: i8 102, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 101, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 100, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 99, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 98, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 97, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 70, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 69, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 68, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 67, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 66, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 65, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: ]
// APPROX: if.end31.i.i.i:
-// APPROX-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I34_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
// APPROX-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I32_I_I]], 4
-// APPROX-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// APPROX-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// APPROX-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[DOTFR14]] to i64
+// APPROX-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP6]]
// APPROX-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
// APPROX-NEXT: [[INCDEC_PTR_I40_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I31_I_I]], i64 1
// APPROX-NEXT: br label [[CLEANUP_I36_I_I]]
// APPROX: cleanup.i36.i.i:
-// APPROX-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[IF_ELSE17_I_I_I]] ]
-// APPROX-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[IF_ELSE17_I_I_I]] ]
-// APPROX-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
+// APPROX-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[SWITCH_EARLY_TEST]] ]
+// APPROX-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[SWITCH_EARLY_TEST]] ]
+// APPROX-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// APPROX-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I30_I_I]], label [[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP11]]
// APPROX: while.cond.i.i.i:
// APPROX-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// APPROX-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// APPROX-NEXT: [[TMP6:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
-// APPROX-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// APPROX-NEXT: [[TMP7:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
+// APPROX-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 0
// APPROX-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I_I_I:%.*]]
// APPROX: while.body.i.i.i:
-// APPROX-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// APPROX-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// APPROX-NEXT: [[TMP8:%.*]] = and i8 [[TMP7]], -8
+// APPROX-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 48
// APPROX-NEXT: br i1 [[OR_COND_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[CLEANUP_I_I_I]]
// APPROX: if.then.i.i.i:
// APPROX-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// APPROX-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// APPROX-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP7]] to i64
// APPROX-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// APPROX-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// APPROX-NEXT: [[INCDEC_PTR_I_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I]], i64 1
@@ -3935,16 +3999,16 @@ extern "C" __device__ double test_modf(double x, double* y) {
// APPROX: while.cond.i14.i.i:
// APPROX-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I21_I_I:%.*]], [[CLEANUP_I20_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// APPROX-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I22_I_I:%.*]], [[CLEANUP_I20_I_I]] ], [ 0, [[ENTRY]] ]
-// APPROX-NEXT: [[TMP8:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
-// APPROX-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// APPROX-NEXT: [[TMP9:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
+// APPROX-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP9]], 0
// APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I18_I_I:%.*]]
// APPROX: while.body.i18.i.i:
-// APPROX-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// APPROX-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// APPROX-NEXT: [[TMP10:%.*]] = add i8 [[TMP9]], -48
+// APPROX-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP10]], 10
// APPROX-NEXT: br i1 [[OR_COND_I19_I_I]], label [[IF_THEN_I24_I_I:%.*]], label [[CLEANUP_I20_I_I]]
// APPROX: if.then.i24.i.i:
// APPROX-NEXT: [[MUL_I25_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// APPROX-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// APPROX-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP9]] to i64
// APPROX-NEXT: [[ADD_I27_I_I:%.*]] = add i64 [[MUL_I25_I_I]], -48
// APPROX-NEXT: [[SUB_I28_I_I:%.*]] = add i64 [[ADD_I27_I_I]], [[CONV5_I26_I_I]]
// APPROX-NEXT: [[INCDEC_PTR_I29_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I]], i64 1
@@ -3958,8 +4022,8 @@ extern "C" __device__ double test_modf(double x, double* y) {
// APPROX-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32
// APPROX-NEXT: [[BF_VALUE_I:%.*]] = and i32 [[CONV_I]], 4194303
// APPROX-NEXT: [[BF_SET9_I:%.*]] = or disjoint i32 [[BF_VALUE_I]], 2143289344
-// APPROX-NEXT: [[TMP10:%.*]] = bitcast i32 [[BF_SET9_I]] to float
-// APPROX-NEXT: ret float [[TMP10]]
+// APPROX-NEXT: [[TMP11:%.*]] = bitcast i32 [[BF_SET9_I]] to float
+// APPROX-NEXT: ret float [[TMP11]]
//
// NCRDIV-LABEL: @test_nanf(
// NCRDIV-NEXT: entry:
@@ -3979,46 +4043,57 @@ extern "C" __device__ double test_modf(double x, double* y) {
// NCRDIV-NEXT: [[__TAGP_ADDR_0_I31_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I37_I_I:%.*]], [[CLEANUP_I36_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I30_I_I_PREHEADER]] ]
// NCRDIV-NEXT: [[__R_0_I32_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I36_I_I]] ], [ 0, [[WHILE_COND_I30_I_I_PREHEADER]] ]
// NCRDIV-NEXT: [[TMP2:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I31_I_I]], align 1, !tbaa [[TBAA4]]
-// NCRDIV-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
+// NCRDIV-NEXT: [[DOTFR14:%.*]] = freeze i8 [[TMP2]]
+// NCRDIV-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[DOTFR14]], 0
// NCRDIV-NEXT: br i1 [[CMP_NOT_I33_I_I]], label [[_ZL4NANFPKC_EXIT:%.*]], label [[WHILE_BODY_I34_I_I:%.*]]
// NCRDIV: while.body.i34.i.i:
-// NCRDIV-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
+// NCRDIV-NEXT: [[TMP3:%.*]] = add i8 [[DOTFR14]], -48
// NCRDIV-NEXT: [[OR_COND_I35_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// NCRDIV-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// NCRDIV: if.else.i.i.i:
-// NCRDIV-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
+// NCRDIV-NEXT: [[TMP4:%.*]] = add i8 [[DOTFR14]], -97
// NCRDIV-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// NCRDIV-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// NCRDIV: if.else17.i.i.i:
-// NCRDIV-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
-// NCRDIV-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// NCRDIV-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I36_I_I]]
+// NCRDIV-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// NCRDIV-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND_I35_I_I]], i64 -48, i64 [[TMP5]]
+// NCRDIV-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// NCRDIV: switch.early.test:
+// NCRDIV-NEXT: switch i8 [[DOTFR14]], label [[CLEANUP_I36_I_I]] [
+// NCRDIV-NEXT: i8 102, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 101, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 100, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 99, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 98, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 97, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 70, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 69, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 68, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 67, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 66, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 65, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: ]
// NCRDIV: if.end31.i.i.i:
-// NCRDIV-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I34_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
// NCRDIV-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I32_I_I]], 4
-// NCRDIV-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// NCRDIV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// NCRDIV-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[DOTFR14]] to i64
+// NCRDIV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP6]]
// NCRDIV-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
// NCRDIV-NEXT: [[INCDEC_PTR_I40_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I31_I_I]], i64 1
// NCRDIV-NEXT: br label [[CLEANUP_I36_I_I]]
// NCRDIV: cleanup.i36.i.i:
-// NCRDIV-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[IF_ELSE17_I_I_I]] ]
-// NCRDIV-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[IF_ELSE17_I_I_I]] ]
-// NCRDIV-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
+// NCRDIV-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[SWITCH_EARLY_TEST]] ]
+// NCRDIV-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[SWITCH_EARLY_TEST]] ]
+// NCRDIV-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// NCRDIV-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I30_I_I]], label [[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP11]]
// NCRDIV: while.cond.i.i.i:
// NCRDIV-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// NCRDIV-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// NCRDIV-NEXT: [[TMP6:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
-// NCRDIV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// NCRDIV-NEXT: [[TMP7:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
+// NCRDIV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 0
// NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I_I_I:%.*]]
// NCRDIV: while.body.i.i.i:
-// NCRDIV-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// NCRDIV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// NCRDIV-NEXT: [[TMP8:%.*]] = and i8 [[TMP7]], -8
+// NCRDIV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 48
// NCRDIV-NEXT: br i1 [[OR_COND_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[CLEANUP_I_I_I]]
// NCRDIV: if.then.i.i.i:
// NCRDIV-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// NCRDIV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// NCRDIV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP7]] to i64
// NCRDIV-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// NCRDIV-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// NCRDIV-NEXT: [[INCDEC_PTR_I_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I]], i64 1
@@ -4030,16 +4105,16 @@ extern "C" __device__ double test_modf(double x, double* y) {
// NCRDIV: while.cond.i14.i.i:
// NCRDIV-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I21_I_I:%.*]], [[CLEANUP_I20_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// NCRDIV-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I22_I_I:%.*]], [[CLEANUP_I20_I_I]] ], [ 0, [[ENTRY]] ]
-// NCRDIV-NEXT: [[TMP8:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
-// NCRDIV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// NCRDIV-NEXT: [[TMP9:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
+// NCRDIV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP9]], 0
// NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I18_I_I:%.*]]
// NCRDIV: while.body.i18.i.i:
-// NCRDIV-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// NCRDIV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// NCRDIV-NEXT: [[TMP10:%.*]] = add i8 [[TMP9]], -48
+// NCRDIV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP10]], 10
// NCRDIV-NEXT: br i1 [[OR_COND_I19_I_I]], label [[IF_THEN_I24_I_I:%.*]], label [[CLEANUP_I20_I_I]]
// NCRDIV: if.then.i24.i.i:
// NCRDIV-NEXT: [[MUL_I25_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// NCRDIV-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// NCRDIV-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP9]] to i64
// NCRDIV-NEXT: [[ADD_I27_I_I:%.*]] = add i64 [[MUL_I25_I_I]], -48
// NCRDIV-NEXT: [[SUB_I28_I_I:%.*]] = add i64 [[ADD_I27_I_I]], [[CONV5_I26_I_I]]
// NCRDIV-NEXT: [[INCDEC_PTR_I29_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I]], i64 1
@@ -4053,8 +4128,8 @@ extern "C" __device__ double test_modf(double x, double* y) {
// NCRDIV-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32
// NCRDIV-NEXT: [[BF_VALUE_I:%.*]] = and i32 [[CONV_I]], 4194303
// NCRDIV-NEXT: [[BF_SET9_I:%.*]] = or disjoint i32 [[BF_VALUE_I]], 2143289344
-// NCRDIV-NEXT: [[TMP10:%.*]] = bitcast i32 [[BF_SET9_I]] to float
-// NCRDIV-NEXT: ret float [[TMP10]]
+// NCRDIV-NEXT: [[TMP11:%.*]] = bitcast i32 [[BF_SET9_I]] to float
+// NCRDIV-NEXT: ret float [[TMP11]]
//
// AMDGCNSPIRV-LABEL: @test_nanf(
// AMDGCNSPIRV-NEXT: entry:
@@ -4071,47 +4146,57 @@ extern "C" __device__ double test_modf(double x, double* y) {
// AMDGCNSPIRV: while.cond.i28.i.i.preheader:
// AMDGCNSPIRV-NEXT: br label [[WHILE_COND_I28_I_I:%.*]]
// AMDGCNSPIRV: while.cond.i28.i.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I29_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I34_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I28_I_I_PREHEADER]] ]
-// AMDGCNSPIRV-NEXT: [[__R_0_I30_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[WHILE_COND_I28_I_I_PREHEADER]] ]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I29_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I34_I_I:%.*]], [[SWITCH_EARLY_TEST:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I28_I_I_PREHEADER]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[WHILE_BODY_I32_I_I:%.*]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ]
+// AMDGCNSPIRV-NEXT: [[__R_0_I30_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[SWITCH_EARLY_TEST]] ], [ 0, [[WHILE_COND_I28_I_I_PREHEADER]] ], [ [[__R_2_I_I_I]], [[WHILE_BODY_I32_I_I]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ]
// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I]], align 1, !tbaa [[TBAA5]]
// AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label [[_ZL4NANFPKC_EXIT:%.*]], label [[WHILE_BODY_I32_I_I:%.*]]
+// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label [[_ZL4NANFPKC_EXIT:%.*]], label [[WHILE_BODY_I32_I_I]]
// AMDGCNSPIRV: while.body.i32.i.i:
// AMDGCNSPIRV-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
// AMDGCNSPIRV-NEXT: [[OR_COND_I33_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I33_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// AMDGCNSPIRV: if.else.i.i.i:
// AMDGCNSPIRV-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
// AMDGCNSPIRV-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// AMDGCNSPIRV: if.else17.i.i.i:
// AMDGCNSPIRV-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
// AMDGCNSPIRV-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I_I_I]]
-// AMDGCNSPIRV: if.end31.i.i.i:
-// AMDGCNSPIRV-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I32_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
+// AMDGCNSPIRV-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// AMDGCNSPIRV-NEXT: [[TMP7:%.*]] = select i1 [[OR_COND_I33_I_I]], i64 -48, i64 [[TMP6]]
+// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = or i1 [[OR_COND33_I_I_I]], [[OR_COND34_I_I_I]]
+// AMDGCNSPIRV-NEXT: [[OR_COND13:%.*]] = or i1 [[TMP8]], [[OR_COND_I33_I_I]]
// AMDGCNSPIRV-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I30_I_I]], 4
// AMDGCNSPIRV-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// AMDGCNSPIRV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// AMDGCNSPIRV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP7]]
// AMDGCNSPIRV-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
-// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I37_I_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I]], i64 1
-// AMDGCNSPIRV-NEXT: br label [[CLEANUP_I_I_I]]
-// AMDGCNSPIRV: cleanup.i.i.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I_I]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I37_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I29_I_I]], [[IF_ELSE17_I_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I30_I_I]], [[IF_ELSE17_I_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
-// AMDGCNSPIRV-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I28_I_I]], label [[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP12]]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I_I_IDX:%.*]] = zext i1 [[OR_COND13]] to i64
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I]], i64 [[__TAGP_ADDR_1_I34_I_I_IDX]]
+// AMDGCNSPIRV-NEXT: [[__R_2_I_I_I]] = select i1 [[OR_COND13]], i64 [[ADD28_I_I_I]], i64 [[__R_0_I30_I_I]]
+// AMDGCNSPIRV-NEXT: [[TMP9:%.*]] = freeze i1 [[OR_COND_I33_I_I]]
+// AMDGCNSPIRV-NEXT: br i1 [[TMP9]], label [[WHILE_COND_I28_I_I]], label [[SWITCH_EARLY_TEST]]
+// AMDGCNSPIRV: switch.early.test:
+// AMDGCNSPIRV-NEXT: switch i8 [[TMP2]], label [[_ZL4NANFPKC_EXIT]] [
+// AMDGCNSPIRV-NEXT: i8 102, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 101, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 100, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 99, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 98, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 97, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 70, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 69, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 68, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 67, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 66, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 65, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: ]
// AMDGCNSPIRV: while.cond.i.i.i:
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[WHILE_BODY_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// AMDGCNSPIRV-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[WHILE_BODY_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA5]]
-// AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA5]]
+// AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP10]], 0
// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I_I_I]]
// AMDGCNSPIRV: while.body.i.i.i:
-// AMDGCNSPIRV-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// AMDGCNSPIRV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// AMDGCNSPIRV-NEXT: [[TMP11:%.*]] = and i8 [[TMP10]], -8
+// AMDGCNSPIRV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP11]], 48
// AMDGCNSPIRV-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// AMDGCNSPIRV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// AMDGCNSPIRV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP10]] to i64
// AMDGCNSPIRV-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// AMDGCNSPIRV-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_I_IDX:%.*]] = zext i1 [[OR_COND_I_I_I]] to i64
@@ -4121,14 +4206,14 @@ extern "C" __device__ double test_modf(double x, double* y) {
// AMDGCNSPIRV: while.cond.i14.i.i:
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I25_I_I:%.*]], [[WHILE_BODY_I18_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// AMDGCNSPIRV-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I26_I_I:%.*]], [[WHILE_BODY_I18_I_I]] ], [ 0, [[ENTRY]] ]
-// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA5]]
-// AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// AMDGCNSPIRV-NEXT: [[TMP12:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA5]]
+// AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP12]], 0
// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL4NANFPKC_EXIT]], label [[WHILE_BODY_I18_I_I]]
// AMDGCNSPIRV: while.body.i18.i.i:
-// AMDGCNSPIRV-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// AMDGCNSPIRV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// AMDGCNSPIRV-NEXT: [[TMP13:%.*]] = add i8 [[TMP12]], -48
+// AMDGCNSPIRV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP13]], 10
// AMDGCNSPIRV-NEXT: [[MUL_I20_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// AMDGCNSPIRV-NEXT: [[CONV5_I21_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// AMDGCNSPIRV-NEXT: [[CONV5_I21_I_I:%.*]] = zext nneg i8 [[TMP12]] to i64
// AMDGCNSPIRV-NEXT: [[ADD_I22_I_I:%.*]] = add i64 [[MUL_I20_I_I]], -48
// AMDGCNSPIRV-NEXT: [[SUB_I23_I_I:%.*]] = add i64 [[ADD_I22_I_I]], [[CONV5_I21_I_I]]
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_I_IDX:%.*]] = zext i1 [[OR_COND_I19_I_I]] to i64
@@ -4136,12 +4221,12 @@ extern "C" __device__ double test_modf(double x, double* y) {
// AMDGCNSPIRV-NEXT: [[__R_1_I26_I_I]] = select i1 [[OR_COND_I19_I_I]], i64 [[SUB_I23_I_I]], i64 [[__R_0_I16_I_I]]
// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I_I]], label [[WHILE_COND_I14_I_I]], label [[_ZL4NANFPKC_EXIT]], !llvm.loop [[LOOP11]]
// AMDGCNSPIRV: _ZL4nanfPKc.exit:
-// AMDGCNSPIRV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, [[WHILE_BODY_I_I_I]] ], [ [[__R_0_I_I_I]], [[WHILE_COND_I_I_I]] ], [ 0, [[CLEANUP_I_I_I]] ], [ [[__R_0_I30_I_I]], [[WHILE_COND_I28_I_I]] ], [ 0, [[WHILE_BODY_I18_I_I]] ], [ [[__R_0_I16_I_I]], [[WHILE_COND_I14_I_I]] ]
+// AMDGCNSPIRV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, [[WHILE_BODY_I_I_I]] ], [ [[__R_0_I_I_I]], [[WHILE_COND_I_I_I]] ], [ 0, [[SWITCH_EARLY_TEST]] ], [ [[__R_0_I30_I_I]], [[WHILE_COND_I28_I_I]] ], [ 0, [[WHILE_BODY_I18_I_I]] ], [ [[__R_0_I16_I_I]], [[WHILE_COND_I14_I_I]] ]
// AMDGCNSPIRV-NEXT: [[CONV_I:%.*]] = trunc i64 [[RETVAL_0_I_I]] to i32
// AMDGCNSPIRV-NEXT: [[BF_VALUE_I:%.*]] = and i32 [[CONV_I]], 4194303
// AMDGCNSPIRV-NEXT: [[BF_SET9_I:%.*]] = or disjoint i32 [[BF_VALUE_I]], 2143289344
-// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = bitcast i32 [[BF_SET9_I]] to float
-// AMDGCNSPIRV-NEXT: ret float [[TMP10]]
+// AMDGCNSPIRV-NEXT: [[TMP14:%.*]] = bitcast i32 [[BF_SET9_I]] to float
+// AMDGCNSPIRV-NEXT: ret float [[TMP14]]
//
extern "C" __device__ float test_nanf(const char *tag) {
return nanf(tag);
@@ -4165,46 +4250,57 @@ extern "C" __device__ float test_nanf(const char *tag) {
// DEFAULT-NEXT: [[__TAGP_ADDR_0_I31_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I37_I_I:%.*]], [[CLEANUP_I36_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I30_I_I_PREHEADER]] ]
// DEFAULT-NEXT: [[__R_0_I32_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I36_I_I]] ], [ 0, [[WHILE_COND_I30_I_I_PREHEADER]] ]
// DEFAULT-NEXT: [[TMP2:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I31_I_I]], align 1, !tbaa [[TBAA4]]
-// DEFAULT-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
+// DEFAULT-NEXT: [[DOTFR14:%.*]] = freeze i8 [[TMP2]]
+// DEFAULT-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[DOTFR14]], 0
// DEFAULT-NEXT: br i1 [[CMP_NOT_I33_I_I]], label [[_ZL3NANPKC_EXIT:%.*]], label [[WHILE_BODY_I34_I_I:%.*]]
// DEFAULT: while.body.i34.i.i:
-// DEFAULT-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
+// DEFAULT-NEXT: [[TMP3:%.*]] = add i8 [[DOTFR14]], -48
// DEFAULT-NEXT: [[OR_COND_I35_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// DEFAULT-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// DEFAULT: if.else.i.i.i:
-// DEFAULT-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
+// DEFAULT-NEXT: [[TMP4:%.*]] = add i8 [[DOTFR14]], -97
// DEFAULT-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// DEFAULT-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// DEFAULT: if.else17.i.i.i:
-// DEFAULT-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
-// DEFAULT-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// DEFAULT-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I36_I_I]]
+// DEFAULT-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// DEFAULT-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND_I35_I_I]], i64 -48, i64 [[TMP5]]
+// DEFAULT-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// DEFAULT: switch.early.test:
+// DEFAULT-NEXT: switch i8 [[DOTFR14]], label [[CLEANUP_I36_I_I]] [
+// DEFAULT-NEXT: i8 102, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 101, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 100, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 99, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 98, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 97, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 70, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 69, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 68, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 67, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 66, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: i8 65, label [[IF_END31_I_I_I]]
+// DEFAULT-NEXT: ]
// DEFAULT: if.end31.i.i.i:
-// DEFAULT-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I34_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
// DEFAULT-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I32_I_I]], 4
-// DEFAULT-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// DEFAULT-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// DEFAULT-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[DOTFR14]] to i64
+// DEFAULT-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP6]]
// DEFAULT-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
// DEFAULT-NEXT: [[INCDEC_PTR_I40_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I31_I_I]], i64 1
// DEFAULT-NEXT: br label [[CLEANUP_I36_I_I]]
// DEFAULT: cleanup.i36.i.i:
-// DEFAULT-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[IF_ELSE17_I_I_I]] ]
-// DEFAULT-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[IF_ELSE17_I_I_I]] ]
-// DEFAULT-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
+// DEFAULT-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[SWITCH_EARLY_TEST]] ]
+// DEFAULT-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[SWITCH_EARLY_TEST]] ]
+// DEFAULT-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// DEFAULT-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I30_I_I]], label [[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP11]]
// DEFAULT: while.cond.i.i.i:
// DEFAULT-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// DEFAULT-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// DEFAULT-NEXT: [[TMP6:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
-// DEFAULT-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// DEFAULT-NEXT: [[TMP7:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
+// DEFAULT-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 0
// DEFAULT-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I_I_I:%.*]]
// DEFAULT: while.body.i.i.i:
-// DEFAULT-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// DEFAULT-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// DEFAULT-NEXT: [[TMP8:%.*]] = and i8 [[TMP7]], -8
+// DEFAULT-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 48
// DEFAULT-NEXT: br i1 [[OR_COND_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[CLEANUP_I_I_I]]
// DEFAULT: if.then.i.i.i:
// DEFAULT-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// DEFAULT-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// DEFAULT-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP7]] to i64
// DEFAULT-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// DEFAULT-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// DEFAULT-NEXT: [[INCDEC_PTR_I_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I]], i64 1
@@ -4216,16 +4312,16 @@ extern "C" __device__ float test_nanf(const char *tag) {
// DEFAULT: while.cond.i14.i.i:
// DEFAULT-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I21_I_I:%.*]], [[CLEANUP_I20_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// DEFAULT-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I22_I_I:%.*]], [[CLEANUP_I20_I_I]] ], [ 0, [[ENTRY]] ]
-// DEFAULT-NEXT: [[TMP8:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
-// DEFAULT-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// DEFAULT-NEXT: [[TMP9:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
+// DEFAULT-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP9]], 0
// DEFAULT-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I18_I_I:%.*]]
// DEFAULT: while.body.i18.i.i:
-// DEFAULT-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// DEFAULT-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// DEFAULT-NEXT: [[TMP10:%.*]] = add i8 [[TMP9]], -48
+// DEFAULT-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP10]], 10
// DEFAULT-NEXT: br i1 [[OR_COND_I19_I_I]], label [[IF_THEN_I24_I_I:%.*]], label [[CLEANUP_I20_I_I]]
// DEFAULT: if.then.i24.i.i:
// DEFAULT-NEXT: [[MUL_I25_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// DEFAULT-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// DEFAULT-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP9]] to i64
// DEFAULT-NEXT: [[ADD_I27_I_I:%.*]] = add i64 [[MUL_I25_I_I]], -48
// DEFAULT-NEXT: [[SUB_I28_I_I:%.*]] = add i64 [[ADD_I27_I_I]], [[CONV5_I26_I_I]]
// DEFAULT-NEXT: [[INCDEC_PTR_I29_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I]], i64 1
@@ -4238,8 +4334,8 @@ extern "C" __device__ float test_nanf(const char *tag) {
// DEFAULT-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, [[CLEANUP_I_I_I]] ], [ [[__R_0_I_I_I]], [[WHILE_COND_I_I_I]] ], [ 0, [[CLEANUP_I36_I_I]] ], [ [[__R_0_I32_I_I]], [[WHILE_COND_I30_I_I]] ], [ 0, [[CLEANUP_I20_I_I]] ], [ [[__R_0_I16_I_I]], [[WHILE_COND_I14_I_I]] ]
// DEFAULT-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247
// DEFAULT-NEXT: [[BF_SET9_I:%.*]] = or disjoint i64 [[BF_VALUE_I]], 9221120237041090560
-// DEFAULT-NEXT: [[TMP10:%.*]] = bitcast i64 [[BF_SET9_I]] to double
-// DEFAULT-NEXT: ret double [[TMP10]]
+// DEFAULT-NEXT: [[TMP11:%.*]] = bitcast i64 [[BF_SET9_I]] to double
+// DEFAULT-NEXT: ret double [[TMP11]]
//
// FINITEONLY-LABEL: @test_nan(
// FINITEONLY-NEXT: entry:
@@ -4263,46 +4359,57 @@ extern "C" __device__ float test_nanf(const char *tag) {
// APPROX-NEXT: [[__TAGP_ADDR_0_I31_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I37_I_I:%.*]], [[CLEANUP_I36_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I30_I_I_PREHEADER]] ]
// APPROX-NEXT: [[__R_0_I32_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I36_I_I]] ], [ 0, [[WHILE_COND_I30_I_I_PREHEADER]] ]
// APPROX-NEXT: [[TMP2:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I31_I_I]], align 1, !tbaa [[TBAA4]]
-// APPROX-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
+// APPROX-NEXT: [[DOTFR14:%.*]] = freeze i8 [[TMP2]]
+// APPROX-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[DOTFR14]], 0
// APPROX-NEXT: br i1 [[CMP_NOT_I33_I_I]], label [[_ZL3NANPKC_EXIT:%.*]], label [[WHILE_BODY_I34_I_I:%.*]]
// APPROX: while.body.i34.i.i:
-// APPROX-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
+// APPROX-NEXT: [[TMP3:%.*]] = add i8 [[DOTFR14]], -48
// APPROX-NEXT: [[OR_COND_I35_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// APPROX-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// APPROX: if.else.i.i.i:
-// APPROX-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
+// APPROX-NEXT: [[TMP4:%.*]] = add i8 [[DOTFR14]], -97
// APPROX-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// APPROX-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// APPROX: if.else17.i.i.i:
-// APPROX-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
-// APPROX-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// APPROX-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I36_I_I]]
+// APPROX-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// APPROX-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND_I35_I_I]], i64 -48, i64 [[TMP5]]
+// APPROX-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// APPROX: switch.early.test:
+// APPROX-NEXT: switch i8 [[DOTFR14]], label [[CLEANUP_I36_I_I]] [
+// APPROX-NEXT: i8 102, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 101, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 100, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 99, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 98, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 97, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 70, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 69, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 68, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 67, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 66, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: i8 65, label [[IF_END31_I_I_I]]
+// APPROX-NEXT: ]
// APPROX: if.end31.i.i.i:
-// APPROX-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I34_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
// APPROX-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I32_I_I]], 4
-// APPROX-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// APPROX-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// APPROX-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[DOTFR14]] to i64
+// APPROX-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP6]]
// APPROX-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
// APPROX-NEXT: [[INCDEC_PTR_I40_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I31_I_I]], i64 1
// APPROX-NEXT: br label [[CLEANUP_I36_I_I]]
// APPROX: cleanup.i36.i.i:
-// APPROX-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[IF_ELSE17_I_I_I]] ]
-// APPROX-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[IF_ELSE17_I_I_I]] ]
-// APPROX-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
+// APPROX-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[SWITCH_EARLY_TEST]] ]
+// APPROX-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[SWITCH_EARLY_TEST]] ]
+// APPROX-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// APPROX-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I30_I_I]], label [[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP11]]
// APPROX: while.cond.i.i.i:
// APPROX-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// APPROX-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// APPROX-NEXT: [[TMP6:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
-// APPROX-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// APPROX-NEXT: [[TMP7:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
+// APPROX-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 0
// APPROX-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I_I_I:%.*]]
// APPROX: while.body.i.i.i:
-// APPROX-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// APPROX-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// APPROX-NEXT: [[TMP8:%.*]] = and i8 [[TMP7]], -8
+// APPROX-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 48
// APPROX-NEXT: br i1 [[OR_COND_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[CLEANUP_I_I_I]]
// APPROX: if.then.i.i.i:
// APPROX-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// APPROX-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// APPROX-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP7]] to i64
// APPROX-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// APPROX-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// APPROX-NEXT: [[INCDEC_PTR_I_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I]], i64 1
@@ -4314,16 +4421,16 @@ extern "C" __device__ float test_nanf(const char *tag) {
// APPROX: while.cond.i14.i.i:
// APPROX-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I21_I_I:%.*]], [[CLEANUP_I20_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// APPROX-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I22_I_I:%.*]], [[CLEANUP_I20_I_I]] ], [ 0, [[ENTRY]] ]
-// APPROX-NEXT: [[TMP8:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
-// APPROX-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// APPROX-NEXT: [[TMP9:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
+// APPROX-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP9]], 0
// APPROX-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I18_I_I:%.*]]
// APPROX: while.body.i18.i.i:
-// APPROX-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// APPROX-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// APPROX-NEXT: [[TMP10:%.*]] = add i8 [[TMP9]], -48
+// APPROX-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP10]], 10
// APPROX-NEXT: br i1 [[OR_COND_I19_I_I]], label [[IF_THEN_I24_I_I:%.*]], label [[CLEANUP_I20_I_I]]
// APPROX: if.then.i24.i.i:
// APPROX-NEXT: [[MUL_I25_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// APPROX-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// APPROX-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP9]] to i64
// APPROX-NEXT: [[ADD_I27_I_I:%.*]] = add i64 [[MUL_I25_I_I]], -48
// APPROX-NEXT: [[SUB_I28_I_I:%.*]] = add i64 [[ADD_I27_I_I]], [[CONV5_I26_I_I]]
// APPROX-NEXT: [[INCDEC_PTR_I29_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I]], i64 1
@@ -4336,8 +4443,8 @@ extern "C" __device__ float test_nanf(const char *tag) {
// APPROX-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, [[CLEANUP_I_I_I]] ], [ [[__R_0_I_I_I]], [[WHILE_COND_I_I_I]] ], [ 0, [[CLEANUP_I36_I_I]] ], [ [[__R_0_I32_I_I]], [[WHILE_COND_I30_I_I]] ], [ 0, [[CLEANUP_I20_I_I]] ], [ [[__R_0_I16_I_I]], [[WHILE_COND_I14_I_I]] ]
// APPROX-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247
// APPROX-NEXT: [[BF_SET9_I:%.*]] = or disjoint i64 [[BF_VALUE_I]], 9221120237041090560
-// APPROX-NEXT: [[TMP10:%.*]] = bitcast i64 [[BF_SET9_I]] to double
-// APPROX-NEXT: ret double [[TMP10]]
+// APPROX-NEXT: [[TMP11:%.*]] = bitcast i64 [[BF_SET9_I]] to double
+// APPROX-NEXT: ret double [[TMP11]]
//
// NCRDIV-LABEL: @test_nan(
// NCRDIV-NEXT: entry:
@@ -4357,46 +4464,57 @@ extern "C" __device__ float test_nanf(const char *tag) {
// NCRDIV-NEXT: [[__TAGP_ADDR_0_I31_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I37_I_I:%.*]], [[CLEANUP_I36_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I30_I_I_PREHEADER]] ]
// NCRDIV-NEXT: [[__R_0_I32_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I36_I_I]] ], [ 0, [[WHILE_COND_I30_I_I_PREHEADER]] ]
// NCRDIV-NEXT: [[TMP2:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I31_I_I]], align 1, !tbaa [[TBAA4]]
-// NCRDIV-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
+// NCRDIV-NEXT: [[DOTFR14:%.*]] = freeze i8 [[TMP2]]
+// NCRDIV-NEXT: [[CMP_NOT_I33_I_I:%.*]] = icmp eq i8 [[DOTFR14]], 0
// NCRDIV-NEXT: br i1 [[CMP_NOT_I33_I_I]], label [[_ZL3NANPKC_EXIT:%.*]], label [[WHILE_BODY_I34_I_I:%.*]]
// NCRDIV: while.body.i34.i.i:
-// NCRDIV-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
+// NCRDIV-NEXT: [[TMP3:%.*]] = add i8 [[DOTFR14]], -48
// NCRDIV-NEXT: [[OR_COND_I35_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// NCRDIV-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// NCRDIV: if.else.i.i.i:
-// NCRDIV-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
+// NCRDIV-NEXT: [[TMP4:%.*]] = add i8 [[DOTFR14]], -97
// NCRDIV-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// NCRDIV-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// NCRDIV: if.else17.i.i.i:
-// NCRDIV-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
-// NCRDIV-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// NCRDIV-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I36_I_I]]
+// NCRDIV-NEXT: [[TMP5:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// NCRDIV-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND_I35_I_I]], i64 -48, i64 [[TMP5]]
+// NCRDIV-NEXT: br i1 [[OR_COND_I35_I_I]], label [[IF_END31_I_I_I:%.*]], label [[SWITCH_EARLY_TEST:%.*]]
+// NCRDIV: switch.early.test:
+// NCRDIV-NEXT: switch i8 [[DOTFR14]], label [[CLEANUP_I36_I_I]] [
+// NCRDIV-NEXT: i8 102, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 101, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 100, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 99, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 98, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 97, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 70, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 69, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 68, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 67, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 66, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: i8 65, label [[IF_END31_I_I_I]]
+// NCRDIV-NEXT: ]
// NCRDIV: if.end31.i.i.i:
-// NCRDIV-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I34_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
// NCRDIV-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I32_I_I]], 4
-// NCRDIV-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// NCRDIV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// NCRDIV-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[DOTFR14]] to i64
+// NCRDIV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP6]]
// NCRDIV-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
// NCRDIV-NEXT: [[INCDEC_PTR_I40_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I31_I_I]], i64 1
// NCRDIV-NEXT: br label [[CLEANUP_I36_I_I]]
// NCRDIV: cleanup.i36.i.i:
-// NCRDIV-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[IF_ELSE17_I_I_I]] ]
-// NCRDIV-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[IF_ELSE17_I_I_I]] ]
-// NCRDIV-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
+// NCRDIV-NEXT: [[__TAGP_ADDR_1_I37_I_I]] = phi ptr [ [[INCDEC_PTR_I40_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I31_I_I]], [[SWITCH_EARLY_TEST]] ]
+// NCRDIV-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I32_I_I]], [[SWITCH_EARLY_TEST]] ]
+// NCRDIV-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[SWITCH_EARLY_TEST]] ]
// NCRDIV-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I30_I_I]], label [[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP11]]
// NCRDIV: while.cond.i.i.i:
// NCRDIV-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// NCRDIV-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// NCRDIV-NEXT: [[TMP6:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
-// NCRDIV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// NCRDIV-NEXT: [[TMP7:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA4]]
+// NCRDIV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 0
// NCRDIV-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I_I_I:%.*]]
// NCRDIV: while.body.i.i.i:
-// NCRDIV-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// NCRDIV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// NCRDIV-NEXT: [[TMP8:%.*]] = and i8 [[TMP7]], -8
+// NCRDIV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP8]], 48
// NCRDIV-NEXT: br i1 [[OR_COND_I_I_I]], label [[IF_THEN_I_I_I:%.*]], label [[CLEANUP_I_I_I]]
// NCRDIV: if.then.i.i.i:
// NCRDIV-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// NCRDIV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// NCRDIV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP7]] to i64
// NCRDIV-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// NCRDIV-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// NCRDIV-NEXT: [[INCDEC_PTR_I_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I_I_I]], i64 1
@@ -4408,16 +4526,16 @@ extern "C" __device__ float test_nanf(const char *tag) {
// NCRDIV: while.cond.i14.i.i:
// NCRDIV-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr [ [[__TAGP_ADDR_1_I21_I_I:%.*]], [[CLEANUP_I20_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// NCRDIV-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I22_I_I:%.*]], [[CLEANUP_I20_I_I]] ], [ 0, [[ENTRY]] ]
-// NCRDIV-NEXT: [[TMP8:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
-// NCRDIV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// NCRDIV-NEXT: [[TMP9:%.*]] = load i8, ptr [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA4]]
+// NCRDIV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP9]], 0
// NCRDIV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I18_I_I:%.*]]
// NCRDIV: while.body.i18.i.i:
-// NCRDIV-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// NCRDIV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// NCRDIV-NEXT: [[TMP10:%.*]] = add i8 [[TMP9]], -48
+// NCRDIV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP10]], 10
// NCRDIV-NEXT: br i1 [[OR_COND_I19_I_I]], label [[IF_THEN_I24_I_I:%.*]], label [[CLEANUP_I20_I_I]]
// NCRDIV: if.then.i24.i.i:
// NCRDIV-NEXT: [[MUL_I25_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// NCRDIV-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// NCRDIV-NEXT: [[CONV5_I26_I_I:%.*]] = zext nneg i8 [[TMP9]] to i64
// NCRDIV-NEXT: [[ADD_I27_I_I:%.*]] = add i64 [[MUL_I25_I_I]], -48
// NCRDIV-NEXT: [[SUB_I28_I_I:%.*]] = add i64 [[ADD_I27_I_I]], [[CONV5_I26_I_I]]
// NCRDIV-NEXT: [[INCDEC_PTR_I29_I_I:%.*]] = getelementptr inbounds nuw i8, ptr [[__TAGP_ADDR_0_I15_I_I]], i64 1
@@ -4430,8 +4548,8 @@ extern "C" __device__ float test_nanf(const char *tag) {
// NCRDIV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, [[CLEANUP_I_I_I]] ], [ [[__R_0_I_I_I]], [[WHILE_COND_I_I_I]] ], [ 0, [[CLEANUP_I36_I_I]] ], [ [[__R_0_I32_I_I]], [[WHILE_COND_I30_I_I]] ], [ 0, [[CLEANUP_I20_I_I]] ], [ [[__R_0_I16_I_I]], [[WHILE_COND_I14_I_I]] ]
// NCRDIV-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247
// NCRDIV-NEXT: [[BF_SET9_I:%.*]] = or disjoint i64 [[BF_VALUE_I]], 9221120237041090560
-// NCRDIV-NEXT: [[TMP10:%.*]] = bitcast i64 [[BF_SET9_I]] to double
-// NCRDIV-NEXT: ret double [[TMP10]]
+// NCRDIV-NEXT: [[TMP11:%.*]] = bitcast i64 [[BF_SET9_I]] to double
+// NCRDIV-NEXT: ret double [[TMP11]]
//
// AMDGCNSPIRV-LABEL: @test_nan(
// AMDGCNSPIRV-NEXT: entry:
@@ -4448,47 +4566,57 @@ extern "C" __device__ float test_nanf(const char *tag) {
// AMDGCNSPIRV: while.cond.i28.i.i.preheader:
// AMDGCNSPIRV-NEXT: br label [[WHILE_COND_I28_I_I:%.*]]
// AMDGCNSPIRV: while.cond.i28.i.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I29_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I34_I_I:%.*]], [[CLEANUP_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I28_I_I_PREHEADER]] ]
-// AMDGCNSPIRV-NEXT: [[__R_0_I30_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[CLEANUP_I_I_I]] ], [ 0, [[WHILE_COND_I28_I_I_PREHEADER]] ]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I29_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I34_I_I:%.*]], [[SWITCH_EARLY_TEST:%.*]] ], [ [[INCDEC_PTR_I_I]], [[WHILE_COND_I28_I_I_PREHEADER]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[WHILE_BODY_I32_I_I:%.*]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__TAGP_ADDR_1_I34_I_I]], [[SWITCH_EARLY_TEST]] ]
+// AMDGCNSPIRV-NEXT: [[__R_0_I30_I_I:%.*]] = phi i64 [ [[__R_2_I_I_I:%.*]], [[SWITCH_EARLY_TEST]] ], [ 0, [[WHILE_COND_I28_I_I_PREHEADER]] ], [ [[__R_2_I_I_I]], [[WHILE_BODY_I32_I_I]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ], [ [[__R_2_I_I_I]], [[SWITCH_EARLY_TEST]] ]
// AMDGCNSPIRV-NEXT: [[TMP2:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I]], align 1, !tbaa [[TBAA5]]
// AMDGCNSPIRV-NEXT: [[CMP_NOT_I31_I_I:%.*]] = icmp eq i8 [[TMP2]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label [[_ZL3NANPKC_EXIT:%.*]], label [[WHILE_BODY_I32_I_I:%.*]]
+// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I31_I_I]], label [[_ZL3NANPKC_EXIT:%.*]], label [[WHILE_BODY_I32_I_I]]
// AMDGCNSPIRV: while.body.i32.i.i:
// AMDGCNSPIRV-NEXT: [[TMP3:%.*]] = add i8 [[TMP2]], -48
// AMDGCNSPIRV-NEXT: [[OR_COND_I33_I_I:%.*]] = icmp ult i8 [[TMP3]], 10
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I33_I_I]], label [[IF_END31_I_I_I:%.*]], label [[IF_ELSE_I_I_I:%.*]]
-// AMDGCNSPIRV: if.else.i.i.i:
// AMDGCNSPIRV-NEXT: [[TMP4:%.*]] = add i8 [[TMP2]], -97
// AMDGCNSPIRV-NEXT: [[OR_COND33_I_I_I:%.*]] = icmp ult i8 [[TMP4]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND33_I_I_I]], label [[IF_END31_I_I_I]], label [[IF_ELSE17_I_I_I:%.*]]
-// AMDGCNSPIRV: if.else17.i.i.i:
// AMDGCNSPIRV-NEXT: [[TMP5:%.*]] = add i8 [[TMP2]], -65
// AMDGCNSPIRV-NEXT: [[OR_COND34_I_I_I:%.*]] = icmp ult i8 [[TMP5]], 6
-// AMDGCNSPIRV-NEXT: br i1 [[OR_COND34_I_I_I]], label [[IF_END31_I_I_I]], label [[CLEANUP_I_I_I]]
-// AMDGCNSPIRV: if.end31.i.i.i:
-// AMDGCNSPIRV-NEXT: [[DOTSINK:%.*]] = phi i64 [ -48, [[WHILE_BODY_I32_I_I]] ], [ -87, [[IF_ELSE_I_I_I]] ], [ -55, [[IF_ELSE17_I_I_I]] ]
+// AMDGCNSPIRV-NEXT: [[TMP6:%.*]] = select i1 [[OR_COND33_I_I_I]], i64 -87, i64 -55
+// AMDGCNSPIRV-NEXT: [[TMP7:%.*]] = select i1 [[OR_COND_I33_I_I]], i64 -48, i64 [[TMP6]]
+// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = or i1 [[OR_COND33_I_I_I]], [[OR_COND34_I_I_I]]
+// AMDGCNSPIRV-NEXT: [[OR_COND13:%.*]] = or i1 [[TMP8]], [[OR_COND_I33_I_I]]
// AMDGCNSPIRV-NEXT: [[MUL24_I_I_I:%.*]] = shl i64 [[__R_0_I30_I_I]], 4
// AMDGCNSPIRV-NEXT: [[CONV25_I_I_I:%.*]] = zext nneg i8 [[TMP2]] to i64
-// AMDGCNSPIRV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[DOTSINK]]
+// AMDGCNSPIRV-NEXT: [[ADD26_I_I_I:%.*]] = add i64 [[MUL24_I_I_I]], [[TMP7]]
// AMDGCNSPIRV-NEXT: [[ADD28_I_I_I:%.*]] = add i64 [[ADD26_I_I_I]], [[CONV25_I_I_I]]
-// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I37_I_I:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I]], i64 1
-// AMDGCNSPIRV-NEXT: br label [[CLEANUP_I_I_I]]
-// AMDGCNSPIRV: cleanup.i.i.i:
-// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I_I]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I37_I_I]], [[IF_END31_I_I_I]] ], [ [[__TAGP_ADDR_0_I29_I_I]], [[IF_ELSE17_I_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[__R_2_I_I_I]] = phi i64 [ [[ADD28_I_I_I]], [[IF_END31_I_I_I]] ], [ [[__R_0_I30_I_I]], [[IF_ELSE17_I_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[COND_I_I_I:%.*]] = phi i1 [ true, [[IF_END31_I_I_I]] ], [ false, [[IF_ELSE17_I_I_I]] ]
-// AMDGCNSPIRV-NEXT: br i1 [[COND_I_I_I]], label [[WHILE_COND_I28_I_I]], label [[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP12]]
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I_I_IDX:%.*]] = zext i1 [[OR_COND13]] to i64
+// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I34_I_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__TAGP_ADDR_0_I29_I_I]], i64 [[__TAGP_ADDR_1_I34_I_I_IDX]]
+// AMDGCNSPIRV-NEXT: [[__R_2_I_I_I]] = select i1 [[OR_COND13]], i64 [[ADD28_I_I_I]], i64 [[__R_0_I30_I_I]]
+// AMDGCNSPIRV-NEXT: [[TMP9:%.*]] = freeze i1 [[OR_COND_I33_I_I]]
+// AMDGCNSPIRV-NEXT: br i1 [[TMP9]], label [[WHILE_COND_I28_I_I]], label [[SWITCH_EARLY_TEST]]
+// AMDGCNSPIRV: switch.early.test:
+// AMDGCNSPIRV-NEXT: switch i8 [[TMP2]], label [[_ZL3NANPKC_EXIT]] [
+// AMDGCNSPIRV-NEXT: i8 102, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 101, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 100, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 99, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 98, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 97, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 70, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 69, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 68, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 67, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 66, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: i8 65, label [[WHILE_COND_I28_I_I]]
+// AMDGCNSPIRV-NEXT: ]
// AMDGCNSPIRV: while.cond.i.i.i:
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I_I_I:%.*]], [[WHILE_BODY_I_I_I:%.*]] ], [ [[INCDEC_PTR_I_I]], [[IF_THEN_I_I]] ]
// AMDGCNSPIRV-NEXT: [[__R_0_I_I_I:%.*]] = phi i64 [ [[__R_1_I_I_I:%.*]], [[WHILE_BODY_I_I_I]] ], [ 0, [[IF_THEN_I_I]] ]
-// AMDGCNSPIRV-NEXT: [[TMP6:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA5]]
-// AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP6]], 0
+// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I_I_I]], align 1, !tbaa [[TBAA5]]
+// AMDGCNSPIRV-NEXT: [[CMP_NOT_I_I_I:%.*]] = icmp eq i8 [[TMP10]], 0
// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I_I_I]]
// AMDGCNSPIRV: while.body.i.i.i:
-// AMDGCNSPIRV-NEXT: [[TMP7:%.*]] = and i8 [[TMP6]], -8
-// AMDGCNSPIRV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP7]], 48
+// AMDGCNSPIRV-NEXT: [[TMP11:%.*]] = and i8 [[TMP10]], -8
+// AMDGCNSPIRV-NEXT: [[OR_COND_I_I_I:%.*]] = icmp eq i8 [[TMP11]], 48
// AMDGCNSPIRV-NEXT: [[MUL_I_I_I:%.*]] = shl i64 [[__R_0_I_I_I]], 3
-// AMDGCNSPIRV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP6]] to i64
+// AMDGCNSPIRV-NEXT: [[CONV5_I_I_I:%.*]] = zext nneg i8 [[TMP10]] to i64
// AMDGCNSPIRV-NEXT: [[ADD_I_I_I:%.*]] = add i64 [[MUL_I_I_I]], -48
// AMDGCNSPIRV-NEXT: [[SUB_I_I_I:%.*]] = add i64 [[ADD_I_I_I]], [[CONV5_I_I_I]]
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I_I_I_IDX:%.*]] = zext i1 [[OR_COND_I_I_I]] to i64
@@ -4498,14 +4626,14 @@ extern "C" __device__ float test_nanf(const char *tag) {
// AMDGCNSPIRV: while.cond.i14.i.i:
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_0_I15_I_I:%.*]] = phi ptr addrspace(4) [ [[__TAGP_ADDR_1_I25_I_I:%.*]], [[WHILE_BODY_I18_I_I:%.*]] ], [ [[TAG]], [[ENTRY:%.*]] ]
// AMDGCNSPIRV-NEXT: [[__R_0_I16_I_I:%.*]] = phi i64 [ [[__R_1_I26_I_I:%.*]], [[WHILE_BODY_I18_I_I]] ], [ 0, [[ENTRY]] ]
-// AMDGCNSPIRV-NEXT: [[TMP8:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA5]]
-// AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP8]], 0
+// AMDGCNSPIRV-NEXT: [[TMP12:%.*]] = load i8, ptr addrspace(4) [[__TAGP_ADDR_0_I15_I_I]], align 1, !tbaa [[TBAA5]]
+// AMDGCNSPIRV-NEXT: [[CMP_NOT_I17_I_I:%.*]] = icmp eq i8 [[TMP12]], 0
// AMDGCNSPIRV-NEXT: br i1 [[CMP_NOT_I17_I_I]], label [[_ZL3NANPKC_EXIT]], label [[WHILE_BODY_I18_I_I]]
// AMDGCNSPIRV: while.body.i18.i.i:
-// AMDGCNSPIRV-NEXT: [[TMP9:%.*]] = add i8 [[TMP8]], -48
-// AMDGCNSPIRV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP9]], 10
+// AMDGCNSPIRV-NEXT: [[TMP13:%.*]] = add i8 [[TMP12]], -48
+// AMDGCNSPIRV-NEXT: [[OR_COND_I19_I_I:%.*]] = icmp ult i8 [[TMP13]], 10
// AMDGCNSPIRV-NEXT: [[MUL_I20_I_I:%.*]] = mul i64 [[__R_0_I16_I_I]], 10
-// AMDGCNSPIRV-NEXT: [[CONV5_I21_I_I:%.*]] = zext nneg i8 [[TMP8]] to i64
+// AMDGCNSPIRV-NEXT: [[CONV5_I21_I_I:%.*]] = zext nneg i8 [[TMP12]] to i64
// AMDGCNSPIRV-NEXT: [[ADD_I22_I_I:%.*]] = add i64 [[MUL_I20_I_I]], -48
// AMDGCNSPIRV-NEXT: [[SUB_I23_I_I:%.*]] = add i64 [[ADD_I22_I_I]], [[CONV5_I21_I_I]]
// AMDGCNSPIRV-NEXT: [[__TAGP_ADDR_1_I25_I_I_IDX:%.*]] = zext i1 [[OR_COND_I19_I_I]] to i64
@@ -4513,11 +4641,11 @@ extern "C" __device__ float test_nanf(const char *tag) {
// AMDGCNSPIRV-NEXT: [[__R_1_I26_I_I]] = select i1 [[OR_COND_I19_I_I]], i64 [[SUB_I23_I_I]], i64 [[__R_0_I16_I_I]]
// AMDGCNSPIRV-NEXT: br i1 [[OR_COND_I19_I_I]], label [[WHILE_COND_I14_I_I]], label [[_ZL3NANPKC_EXIT]], !llvm.loop [[LOOP11]]
// AMDGCNSPIRV: _ZL3nanPKc.exit:
-// AMDGCNSPIRV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, [[WHILE_BODY_I_I_I]] ], [ [[__R_0_I_I_I]], [[WHILE_COND_I_I_I]] ], [ 0, [[CLEANUP_I_I_I]] ], [ [[__R_0_I30_I_I]], [[WHILE_COND_I28_I_I]] ], [ 0, [[WHILE_BODY_I18_I_I]] ], [ [[__R_0_I16_I_I]], [[WHILE_COND_I14_I_I]] ]
+// AMDGCNSPIRV-NEXT: [[RETVAL_0_I_I:%.*]] = phi i64 [ 0, [[WHILE_BODY_I_I_I]] ], [ [[__R_0_I_I_I]], [[WHILE_COND_I_I_I]] ], [ 0, [[SWITCH_EARLY_TEST]] ], [ [[__R_0_I30_I_I]], [[WHILE_COND_I28_I_I]] ], [ 0, [[WHILE_BODY_I18_I_I]] ], [ [[__R_0_I16_I_I]], [[WHILE_COND_I14_I_I]] ]
// AMDGCNSPIRV-NEXT: [[BF_VALUE_I:%.*]] = and i64 [[RETVAL_0_I_I]], 2251799813685247
// AMDGCNSPIRV-NEXT: [[BF_SET9_I:%.*]] = or disjoint i64 [[BF_VALUE_I]], 9221120237041090560
-// AMDGCNSPIRV-NEXT: [[TMP10:%.*]] = bitcast i64 [[BF_SET9_I]] to double
-// AMDGCNSPIRV-NEXT: ret double [[TMP10]]
+// AMDGCNSPIRV-NEXT: [[TMP14:%.*]] = bitcast i64 [[BF_SET9_I]] to double
+// AMDGCNSPIRV-NEXT: ret double [[TMP14]]
//
extern "C" __device__ double test_nan(const char *tag) {
return nan(tag);
@@ -4679,22 +4807,22 @@ extern "C" __device__ double test_nearbyint(double x) {
// DEFAULT-LABEL: @test_nextafterf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_nextafterf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_nextafter_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_nextafter_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_nextafterf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_nextafterf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_nextafter_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_nextafterf(
@@ -4708,22 +4836,22 @@ extern "C" __device__ float test_nextafterf(float x, float y) {
// DEFAULT-LABEL: @test_nextafter(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_nextafter(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_nextafter_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_nextafter_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_nextafter(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_nextafter(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_nextafter_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_nextafter(
@@ -4737,22 +4865,22 @@ extern "C" __device__ double test_nextafter(double x, double y) {
// DEFAULT-LABEL: @test_norm3df(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_norm3df(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len3_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len3_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_norm3df(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_norm3df(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_norm3df(
@@ -4766,22 +4894,22 @@ extern "C" __device__ float test_norm3df(float x, float y, float z) {
// DEFAULT-LABEL: @test_norm3d(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_norm3d(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len3_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len3_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_norm3d(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_norm3d(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_norm3d(
@@ -4795,22 +4923,22 @@ extern "C" __device__ double test_norm3d(double x, double y, double z) {
// DEFAULT-LABEL: @test_norm4df(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_norm4df(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len4_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]], float noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_len4_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]], float noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_norm4df(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_norm4df(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_len4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_norm4df(
@@ -4824,22 +4952,22 @@ extern "C" __device__ float test_norm4df(float x, float y, float z, float w) {
// DEFAULT-LABEL: @test_norm4d(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_norm4d(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len4_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]], double noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_len4_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]], double noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_norm4d(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_norm4d(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_len4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_norm4d(
@@ -4853,22 +4981,22 @@ extern "C" __device__ double test_norm4d(double x, double y, double z, double w)
// DEFAULT-LABEL: @test_normcdff(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_normcdff(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdf_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdf_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_normcdff(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_normcdff(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdf_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_normcdff(
@@ -4882,22 +5010,22 @@ extern "C" __device__ float test_normcdff(float x) {
// DEFAULT-LABEL: @test_normcdf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_normcdf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdf_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdf_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_normcdf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_normcdf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdf_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_normcdf(
@@ -4911,22 +5039,22 @@ extern "C" __device__ double test_normcdf(double x) {
// DEFAULT-LABEL: @test_normcdfinvf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_normcdfinvf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdfinv_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_ncdfinv_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_normcdfinvf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_normcdfinvf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_ncdfinv_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_normcdfinvf(
@@ -4940,22 +5068,22 @@ extern "C" __device__ float test_normcdfinvf(float x) {
// DEFAULT-LABEL: @test_normcdfinv(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_normcdfinv(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdfinv_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_ncdfinv_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_normcdfinv(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_normcdfinv(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_ncdfinv_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_normcdfinv(
@@ -5056,12 +5184,12 @@ extern "C" __device__ double test_normcdfinv(double x) {
// AMDGCNSPIRV-NEXT: [[__A_ADDR_0_I3:%.*]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I:%.*]], [[WHILE_BODY_I]] ], [ [[Y:%.*]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[__DIM_ADDR_0_I2:%.*]] = phi i32 [ [[DEC_I:%.*]], [[WHILE_BODY_I]] ], [ [[X]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[DEC_I]] = add nsw i32 [[__DIM_ADDR_0_I2]], -1
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA17]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// AMDGCNSPIRV-NEXT: [[MUL_I:%.*]] = fmul contract float [[TMP0]], [[TMP0]]
// AMDGCNSPIRV-NEXT: [[ADD_I]] = fadd contract float [[__R_0_I4]], [[MUL_I]]
// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__A_ADDR_0_I3]], i64 4
// AMDGCNSPIRV-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5NORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP21:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5NORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP20:![0-9]+]]
// AMDGCNSPIRV: _ZL5normfiPKf.exit:
// AMDGCNSPIRV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = tail call contract noundef addrspace(4) float @llvm.sqrt.f32(float [[__R_0_I_LCSSA]])
@@ -5160,12 +5288,12 @@ extern "C" __device__ float test_normf(int x, const float *y) {
// AMDGCNSPIRV-NEXT: [[__A_ADDR_0_I3:%.*]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I:%.*]], [[WHILE_BODY_I]] ], [ [[Y:%.*]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[__DIM_ADDR_0_I2:%.*]] = phi i32 [ [[DEC_I:%.*]], [[WHILE_BODY_I]] ], [ [[X]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[DEC_I]] = add nsw i32 [[__DIM_ADDR_0_I2]], -1
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA19]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// AMDGCNSPIRV-NEXT: [[MUL_I:%.*]] = fmul contract double [[TMP0]], [[TMP0]]
// AMDGCNSPIRV-NEXT: [[ADD_I]] = fadd contract double [[__R_0_I4]], [[MUL_I]]
// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__A_ADDR_0_I3]], i64 8
// AMDGCNSPIRV-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL4NORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL4NORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP21:![0-9]+]]
// AMDGCNSPIRV: _ZL4normiPKd.exit:
// AMDGCNSPIRV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
// AMDGCNSPIRV-NEXT: [[TMP1:%.*]] = tail call contract noundef addrspace(4) double @llvm.sqrt.f64(double [[__R_0_I_LCSSA]])
@@ -5177,22 +5305,22 @@ extern "C" __device__ double test_norm(int x, const double *y) {
// DEFAULT-LABEL: @test_powf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_powf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_powf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_powf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_powf(
@@ -5206,22 +5334,22 @@ extern "C" __device__ float test_powf(float x, float y) {
// DEFAULT-LABEL: @test_pow(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_pow(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pow_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pow_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_pow(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_pow(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pow_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_pow(
@@ -5235,22 +5363,22 @@ extern "C" __device__ double test_pow(double x, double y) {
// DEFAULT-LABEL: @test_powif(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_powif(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pown_f32(float noundef nofpclass(nan inf) [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pown_f32(float noundef nofpclass(nan inf) [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_powif(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_powif(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pown_f32(float noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_powif(
@@ -5264,22 +5392,22 @@ extern "C" __device__ float test_powif(float x, int y) {
// DEFAULT-LABEL: @test_powi(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_powi(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pown_f64(double noundef nofpclass(nan inf) [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_pown_f64(double noundef nofpclass(nan inf) [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_powi(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_powi(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_pown_f64(double noundef [[X:%.*]], i32 noundef [[Y:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_powi(
@@ -5293,22 +5421,22 @@ extern "C" __device__ double test_powi(double x, int y) {
// DEFAULT-LABEL: @test_rcbrtf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rcbrtf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rcbrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rcbrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_rcbrtf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_rcbrtf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rcbrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rcbrtf(
@@ -5322,22 +5450,22 @@ extern "C" __device__ float test_rcbrtf(float x) {
// DEFAULT-LABEL: @test_rcbrt(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rcbrt(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rcbrt_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rcbrt_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_rcbrt(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_rcbrt(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rcbrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rcbrt(
@@ -5351,22 +5479,22 @@ extern "C" __device__ double test_rcbrt(double x) {
// DEFAULT-LABEL: @test_remainderf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_remainderf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remainder_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remainder_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_remainderf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_remainderf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_remainder_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_remainderf(
@@ -5380,22 +5508,22 @@ extern "C" __device__ float test_remainderf(float x, float y) {
// DEFAULT-LABEL: @test_remainder(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_remainder(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remainder_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remainder_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_remainder(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_remainder(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_remainder_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_remainder(
@@ -5410,41 +5538,41 @@ extern "C" __device__ double test_remainder(double x, double y) {
// DEFAULT-LABEL: @test_remquof(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA12]]
// DEFAULT-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA12]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_remquof(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remquo_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_remquo_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA12]]
// FINITEONLY-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA12]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_remquof(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA12]]
// APPROX-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA12]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_remquof(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef float @__ocml_remquo_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA13]]
// NCRDIV-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA13]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_remquof(
@@ -5453,8 +5581,8 @@ extern "C" __device__ double test_remainder(double x, double y) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func noundef addrspace(4) float @__ocml_remquo_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA13]]
-// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA13]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA12]]
+// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA12]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret float [[CALL_I]]
//
@@ -5465,41 +5593,41 @@ extern "C" __device__ float test_remquof(float x, float y, int* z) {
// DEFAULT-LABEL: @test_remquo(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA12]]
// DEFAULT-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA12]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_remquo(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remquo_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_remquo_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA12]]
// FINITEONLY-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA12]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_remquo(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA12]]
// APPROX-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA12]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_remquo(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca i32, align 4, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract noundef double @__ocml_remquo_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA13]]
// NCRDIV-NEXT: store i32 [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA13]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_remquo(
@@ -5508,8 +5636,8 @@ extern "C" __device__ float test_remquof(float x, float y, int* z) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func noundef addrspace(4) double @__ocml_remquo_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA13]]
-// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA13]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA12]]
+// AMDGCNSPIRV-NEXT: store i32 [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA12]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret double [[CALL_I]]
//
@@ -5519,22 +5647,22 @@ extern "C" __device__ double test_remquo(double x, double y, int* z) {
// DEFAULT-LABEL: @test_rhypotf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rhypotf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rhypot_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rhypot_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_rhypotf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_rhypotf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rhypot_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rhypotf(
@@ -5548,22 +5676,22 @@ extern "C" __device__ float test_rhypotf(float x, float y) {
// DEFAULT-LABEL: @test_rhypot(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rhypot(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rhypot_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rhypot_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_rhypot(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_rhypot(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rhypot_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rhypot(
@@ -5650,7 +5778,7 @@ extern "C" __device__ double test_rint(double x) {
// DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
// DEFAULT: _ZL6rnormfiPKf.exit:
// DEFAULT-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rnormf(
@@ -5670,7 +5798,7 @@ extern "C" __device__ double test_rint(double x) {
// FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
// FINITEONLY: _ZL6rnormfiPKf.exit:
// FINITEONLY-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_rnormf(
@@ -5690,7 +5818,7 @@ extern "C" __device__ double test_rint(double x) {
// APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
// APPROX: _ZL6rnormfiPKf.exit:
// APPROX-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_rnormf(
@@ -5710,7 +5838,7 @@ extern "C" __device__ double test_rint(double x) {
// NCRDIV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP24:![0-9]+]]
// NCRDIV: _ZL6rnormfiPKf.exit:
// NCRDIV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rnormf(
@@ -5722,12 +5850,12 @@ extern "C" __device__ double test_rint(double x) {
// AMDGCNSPIRV-NEXT: [[__A_ADDR_0_I3:%.*]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I:%.*]], [[WHILE_BODY_I]] ], [ [[Y:%.*]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[__DIM_ADDR_0_I2:%.*]] = phi i32 [ [[DEC_I:%.*]], [[WHILE_BODY_I]] ], [ [[X]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[DEC_I]] = add nsw i32 [[__DIM_ADDR_0_I2]], -1
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA17]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// AMDGCNSPIRV-NEXT: [[MUL_I:%.*]] = fmul contract float [[TMP0]], [[TMP0]]
// AMDGCNSPIRV-NEXT: [[ADD_I]] = fadd contract float [[__R_0_I4]], [[MUL_I]]
// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__A_ADDR_0_I3]], i64 4
// AMDGCNSPIRV-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
// AMDGCNSPIRV: _ZL6rnormfiPKf.exit:
// AMDGCNSPIRV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = tail call contract spir_func noundef addrspace(4) float @__ocml_rsqrt_f32(float noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
@@ -5754,7 +5882,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
// DEFAULT: _ZL5rnormiPKd.exit:
// DEFAULT-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rnorm(
@@ -5774,7 +5902,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
// FINITEONLY: _ZL5rnormiPKd.exit:
// FINITEONLY-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[__R_0_I_LCSSA]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_rnorm(
@@ -5794,7 +5922,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
// APPROX: _ZL5rnormiPKd.exit:
// APPROX-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_rnorm(
@@ -5814,7 +5942,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// NCRDIV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP25:![0-9]+]]
// NCRDIV: _ZL5rnormiPKd.exit:
// NCRDIV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rnorm(
@@ -5826,12 +5954,12 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// AMDGCNSPIRV-NEXT: [[__A_ADDR_0_I3:%.*]] = phi ptr addrspace(4) [ [[INCDEC_PTR_I:%.*]], [[WHILE_BODY_I]] ], [ [[Y:%.*]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[__DIM_ADDR_0_I2:%.*]] = phi i32 [ [[DEC_I:%.*]], [[WHILE_BODY_I]] ], [ [[X]], [[ENTRY]] ]
// AMDGCNSPIRV-NEXT: [[DEC_I]] = add nsw i32 [[__DIM_ADDR_0_I2]], -1
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA19]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// AMDGCNSPIRV-NEXT: [[MUL_I:%.*]] = fmul contract double [[TMP0]], [[TMP0]]
// AMDGCNSPIRV-NEXT: [[ADD_I]] = fadd contract double [[__R_0_I4]], [[MUL_I]]
// AMDGCNSPIRV-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr addrspace(4) [[__A_ADDR_0_I3]], i64 8
// AMDGCNSPIRV-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
-// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP24:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
// AMDGCNSPIRV: _ZL5rnormiPKd.exit:
// AMDGCNSPIRV-NEXT: [[__R_0_I_LCSSA:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[ADD_I]], [[WHILE_BODY_I]] ]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = tail call contract spir_func noundef addrspace(4) double @__ocml_rsqrt_f64(double noundef [[__R_0_I_LCSSA]]) #[[ATTR13]]
@@ -5843,22 +5971,22 @@ extern "C" __device__ double test_rnorm(int x, const double* y) {
// DEFAULT-LABEL: @test_rnorm3df(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rnorm3df(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen3_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen3_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_rnorm3df(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_rnorm3df(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen3_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rnorm3df(
@@ -5872,22 +6000,22 @@ extern "C" __device__ float test_rnorm3df(float x, float y, float z) {
// DEFAULT-LABEL: @test_rnorm3d(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rnorm3d(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen3_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen3_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_rnorm3d(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_rnorm3d(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen3_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rnorm3d(
@@ -5901,22 +6029,22 @@ extern "C" __device__ double test_rnorm3d(double x, double y, double z) {
// DEFAULT-LABEL: @test_rnorm4df(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rnorm4df(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen4_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]], float noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rlen4_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]], float noundef nofpclass(nan inf) [[Z:%.*]], float noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_rnorm4df(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_rnorm4df(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rlen4_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]], float noundef [[Z:%.*]], float noundef [[W:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rnorm4df(
@@ -5930,22 +6058,22 @@ extern "C" __device__ float test_rnorm4df(float x, float y, float z, float w) {
// DEFAULT-LABEL: @test_rnorm4d(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rnorm4d(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen4_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]], double noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rlen4_f64(double noundef nofpclass(nan inf) [[X:%.*]], double noundef nofpclass(nan inf) [[Y:%.*]], double noundef nofpclass(nan inf) [[Z:%.*]], double noundef nofpclass(nan inf) [[W:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_rnorm4d(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_rnorm4d(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rlen4_f64(double noundef [[X:%.*]], double noundef [[Y:%.*]], double noundef [[Z:%.*]], double noundef [[W:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rnorm4d(
@@ -6017,22 +6145,22 @@ extern "C" __device__ double test_round(double x) {
// DEFAULT-LABEL: @test_rsqrtf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rsqrtf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_rsqrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_rsqrtf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_rsqrtf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_rsqrt_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rsqrtf(
@@ -6046,22 +6174,22 @@ extern "C" __device__ float test_rsqrtf(float x) {
// DEFAULT-LABEL: @test_rsqrt(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_rsqrt(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_rsqrt_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_rsqrt(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_rsqrt(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_rsqrt_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_rsqrt(
@@ -6246,45 +6374,45 @@ extern "C" __device__ BOOL_TYPE test___signbit(double x) {
// DEFAULT-LABEL: @test_sincosf(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16]]
// DEFAULT-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret void
//
// FINITEONLY-LABEL: @test_sincosf(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincos_f32(float noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincos_f32(float noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16]]
// FINITEONLY-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret void
//
// APPROX-LABEL: @test_sincosf(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// APPROX-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16]]
// APPROX-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret void
//
// NCRDIV-LABEL: @test_sincosf(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincos_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA17]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA17]]
// NCRDIV-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA17]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret void
//
// AMDGCNSPIRV-LABEL: @test_sincosf(
@@ -6293,9 +6421,9 @@ extern "C" __device__ BOOL_TYPE test___signbit(double x) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func addrspace(4) float @__ocml_sincos_f32(float noundef [[X:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: store float [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA17]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA17]]
-// AMDGCNSPIRV-NEXT: store float [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA17]]
+// AMDGCNSPIRV-NEXT: store float [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA16]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA16]]
+// AMDGCNSPIRV-NEXT: store float [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA16]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret void
//
@@ -6306,45 +6434,45 @@ extern "C" __device__ void test_sincosf(float x, float *y, float *z) {
// DEFAULT-LABEL: @test_sincos(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18]]
// DEFAULT-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA18]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret void
//
// FINITEONLY-LABEL: @test_sincos(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincos_f64(double noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincos_f64(double noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18]]
// FINITEONLY-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA18]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret void
//
// APPROX-LABEL: @test_sincos(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
// APPROX-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18]]
// APPROX-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA18]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret void
//
// NCRDIV-LABEL: @test_sincos(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincos_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA19]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA19]]
// NCRDIV-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA19]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret void
//
// AMDGCNSPIRV-LABEL: @test_sincos(
@@ -6353,9 +6481,9 @@ extern "C" __device__ void test_sincosf(float x, float *y, float *z) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 8, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func addrspace(4) double @__ocml_sincos_f64(double noundef [[X:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: store double [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 8, !tbaa [[TBAA19]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__TMP_ASCAST_I]], align 8, !tbaa [[TBAA19]]
-// AMDGCNSPIRV-NEXT: store double [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 8, !tbaa [[TBAA19]]
+// AMDGCNSPIRV-NEXT: store double [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 8, !tbaa [[TBAA18]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__TMP_ASCAST_I]], align 8, !tbaa [[TBAA18]]
+// AMDGCNSPIRV-NEXT: store double [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 8, !tbaa [[TBAA18]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 8, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret void
//
@@ -6366,45 +6494,45 @@ extern "C" __device__ void test_sincos(double x, double *y, double *z) {
// DEFAULT-LABEL: @test_sincospif(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16]]
// DEFAULT-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret void
//
// FINITEONLY-LABEL: @test_sincospif(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincospi_f32(float noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) float @__ocml_sincospi_f32(float noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16]]
// FINITEONLY-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret void
//
// APPROX-LABEL: @test_sincospif(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// APPROX-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA16]]
// APPROX-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret void
//
// NCRDIV-LABEL: @test_sincospif(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca float, align 4, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract float @__ocml_sincospi_f32(float noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA17]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(5) [[__TMP_I]], align 4, !tbaa [[TBAA17]]
// NCRDIV-NEXT: store float [[TMP0]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA17]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 4, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret void
//
// AMDGCNSPIRV-LABEL: @test_sincospif(
@@ -6413,9 +6541,9 @@ extern "C" __device__ void test_sincos(double x, double *y, double *z) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func addrspace(4) float @__ocml_sincospi_f32(float noundef [[X:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: store float [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA17]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA17]]
-// AMDGCNSPIRV-NEXT: store float [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA17]]
+// AMDGCNSPIRV-NEXT: store float [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA16]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load float, ptr addrspace(4) [[__TMP_ASCAST_I]], align 4, !tbaa [[TBAA16]]
+// AMDGCNSPIRV-NEXT: store float [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA16]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 4, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret void
//
@@ -6426,45 +6554,45 @@ extern "C" __device__ void test_sincospif(float x, float *y, float *z) {
// DEFAULT-LABEL: @test_sincospi(
// DEFAULT-NEXT: entry:
// DEFAULT-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// DEFAULT-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// DEFAULT-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
// DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18]]
// DEFAULT-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA18]]
-// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// DEFAULT-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// DEFAULT-NEXT: ret void
//
// FINITEONLY-LABEL: @test_sincospi(
// FINITEONLY-NEXT: entry:
// FINITEONLY-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincospi_f64(double noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = call nnan ninf contract nofpclass(nan inf) double @__ocml_sincospi_f64(double noundef nofpclass(nan inf) [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// FINITEONLY-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18]]
// FINITEONLY-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA18]]
-// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// FINITEONLY-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// FINITEONLY-NEXT: ret void
//
// APPROX-LABEL: @test_sincospi(
// APPROX-NEXT: entry:
// APPROX-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// APPROX-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// APPROX-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// APPROX-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA18]]
// APPROX-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA18]]
// APPROX-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA18]]
-// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// APPROX-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// APPROX-NEXT: ret void
//
// NCRDIV-LABEL: @test_sincospi(
// NCRDIV-NEXT: entry:
// NCRDIV-NEXT: [[__TMP_I:%.*]] = alloca double, align 8, addrspace(5)
-// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
-// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR14]]
+// NCRDIV-NEXT: call void @llvm.lifetime.start.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = call contract double @__ocml_sincospi_f64(double noundef [[X:%.*]], ptr addrspace(5) noundef [[__TMP_I]]) #[[ATTR16]]
// NCRDIV-NEXT: store double [[CALL_I]], ptr [[Y:%.*]], align 8, !tbaa [[TBAA19]]
// NCRDIV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(5) [[__TMP_I]], align 8, !tbaa [[TBAA19]]
// NCRDIV-NEXT: store double [[TMP0]], ptr [[Z:%.*]], align 8, !tbaa [[TBAA19]]
-// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR15]]
+// NCRDIV-NEXT: call void @llvm.lifetime.end.p5(i64 8, ptr addrspace(5) [[__TMP_I]]) #[[ATTR17]]
// NCRDIV-NEXT: ret void
//
// AMDGCNSPIRV-LABEL: @test_sincospi(
@@ -6473,9 +6601,9 @@ extern "C" __device__ void test_sincospif(float x, float *y, float *z) {
// AMDGCNSPIRV-NEXT: [[__TMP_ASCAST_I:%.*]] = addrspacecast ptr [[__TMP_I]] to ptr addrspace(4)
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.start.p0(i64 8, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = call contract spir_func addrspace(4) double @__ocml_sincospi_f64(double noundef [[X:%.*]], ptr noundef nonnull [[__TMP_I]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: store double [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 8, !tbaa [[TBAA19]]
-// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__TMP_ASCAST_I]], align 8, !tbaa [[TBAA19]]
-// AMDGCNSPIRV-NEXT: store double [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 8, !tbaa [[TBAA19]]
+// AMDGCNSPIRV-NEXT: store double [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 8, !tbaa [[TBAA18]]
+// AMDGCNSPIRV-NEXT: [[TMP0:%.*]] = load double, ptr addrspace(4) [[__TMP_ASCAST_I]], align 8, !tbaa [[TBAA18]]
+// AMDGCNSPIRV-NEXT: store double [[TMP0]], ptr addrspace(4) [[Z:%.*]], align 8, !tbaa [[TBAA18]]
// AMDGCNSPIRV-NEXT: call addrspace(4) void @llvm.lifetime.end.p0(i64 8, ptr nonnull [[__TMP_I]]) #[[ATTR15]]
// AMDGCNSPIRV-NEXT: ret void
//
@@ -6485,22 +6613,22 @@ extern "C" __device__ void test_sincospi(double x, double *y, double *z) {
// DEFAULT-LABEL: @test_sinf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_sinf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_sinf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I1:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I1]]
//
// NCRDIV-LABEL: @test_sinf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_sinf(
@@ -6514,22 +6642,22 @@ extern "C" __device__ float test_sinf(float x) {
// DEFAULT-LABEL: @test_sin(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_sin(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sin_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sin_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_sin(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_sin(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sin_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_sin(
@@ -6543,22 +6671,22 @@ extern "C" __device__ double test_sin(double x) {
// DEFAULT-LABEL: @test_sinpif(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_sinpif(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sinpi_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_sinpi_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_sinpif(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_sinpif(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_sinpi_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_sinpif(
@@ -6572,22 +6700,22 @@ extern "C" __device__ float test_sinpif(float x) {
// DEFAULT-LABEL: @test_sinpi(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_sinpi(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sinpi_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_sinpi_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_sinpi(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_sinpi(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_sinpi_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_sinpi(
@@ -6659,22 +6787,22 @@ extern "C" __device__ double test_sqrt(double x) {
// DEFAULT-LABEL: @test_tanf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_tanf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tan_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tan_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_tanf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_tanf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tan_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_tanf(
@@ -6688,22 +6816,22 @@ extern "C" __device__ float test_tanf(float x) {
// DEFAULT-LABEL: @test_tan(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_tan(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tan_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tan_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_tan(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_tan(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tan_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_tan(
@@ -6717,22 +6845,22 @@ extern "C" __device__ double test_tan(double x) {
// DEFAULT-LABEL: @test_tanhf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_tanhf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tanh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tanh_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_tanhf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_tanhf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tanh_f32(float noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_tanhf(
@@ -6746,22 +6874,22 @@ extern "C" __device__ float test_tanhf(float x) {
// DEFAULT-LABEL: @test_tanh(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_tanh(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tanh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tanh_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_tanh(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_tanh(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tanh_f64(double noundef [[X:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_tanh(
@@ -6775,22 +6903,22 @@ extern "C" __device__ double test_tanh(double x) {
// DEFAULT-LABEL: @test_tgammaf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_tgammaf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tgamma_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_tgamma_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_tgammaf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_tgammaf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_tgamma_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_tgammaf(
@@ -6804,22 +6932,22 @@ extern "C" __device__ float test_tgammaf(float x) {
// DEFAULT-LABEL: @test_tgamma(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_tgamma(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tgamma_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_tgamma_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_tgamma(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_tgamma(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_tgamma_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_tgamma(
@@ -6891,22 +7019,22 @@ extern "C" __device__ double test_trunc(double x) {
// DEFAULT-LABEL: @test_y0f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_y0f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_y0f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_y0f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_y0f(
@@ -6920,22 +7048,22 @@ extern "C" __device__ float test_y0f(float x) {
// DEFAULT-LABEL: @test_y0(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_y0(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_y0(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_y0(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_y0(
@@ -6949,22 +7077,22 @@ extern "C" __device__ double test_y0(double x) {
// DEFAULT-LABEL: @test_y1f(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test_y1f(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test_y1f(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test_y1f(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_y1f(
@@ -6978,22 +7106,22 @@ extern "C" __device__ float test_y1f(float x) {
// DEFAULT-LABEL: @test_y1(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret double [[CALL_I]]
//
// FINITEONLY-LABEL: @test_y1(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret double [[CALL_I]]
//
// APPROX-LABEL: @test_y1(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret double [[CALL_I]]
//
// NCRDIV-LABEL: @test_y1(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret double [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test_y1(
@@ -7012,14 +7140,14 @@ extern "C" __device__ double test_y1(double x) {
// DEFAULT-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// DEFAULT-NEXT: ]
// DEFAULT: if.then.i:
-// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL3YNFIF_EXIT:%.*]]
// DEFAULT: if.then2.i:
-// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL3YNFIF_EXIT]]
// DEFAULT: if.end4.i:
-// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]]
-// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]]
+// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// DEFAULT-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3YNFIF_EXIT]]
// DEFAULT: for.body.i:
@@ -7045,14 +7173,14 @@ extern "C" __device__ double test_y1(double x) {
// FINITEONLY-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// FINITEONLY-NEXT: ]
// FINITEONLY: if.then.i:
-// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL3YNFIF_EXIT:%.*]]
// FINITEONLY: if.then2.i:
-// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL3YNFIF_EXIT]]
// FINITEONLY: if.end4.i:
-// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
-// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y0_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
+// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_y1_f32(float noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// FINITEONLY-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3YNFIF_EXIT]]
// FINITEONLY: for.body.i:
@@ -7078,14 +7206,14 @@ extern "C" __device__ double test_y1(double x) {
// APPROX-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// APPROX-NEXT: ]
// APPROX: if.then.i:
-// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y:%.*]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL3YNFIF_EXIT:%.*]]
// APPROX: if.then2.i:
-// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL3YNFIF_EXIT]]
// APPROX: if.end4.i:
-// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]]
-// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]]
+// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// APPROX-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3YNFIF_EXIT]]
// APPROX: for.body.i:
@@ -7111,14 +7239,14 @@ extern "C" __device__ double test_y1(double x) {
// NCRDIV-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// NCRDIV-NEXT: ]
// NCRDIV: if.then.i:
-// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL3YNFIF_EXIT:%.*]]
// NCRDIV: if.then2.i:
-// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL3YNFIF_EXIT]]
// NCRDIV: if.end4.i:
-// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR14]]
-// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_y0_f32(float noundef [[Y]]) #[[ATTR16]]
+// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef float @__ocml_y1_f32(float noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// NCRDIV-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL3YNFIF_EXIT]]
// NCRDIV: for.body.i:
@@ -7165,7 +7293,7 @@ extern "C" __device__ double test_y1(double x) {
// AMDGCNSPIRV-NEXT: [[SUB_I]] = fsub contract float [[MUL8_I]], [[__X0_0_I2]]
// AMDGCNSPIRV-NEXT: [[INC_I]] = add nuw nsw i32 [[__I_0_I4]], 1
// AMDGCNSPIRV-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC_I]], [[X]]
-// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL3YNFIF_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP25:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL3YNFIF_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP24:![0-9]+]]
// AMDGCNSPIRV: _ZL3ynfif.exit:
// AMDGCNSPIRV-NEXT: [[RETVAL_0_I:%.*]] = phi float [ [[CALL_I20_I]], [[IF_THEN_I]] ], [ [[CALL_I22_I]], [[IF_THEN2_I]] ], [ [[CALL_I21_I]], [[IF_END4_I]] ], [ [[SUB_I]], [[FOR_BODY_I]] ]
// AMDGCNSPIRV-NEXT: ret float [[RETVAL_0_I]]
@@ -7181,14 +7309,14 @@ extern "C" __device__ float test_ynf(int x, float y) {
// DEFAULT-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// DEFAULT-NEXT: ]
// DEFAULT: if.then.i:
-// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL2YNID_EXIT:%.*]]
// DEFAULT: if.then2.i:
-// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: br label [[_ZL2YNID_EXIT]]
// DEFAULT: if.end4.i:
-// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]]
-// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]]
+// DEFAULT-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]]
// DEFAULT-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// DEFAULT-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2YNID_EXIT]]
// DEFAULT: for.body.i:
@@ -7214,14 +7342,14 @@ extern "C" __device__ float test_ynf(int x, float y) {
// FINITEONLY-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// FINITEONLY-NEXT: ]
// FINITEONLY: if.then.i:
-// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I20_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL2YNID_EXIT:%.*]]
// FINITEONLY: if.then2.i:
-// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I22_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: br label [[_ZL2YNID_EXIT]]
// FINITEONLY: if.end4.i:
-// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
-// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y0_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
+// FINITEONLY-NEXT: [[CALL_I21_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) double @__ocml_y1_f64(double noundef nofpclass(nan inf) [[Y]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// FINITEONLY-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2YNID_EXIT]]
// FINITEONLY: for.body.i:
@@ -7247,14 +7375,14 @@ extern "C" __device__ float test_ynf(int x, float y) {
// APPROX-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// APPROX-NEXT: ]
// APPROX: if.then.i:
-// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y:%.*]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL2YNID_EXIT:%.*]]
// APPROX: if.then2.i:
-// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: br label [[_ZL2YNID_EXIT]]
// APPROX: if.end4.i:
-// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]]
-// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]]
+// APPROX-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]]
// APPROX-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// APPROX-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2YNID_EXIT]]
// APPROX: for.body.i:
@@ -7280,14 +7408,14 @@ extern "C" __device__ float test_ynf(int x, float y) {
// NCRDIV-NEXT: i32 1, label [[IF_THEN2_I:%.*]]
// NCRDIV-NEXT: ]
// NCRDIV: if.then.i:
-// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I20_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL2YNID_EXIT:%.*]]
// NCRDIV: if.then2.i:
-// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I22_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: br label [[_ZL2YNID_EXIT]]
// NCRDIV: if.end4.i:
-// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR14]]
-// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef double @__ocml_y0_f64(double noundef [[Y]]) #[[ATTR16]]
+// NCRDIV-NEXT: [[CALL_I21_I:%.*]] = tail call contract noundef double @__ocml_y1_f64(double noundef [[Y]]) #[[ATTR16]]
// NCRDIV-NEXT: [[CMP7_I1:%.*]] = icmp sgt i32 [[X]], 1
// NCRDIV-NEXT: br i1 [[CMP7_I1]], label [[FOR_BODY_I:%.*]], label [[_ZL2YNID_EXIT]]
// NCRDIV: for.body.i:
@@ -7334,7 +7462,7 @@ extern "C" __device__ float test_ynf(int x, float y) {
// AMDGCNSPIRV-NEXT: [[SUB_I]] = fsub contract double [[MUL8_I]], [[__X0_0_I2]]
// AMDGCNSPIRV-NEXT: [[INC_I]] = add nuw nsw i32 [[__I_0_I4]], 1
// AMDGCNSPIRV-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i32 [[INC_I]], [[X]]
-// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL2YNID_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP26:![0-9]+]]
+// AMDGCNSPIRV-NEXT: br i1 [[EXITCOND_NOT]], label [[_ZL2YNID_EXIT]], label [[FOR_BODY_I]], !llvm.loop [[LOOP25:![0-9]+]]
// AMDGCNSPIRV: _ZL2ynid.exit:
// AMDGCNSPIRV-NEXT: [[RETVAL_0_I:%.*]] = phi double [ [[CALL_I20_I]], [[IF_THEN_I]] ], [ [[CALL_I22_I]], [[IF_THEN2_I]] ], [ [[CALL_I21_I]], [[IF_END4_I]] ], [ [[SUB_I]], [[FOR_BODY_I]] ]
// AMDGCNSPIRV-NEXT: ret double [[RETVAL_0_I]]
@@ -7345,22 +7473,22 @@ extern "C" __device__ double test_yn(int x, double y) {
// DEFAULT-LABEL: @test___cosf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test___cosf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test___cosf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test___cosf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test___cosf(
@@ -7616,22 +7744,22 @@ extern "C" __device__ float test___frsqrt_rn(float x) {
// DEFAULT-LABEL: @test___fsqrt_rn(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test___fsqrt_rn(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sqrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR12]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sqrt_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test___fsqrt_rn(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test___fsqrt_rn(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X:%.*]]) #[[ATTR12]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sqrt_f32(float noundef [[X:%.*]]) #[[ATTR14]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test___fsqrt_rn(
@@ -7761,22 +7889,22 @@ extern "C" __device__ float test___logf(float x) {
// DEFAULT-LABEL: @test___powf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR13]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR15]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test___powf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR13]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_pow_f32(float noundef nofpclass(nan inf) [[X:%.*]], float noundef nofpclass(nan inf) [[Y:%.*]]) #[[ATTR15]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test___powf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR13]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR15]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test___powf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR13]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_pow_f32(float noundef [[X:%.*]], float noundef [[Y:%.*]]) #[[ATTR15]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test___powf(
@@ -7834,42 +7962,42 @@ extern "C" __device__ float test___saturatef(float x) {
// DEFAULT-LABEL: @test___sincosf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
-// DEFAULT-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]]
// DEFAULT-NEXT: store float [[CALL1_I]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
// DEFAULT-NEXT: ret void
//
// FINITEONLY-LABEL: @test___sincosf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
-// FINITEONLY-NEXT: [[CALL1_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL1_I:%.*]] = tail call nnan ninf contract nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]]
// FINITEONLY-NEXT: store float [[CALL1_I]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
// FINITEONLY-NEXT: ret void
//
// APPROX-LABEL: @test___sincosf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA16]]
-// APPROX-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]]
// APPROX-NEXT: store float [[CALL1_I]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA16]]
// APPROX-NEXT: ret void
//
// NCRDIV-LABEL: @test___sincosf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: store float [[CALL_I]], ptr [[Y:%.*]], align 4, !tbaa [[TBAA17]]
-// NCRDIV-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL1_I:%.*]] = tail call contract float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]]
// NCRDIV-NEXT: store float [[CALL1_I]], ptr [[Z:%.*]], align 4, !tbaa [[TBAA17]]
// NCRDIV-NEXT: ret void
//
// AMDGCNSPIRV-LABEL: @test___sincosf(
// AMDGCNSPIRV-NEXT: entry:
// AMDGCNSPIRV-NEXT: [[CALL_I:%.*]] = tail call contract spir_func addrspace(4) float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: store float [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA17]]
+// AMDGCNSPIRV-NEXT: store float [[CALL_I]], ptr addrspace(4) [[Y:%.*]], align 4, !tbaa [[TBAA16]]
// AMDGCNSPIRV-NEXT: [[CALL1_I:%.*]] = tail call contract spir_func addrspace(4) float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]]
-// AMDGCNSPIRV-NEXT: store float [[CALL1_I]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA17]]
+// AMDGCNSPIRV-NEXT: store float [[CALL1_I]], ptr addrspace(4) [[Z:%.*]], align 4, !tbaa [[TBAA16]]
// AMDGCNSPIRV-NEXT: ret void
//
extern "C" __device__ void test___sincosf(float x, float *y, float *z) {
@@ -7878,22 +8006,22 @@ extern "C" __device__ void test___sincosf(float x, float *y, float *z) {
// DEFAULT-LABEL: @test___sinf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// DEFAULT-NEXT: ret float [[CALL_I]]
//
// FINITEONLY-LABEL: @test___sinf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
// FINITEONLY-NEXT: ret float [[CALL_I]]
//
// APPROX-LABEL: @test___sinf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// APPROX-NEXT: ret float [[CALL_I]]
//
// NCRDIV-LABEL: @test___sinf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
// NCRDIV-NEXT: ret float [[CALL_I]]
//
// AMDGCNSPIRV-LABEL: @test___sinf(
@@ -7907,32 +8035,32 @@ extern "C" __device__ float test___sinf(float x) {
// DEFAULT-LABEL: @test___tanf(
// DEFAULT-NEXT: entry:
-// DEFAULT-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
-// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]]
+// DEFAULT-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
+// DEFAULT-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]]
// DEFAULT-NEXT: [[TMP0:%.*]] = tail call contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]])
// DEFAULT-NEXT: [[MUL_I:%.*]] = fmul contract float [[CALL_I3_I]], [[TMP0]]
// DEFAULT-NEXT: ret float [[MUL_I]]
//
// FINITEONLY-LABEL: @test___tanf(
// FINITEONLY-NEXT: entry:
-// FINITEONLY-NEXT: [[CALL_I3_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR14]]
-// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR14]]
+// FINITEONLY-NEXT: [[CALL_I3_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_sin_f32(float noundef nofpclass(nan inf) [[X:%.*]]) #[[ATTR16]]
+// FINITEONLY-NEXT: [[CALL_I_I:%.*]] = tail call nnan ninf contract noundef nofpclass(nan inf) float @__ocml_native_cos_f32(float noundef nofpclass(nan inf) [[X]]) #[[ATTR16]]
// FINITEONLY-NEXT: [[TMP0:%.*]] = tail call nnan ninf contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]])
// FINITEONLY-NEXT: [[MUL_I:%.*]] = fmul nnan ninf contract float [[CALL_I3_I]], [[TMP0]]
// FINITEONLY-NEXT: ret float [[MUL_I]]
//
// APPROX-LABEL: @test___tanf(
// APPROX-NEXT: entry:
-// APPROX-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
-// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]]
+// APPROX-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
+// APPROX-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]]
// APPROX-NEXT: [[TMP0:%.*]] = tail call contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]])
// APPROX-NEXT: [[MUL_I:%.*]] = fmul contract float [[CALL_I3_I]], [[TMP0]]
// APPROX-NEXT: ret float [[MUL_I]]
//
// NCRDIV-LABEL: @test___tanf(
// NCRDIV-NEXT: entry:
-// NCRDIV-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR14]]
-// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR14]]
+// NCRDIV-NEXT: [[CALL_I3_I:%.*]] = tail call contract noundef float @__ocml_native_sin_f32(float noundef [[X:%.*]]) #[[ATTR16]]
+// NCRDIV-NEXT: [[CALL_I_I:%.*]] = tail call contract noundef float @__ocml_native_cos_f32(float noundef [[X]]) #[[ATTR16]]
// NCRDIV-NEXT: [[TMP0:%.*]] = tail call contract float @llvm.amdgcn.rcp.f32(float [[CALL_I_I]])
// NCRDIV-NEXT: [[MUL_I:%.*]] = fmul contract float [[CALL_I3_I]], [[TMP0]]
// NCRDIV-NEXT: ret float [[MUL_I]]
>From ffae627ec8b0a3f3647177d1d1589e50879b8cdc Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Mon, 16 Jun 2025 21:38:50 -0700
Subject: [PATCH 4/8] Rebase && Fix AArch64 postidx test
---
.../AArch64/postidx-load.ll | 104 ++++++++----------
1 file changed, 46 insertions(+), 58 deletions(-)
diff --git a/llvm/test/Transforms/LoopStrengthReduce/AArch64/postidx-load.ll b/llvm/test/Transforms/LoopStrengthReduce/AArch64/postidx-load.ll
index 5976658ccdf86..c33c8f570076e 100644
--- a/llvm/test/Transforms/LoopStrengthReduce/AArch64/postidx-load.ll
+++ b/llvm/test/Transforms/LoopStrengthReduce/AArch64/postidx-load.ll
@@ -7,25 +7,18 @@
define i32 @i32_initially_postidx(ptr %p, i64 %n) {
; CHECK-LABEL: i32_initially_postidx:
; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: cmp x1, #1
-; CHECK-NEXT: b.lt .LBB0_5
-; CHECK-NEXT: // %bb.1: // %for.body.preheader
; CHECK-NEXT: mov w8, wzr
-; CHECK-NEXT: .LBB0_2: // %for.body
+; CHECK-NEXT: subs x9, x1, #1
+; CHECK-NEXT: b.lt .LBB0_2
+; CHECK-NEXT: .LBB0_1: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: ldr w9, [x0], #4
-; CHECK-NEXT: add w8, w8, w9
+; CHECK-NEXT: ldr w10, [x0], #4
+; CHECK-NEXT: add w8, w8, w10
; CHECK-NEXT: cmp w8, #0
-; CHECK-NEXT: b.lo .LBB0_5
-; CHECK-NEXT: // %bb.3: // %for.inc
-; CHECK-NEXT: // in Loop: Header=BB0_2 Depth=1
-; CHECK-NEXT: subs x1, x1, #1
-; CHECK-NEXT: b.ne .LBB0_2
-; CHECK-NEXT: // %bb.4: // %cleanup
-; CHECK-NEXT: mov w0, w8
-; CHECK-NEXT: ret
-; CHECK-NEXT: .LBB0_5:
-; CHECK-NEXT: mov w8, wzr
+; CHECK-NEXT: ccmp x9, #0, #4, hs
+; CHECK-NEXT: sub x9, x9, #1
+; CHECK-NEXT: b.ne .LBB0_1
+; CHECK-NEXT: .LBB0_2: // %cleanup
; CHECK-NEXT: mov w0, w8
; CHECK-NEXT: ret
entry:
@@ -55,25 +48,18 @@ cleanup:
define i32 @i32_initially_offset(ptr %p, i64 %n) {
; CHECK-LABEL: i32_initially_offset:
; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: cmp x1, #1
-; CHECK-NEXT: b.lt .LBB1_5
-; CHECK-NEXT: // %bb.1: // %for.body.preheader
; CHECK-NEXT: mov w8, wzr
-; CHECK-NEXT: .LBB1_2: // %for.body
+; CHECK-NEXT: subs x9, x1, #1
+; CHECK-NEXT: b.lt .LBB1_2
+; CHECK-NEXT: .LBB1_1: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: ldr w9, [x0], #4
-; CHECK-NEXT: add w8, w8, w9
+; CHECK-NEXT: ldr w10, [x0], #4
+; CHECK-NEXT: add w8, w8, w10
; CHECK-NEXT: cmp w8, #0
-; CHECK-NEXT: b.lo .LBB1_5
-; CHECK-NEXT: // %bb.3: // %for.cond
-; CHECK-NEXT: // in Loop: Header=BB1_2 Depth=1
-; CHECK-NEXT: subs x1, x1, #1
-; CHECK-NEXT: b.ne .LBB1_2
-; CHECK-NEXT: // %bb.4: // %cleanup
-; CHECK-NEXT: mov w0, w8
-; CHECK-NEXT: ret
-; CHECK-NEXT: .LBB1_5:
-; CHECK-NEXT: mov w8, wzr
+; CHECK-NEXT: ccmp x9, #0, #4, hs
+; CHECK-NEXT: sub x9, x9, #1
+; CHECK-NEXT: b.ne .LBB1_1
+; CHECK-NEXT: .LBB1_2: // %cleanup
; CHECK-NEXT: mov w0, w8
; CHECK-NEXT: ret
entry:
@@ -102,20 +88,21 @@ cleanup:
define float @float_initially_postidx(ptr %p, i64 %n) {
; CHECK-LABEL: float_initially_postidx:
; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: movi d0, #0000000000000000
-; CHECK-NEXT: cmp x1, #1
-; CHECK-NEXT: b.lt .LBB2_3
-; CHECK-NEXT: .LBB2_1: // %for.body
+; CHECK-NEXT: subs x8, x1, #1
+; CHECK-NEXT: b.lt .LBB2_4
+; CHECK-NEXT: // %bb.1: // %for.body.preheader
+; CHECK-NEXT: movi d1, #0000000000000000
+; CHECK-NEXT: movi d2, #0000000000000000
+; CHECK-NEXT: .LBB2_2: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: ldr s1, [x0], #4
-; CHECK-NEXT: fadd s0, s0, s1
-; CHECK-NEXT: fcmp s0, #0.0
-; CHECK-NEXT: b.mi .LBB2_4
-; CHECK-NEXT: // %bb.2: // %for.inc
-; CHECK-NEXT: // in Loop: Header=BB2_1 Depth=1
-; CHECK-NEXT: subs x1, x1, #1
-; CHECK-NEXT: b.ne .LBB2_1
-; CHECK-NEXT: .LBB2_3: // %cleanup
+; CHECK-NEXT: ldr s0, [x0], #4
+; CHECK-NEXT: fadd s2, s2, s0
+; CHECK-NEXT: fcmp s2, #0.0
+; CHECK-NEXT: fcsel s0, s1, s2, mi
+; CHECK-NEXT: ccmp x8, #0, #4, pl
+; CHECK-NEXT: sub x8, x8, #1
+; CHECK-NEXT: b.ne .LBB2_2
+; CHECK-NEXT: // %bb.3: // %cleanup
; CHECK-NEXT: ret
; CHECK-NEXT: .LBB2_4:
; CHECK-NEXT: movi d0, #0000000000000000
@@ -147,20 +134,21 @@ cleanup:
define float @float_initially_offset(ptr %p, i64 %n) {
; CHECK-LABEL: float_initially_offset:
; CHECK: // %bb.0: // %entry
-; CHECK-NEXT: movi d0, #0000000000000000
-; CHECK-NEXT: cmp x1, #1
-; CHECK-NEXT: b.lt .LBB3_3
-; CHECK-NEXT: .LBB3_1: // %for.body
+; CHECK-NEXT: subs x8, x1, #1
+; CHECK-NEXT: b.lt .LBB3_4
+; CHECK-NEXT: // %bb.1: // %for.body.preheader
+; CHECK-NEXT: movi d1, #0000000000000000
+; CHECK-NEXT: movi d2, #0000000000000000
+; CHECK-NEXT: .LBB3_2: // %for.body
; CHECK-NEXT: // =>This Inner Loop Header: Depth=1
-; CHECK-NEXT: ldr s1, [x0], #4
-; CHECK-NEXT: fadd s0, s0, s1
-; CHECK-NEXT: fcmp s0, #0.0
-; CHECK-NEXT: b.mi .LBB3_4
-; CHECK-NEXT: // %bb.2: // %for.cond
-; CHECK-NEXT: // in Loop: Header=BB3_1 Depth=1
-; CHECK-NEXT: subs x1, x1, #1
-; CHECK-NEXT: b.ne .LBB3_1
-; CHECK-NEXT: .LBB3_3: // %cleanup
+; CHECK-NEXT: ldr s0, [x0], #4
+; CHECK-NEXT: fadd s2, s2, s0
+; CHECK-NEXT: fcmp s2, #0.0
+; CHECK-NEXT: fcsel s0, s1, s2, mi
+; CHECK-NEXT: ccmp x8, #0, #4, pl
+; CHECK-NEXT: sub x8, x8, #1
+; CHECK-NEXT: b.ne .LBB3_2
+; CHECK-NEXT: // %bb.3: // %cleanup
; CHECK-NEXT: ret
; CHECK-NEXT: .LBB3_4:
; CHECK-NEXT: movi d0, #0000000000000000
>From b462feeb3ad16fc367064414bd816f82a8e589ee Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Fri, 20 Jun 2025 12:08:38 -0700
Subject: [PATCH 5/8] Passing TTI through simplifyLoop so we can count
instruction cost
---
.../llvm/Transforms/Utils/LoopConstrainer.h | 4 +-
llvm/include/llvm/Transforms/Utils/LoopPeel.h | 2 +-
.../llvm/Transforms/Utils/LoopSimplify.h | 3 +-
.../Scalar/InductiveRangeCheckElimination.cpp | 12 +++---
llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp | 10 +++--
llvm/lib/Transforms/Scalar/LoopFuse.cpp | 4 +-
.../Transforms/Scalar/LoopLoadElimination.cpp | 7 ++--
llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp | 4 +-
llvm/lib/Transforms/Utils/LoopConstrainer.cpp | 6 +--
llvm/lib/Transforms/Utils/LoopPeel.cpp | 6 +--
llvm/lib/Transforms/Utils/LoopSimplify.cpp | 17 +++++----
llvm/lib/Transforms/Utils/LoopUnroll.cpp | 4 +-
.../Transforms/Vectorize/LoopVectorize.cpp | 2 +-
.../Transforms/LoopSimplify/merge-exits.ll | 37 ++++++++-----------
.../LoopUnroll/runtime-loop-multiple-exits.ll | 7 +++-
llvm/test/Transforms/LoopUnroll/scevunroll.ll | 18 ++++-----
16 files changed, 77 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h b/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h
index 64db907e9a0f4..1e7e9adedc98d 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h
@@ -10,6 +10,7 @@
#define LLVM_TRANSFORMS_UTILS_LOOP_CONSTRAINER_H
#include "llvm/Support/Casting.h"
+#include "llvm/Transforms/Utils/Local.h"
#include "llvm/Transforms/Utils/ValueMapper.h"
#include <optional>
@@ -190,6 +191,7 @@ class LoopConstrainer {
Function &F;
LLVMContext &Ctx;
ScalarEvolution &SE;
+ TargetTransformInfo &TTI;
DominatorTree &DT;
LoopInfo &LI;
function_ref<void(Loop *, bool)> LPMAddNewLoop;
@@ -216,7 +218,7 @@ class LoopConstrainer {
LoopConstrainer(Loop &L, LoopInfo &LI,
function_ref<void(Loop *, bool)> LPMAddNewLoop,
const LoopStructure &LS, ScalarEvolution &SE,
- DominatorTree &DT, Type *T, SubRanges SR);
+ TargetTransformInfo &TTI, DominatorTree &DT, Type *T, SubRanges SR);
// Entry point for the algorithm. Returns true on success.
bool run();
diff --git a/llvm/include/llvm/Transforms/Utils/LoopPeel.h b/llvm/include/llvm/Transforms/Utils/LoopPeel.h
index 49dbc9aa1f2a9..73d3b84b36899 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopPeel.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopPeel.h
@@ -31,7 +31,7 @@ bool canPeelLastIteration(const Loop &L, ScalarEvolution &SE);
/// off the last \p PeelCount iterations from \p L (canPeelLastIteration must be
/// true for \p L), otherwise peel off the first \p PeelCount iterations.
bool peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
- ScalarEvolution *SE, DominatorTree &DT, AssumptionCache *AC,
+ ScalarEvolution *SE, const TargetTransformInfo *TTI, DominatorTree &DT, AssumptionCache *AC,
bool PreserveLCSSA, ValueToValueMapTy &VMap);
TargetTransformInfo::PeelingPreferences
diff --git a/llvm/include/llvm/Transforms/Utils/LoopSimplify.h b/llvm/include/llvm/Transforms/Utils/LoopSimplify.h
index 8f3fa1f2b18ef..230fac2f8a0ac 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopSimplify.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopSimplify.h
@@ -40,6 +40,7 @@
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
+#include "llvm/Transforms/Utils/Local.h"
namespace llvm {
@@ -64,7 +65,7 @@ class LoopSimplifyPass : public PassInfoMixin<LoopSimplifyPass> {
/// analyses if they're non-null, and LCSSA if \c PreserveLCSSA is true.
LLVM_ABI bool simplifyLoop(Loop *L, DominatorTree *DT, LoopInfo *LI,
ScalarEvolution *SE, AssumptionCache *AC,
- MemorySSAUpdater *MSSAU, bool PreserveLCSSA);
+ MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI, bool PreserveLCSSA);
} // end namespace llvm
diff --git a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
index e706a6f83b1e7..b56646af58313 100644
--- a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
+++ b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
@@ -234,6 +234,7 @@ class InductiveRangeCheck {
class InductiveRangeCheckElimination {
ScalarEvolution &SE;
BranchProbabilityInfo *BPI;
+ TargetTransformInfo &TTI;
DominatorTree &DT;
LoopInfo &LI;
@@ -248,9 +249,9 @@ class InductiveRangeCheckElimination {
public:
InductiveRangeCheckElimination(ScalarEvolution &SE,
- BranchProbabilityInfo *BPI, DominatorTree &DT,
+ BranchProbabilityInfo *BPI, TargetTransformInfo &TTI, DominatorTree &DT,
LoopInfo &LI, GetBFIFunc GetBFI = std::nullopt)
- : SE(SE), BPI(BPI), DT(DT), LI(LI), GetBFI(GetBFI) {}
+ : SE(SE), BPI(BPI), TTI(TTI), DT(DT), LI(LI), GetBFI(GetBFI) {}
bool run(Loop *L, function_ref<void(Loop *, bool)> LPMAddNewLoop);
};
@@ -900,6 +901,7 @@ IntersectUnsignedRange(ScalarEvolution &SE,
PreservedAnalyses IRCEPass::run(Function &F, FunctionAnalysisManager &AM) {
auto &DT = AM.getResult<DominatorTreeAnalysis>(F);
+ auto &TTI = AM.getResult<TargetIRAnalysis>(F);
LoopInfo &LI = AM.getResult<LoopAnalysis>(F);
// There are no loops in the function. Return before computing other expensive
// analyses.
@@ -913,14 +915,14 @@ PreservedAnalyses IRCEPass::run(Function &F, FunctionAnalysisManager &AM) {
auto getBFI = [&F, &AM ]()->BlockFrequencyInfo & {
return AM.getResult<BlockFrequencyAnalysis>(F);
};
- InductiveRangeCheckElimination IRCE(SE, &BPI, DT, LI, { getBFI });
+ InductiveRangeCheckElimination IRCE(SE, &BPI, TTI, DT, LI, { getBFI });
bool Changed = false;
{
bool CFGChanged = false;
for (const auto &L : LI) {
CFGChanged |= simplifyLoop(L, &DT, &LI, &SE, nullptr, nullptr,
- /*PreserveLCSSA=*/false);
+ &TTI, /*PreserveLCSSA=*/false);
Changed |= formLCSSARecursively(*L, DT, &LI, &SE);
}
Changed |= CFGChanged;
@@ -1080,7 +1082,7 @@ bool InductiveRangeCheckElimination::run(
return false;
}
- LoopConstrainer LC(*L, LI, LPMAddNewLoop, LS, SE, DT,
+ LoopConstrainer LC(*L, LI, LPMAddNewLoop, LS, SE, TTI, DT,
SafeIterRange->getBegin()->getType(), *MaybeSR);
if (LC.run()) {
diff --git a/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp b/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp
index 73f1942849ac2..756a6d81f666a 100644
--- a/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp
@@ -285,7 +285,7 @@ static BranchInst *findSplitCandidate(const Loop &L, ScalarEvolution &SE,
}
static bool splitLoopBound(Loop &L, DominatorTree &DT, LoopInfo &LI,
- ScalarEvolution &SE, LPMUpdater &U) {
+ ScalarEvolution &SE, LPMUpdater &U, const TargetTransformInfo *TTI) {
ConditionInfo SplitCandidateCond;
ConditionInfo ExitingCond;
@@ -460,8 +460,8 @@ static bool splitLoopBound(Loop &L, DominatorTree &DT, LoopInfo &LI,
SE.forgetLoop(&L);
// Canonicalize loops.
- simplifyLoop(&L, &DT, &LI, &SE, nullptr, nullptr, true);
- simplifyLoop(PostLoop, &DT, &LI, &SE, nullptr, nullptr, true);
+ simplifyLoop(&L, &DT, &LI, &SE, nullptr, nullptr, TTI, true);
+ simplifyLoop(PostLoop, &DT, &LI, &SE, nullptr, nullptr, TTI, true);
// Add new post-loop to loop pass manager.
U.addSiblingLoops(PostLoop);
@@ -478,7 +478,9 @@ PreservedAnalyses LoopBoundSplitPass::run(Loop &L, LoopAnalysisManager &AM,
LLVM_DEBUG(dbgs() << "Spliting bound of loop in " << F.getName() << ": " << L
<< "\n");
- if (!splitLoopBound(L, AR.DT, AR.LI, AR.SE, U))
+ auto &FAM = AM.getResult<FunctionAnalysisManagerLoopProxy>(L, AR);
+ auto TTI = FAM.getCachedResult<TargetIRAnalysis>(F);
+ if (!splitLoopBound(L, AR.DT, AR.LI, AR.SE, U, TTI))
return PreservedAnalyses::all();
assert(AR.DT.verify(DominatorTree::VerificationLevel::Fast));
diff --git a/llvm/lib/Transforms/Scalar/LoopFuse.cpp b/llvm/lib/Transforms/Scalar/LoopFuse.cpp
index d6bd92d520e28..c92accd143d70 100644
--- a/llvm/lib/Transforms/Scalar/LoopFuse.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopFuse.cpp
@@ -791,7 +791,7 @@ struct LoopFuser {
ValueToValueMapTy VMap;
FC0.Peeled =
- peelLoop(FC0.L, PeelCount, false, &LI, &SE, DT, &AC, true, VMap);
+ peelLoop(FC0.L, PeelCount, false, &LI, &SE, &TTI, DT, &AC, true, VMap);
if (FC0.Peeled) {
LLVM_DEBUG(dbgs() << "Done Peeling\n");
@@ -2084,7 +2084,7 @@ PreservedAnalyses LoopFusePass::run(Function &F, FunctionAnalysisManager &AM) {
bool Changed = false;
for (auto &L : LI) {
Changed |=
- simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, false /* PreserveLCSSA */);
+ simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, &TTI, false /* PreserveLCSSA */);
}
if (Changed)
PDT.recalculate(F);
diff --git a/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp b/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp
index 6bdf76f789a49..ec89f55b1b78c 100644
--- a/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp
@@ -645,7 +645,7 @@ static bool eliminateLoadsAcrossLoops(Function &F, LoopInfo &LI,
BlockFrequencyInfo *BFI,
ProfileSummaryInfo *PSI,
ScalarEvolution *SE, AssumptionCache *AC,
- LoopAccessInfoManager &LAIs) {
+ LoopAccessInfoManager &LAIs, const TargetTransformInfo *TTI) {
// Build up a worklist of inner-loops to transform to avoid iterator
// invalidation.
// FIXME: This logic comes from other passes that actually change the loop
@@ -657,7 +657,7 @@ static bool eliminateLoadsAcrossLoops(Function &F, LoopInfo &LI,
for (Loop *TopLevelLoop : LI)
for (Loop *L : depth_first(TopLevelLoop)) {
- Changed |= simplifyLoop(L, &DT, &LI, SE, AC, /*MSSAU*/ nullptr, false);
+ Changed |= simplifyLoop(L, &DT, &LI, SE, AC, /*MSSAU*/ nullptr, TTI, false);
// We only handle inner-most loops.
if (L->isInnermost())
Worklist.push_back(L);
@@ -692,8 +692,9 @@ PreservedAnalyses LoopLoadEliminationPass::run(Function &F,
auto *BFI = (PSI && PSI->hasProfileSummary()) ?
&AM.getResult<BlockFrequencyAnalysis>(F) : nullptr;
LoopAccessInfoManager &LAIs = AM.getResult<LoopAccessAnalysis>(F);
+ auto &TTI = AM.getResult<TargetIRAnalysis>(F);
- bool Changed = eliminateLoadsAcrossLoops(F, LI, DT, BFI, PSI, &SE, &AC, LAIs);
+ bool Changed = eliminateLoadsAcrossLoops(F, LI, DT, BFI, PSI, &SE, &AC, LAIs, &TTI);
if (!Changed)
return PreservedAnalyses::all();
diff --git a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
index a22d84dcf014d..8882acf63f11d 100644
--- a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
@@ -1314,7 +1314,7 @@ tryToUnrollLoop(Loop *L, DominatorTree &DT, LoopInfo *LI, ScalarEvolution &SE,
});
ValueToValueMapTy VMap;
- if (peelLoop(L, PP.PeelCount, PP.PeelLast, LI, &SE, DT, &AC, PreserveLCSSA,
+ if (peelLoop(L, PP.PeelCount, PP.PeelLast, LI, &SE, &TTI, DT, &AC, PreserveLCSSA,
VMap)) {
simplifyLoopAfterUnroll(L, true, LI, &SE, &DT, &AC, &TTI, nullptr);
// If the loop was peeled, we already "used up" the profile information
@@ -1624,7 +1624,7 @@ PreservedAnalyses LoopUnrollPass::run(Function &F,
// unrolled.
for (const auto &L : LI) {
Changed |=
- simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, false /* PreserveLCSSA */);
+ simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, &TTI, false /* PreserveLCSSA */);
Changed |= formLCSSARecursively(*L, DT, &LI, &SE);
}
diff --git a/llvm/lib/Transforms/Utils/LoopConstrainer.cpp b/llvm/lib/Transforms/Utils/LoopConstrainer.cpp
index 8f103153059e8..5afb2a719dc65 100644
--- a/llvm/lib/Transforms/Utils/LoopConstrainer.cpp
+++ b/llvm/lib/Transforms/Utils/LoopConstrainer.cpp
@@ -460,8 +460,8 @@ static void DisableAllLoopOptsOnLoop(Loop &L) {
LoopConstrainer::LoopConstrainer(Loop &L, LoopInfo &LI,
function_ref<void(Loop *, bool)> LPMAddNewLoop,
const LoopStructure &LS, ScalarEvolution &SE,
- DominatorTree &DT, Type *T, SubRanges SR)
- : F(*L.getHeader()->getParent()), Ctx(L.getHeader()->getContext()), SE(SE),
+ TargetTransformInfo &TTI, DominatorTree &DT, Type *T, SubRanges SR)
+ : F(*L.getHeader()->getParent()), Ctx(L.getHeader()->getContext()), SE(SE), TTI(TTI),
DT(DT), LI(LI), LPMAddNewLoop(LPMAddNewLoop), OriginalLoop(L), RangeTy(T),
MainLoopStructure(LS), SR(SR) {}
@@ -872,7 +872,7 @@ bool LoopConstrainer::run() {
// This function canonicalizes the loop into Loop-Simplify and LCSSA forms.
auto CanonicalizeLoop = [&](Loop *L, bool IsOriginalLoop) {
formLCSSARecursively(*L, DT, &LI, &SE);
- simplifyLoop(L, &DT, &LI, &SE, nullptr, nullptr, true);
+ simplifyLoop(L, &DT, &LI, &SE, nullptr, nullptr, &TTI, true);
// Pre/post loops are slow paths, we do not need to perform any loop
// optimizations on them.
if (!IsOriginalLoop)
diff --git a/llvm/lib/Transforms/Utils/LoopPeel.cpp b/llvm/lib/Transforms/Utils/LoopPeel.cpp
index 27e70c5ddc0fc..7082b4e58bb2d 100644
--- a/llvm/lib/Transforms/Utils/LoopPeel.cpp
+++ b/llvm/lib/Transforms/Utils/LoopPeel.cpp
@@ -351,7 +351,7 @@ bool llvm::canPeelLastIteration(const Loop &L, ScalarEvolution &SE) {
m_BasicBlock(Succ1), m_BasicBlock(Succ2))) &&
((Pred == CmpInst::ICMP_EQ && Succ2 == L.getHeader()) ||
(Pred == CmpInst::ICMP_NE && Succ1 == L.getHeader())) &&
- Bound->getType()->isIntegerTy() &&
+ Bound->getType()->isIntegerTy() &&
SE.isLoopInvariant(SE.getSCEV(Bound), &L) &&
match(SE.getSCEV(Inc),
m_scev_AffineAddRec(m_SCEV(), m_scev_One(), m_SpecificLoop(&L)));
@@ -1030,7 +1030,7 @@ llvm::gatherPeelingPreferences(Loop *L, ScalarEvolution &SE,
/// for the bulk of dynamic execution, can be further simplified by scalar
/// optimizations.
bool llvm::peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
- ScalarEvolution *SE, DominatorTree &DT, AssumptionCache *AC,
+ ScalarEvolution *SE, const TargetTransformInfo *TTI, DominatorTree &DT, AssumptionCache *AC,
bool PreserveLCSSA, ValueToValueMapTy &LVMap) {
assert(PeelCount > 0 && "Attempt to peel out zero iterations?");
assert(canPeel(L) && "Attempt to peel a loop which is not peelable?");
@@ -1308,7 +1308,7 @@ bool llvm::peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
#endif
// FIXME: Incrementally update loop-simplify
- simplifyLoop(L, &DT, LI, SE, AC, nullptr, PreserveLCSSA);
+ simplifyLoop(L, &DT, LI, SE, AC, nullptr, TTI, PreserveLCSSA);
NumPeeled++;
NumPeeledEnd += PeelLast;
diff --git a/llvm/lib/Transforms/Utils/LoopSimplify.cpp b/llvm/lib/Transforms/Utils/LoopSimplify.cpp
index 61ffb49a8c010..77013d36185bd 100644
--- a/llvm/lib/Transforms/Utils/LoopSimplify.cpp
+++ b/llvm/lib/Transforms/Utils/LoopSimplify.cpp
@@ -66,7 +66,6 @@
#include "llvm/Support/raw_ostream.h"
#include "llvm/Transforms/Utils.h"
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
-#include "llvm/Transforms/Utils/Local.h"
#include "llvm/Transforms/Utils/LoopUtils.h"
using namespace llvm;
@@ -475,7 +474,7 @@ static BasicBlock *insertUniqueBackedgeBlock(Loop *L, BasicBlock *Preheader,
static bool simplifyOneLoop(Loop *L, SmallVectorImpl<Loop *> &Worklist,
DominatorTree *DT, LoopInfo *LI,
ScalarEvolution *SE, AssumptionCache *AC,
- MemorySSAUpdater *MSSAU, bool PreserveLCSSA) {
+ MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI, bool PreserveLCSSA) {
bool Changed = false;
if (MSSAU && VerifyMemorySSA)
MSSAU->getMemorySSA()->verifyMemorySSA();
@@ -656,7 +655,7 @@ static bool simplifyOneLoop(Loop *L, SmallVectorImpl<Loop *> &Worklist,
// The block has now been cleared of all instructions except for
// a comparison and a conditional branch. SimplifyCFG may be able
// to fold it now.
- if (!foldBranchToCommonDest(BI, /*DTU=*/nullptr, MSSAU))
+ if (!foldBranchToCommonDest(BI, /*DTU=*/nullptr, MSSAU, TTI))
continue;
// Success. The block is now dead, so remove it from the loop,
@@ -696,7 +695,7 @@ static bool simplifyOneLoop(Loop *L, SmallVectorImpl<Loop *> &Worklist,
bool llvm::simplifyLoop(Loop *L, DominatorTree *DT, LoopInfo *LI,
ScalarEvolution *SE, AssumptionCache *AC,
- MemorySSAUpdater *MSSAU, bool PreserveLCSSA) {
+ MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI, bool PreserveLCSSA) {
bool Changed = false;
#ifndef NDEBUG
@@ -724,7 +723,7 @@ bool llvm::simplifyLoop(Loop *L, DominatorTree *DT, LoopInfo *LI,
while (!Worklist.empty())
Changed |= simplifyOneLoop(Worklist.pop_back_val(), Worklist, DT, LI, SE,
- AC, MSSAU, PreserveLCSSA);
+ AC, MSSAU, TTI, PreserveLCSSA);
// Changing exit conditions for blocks may affect exit counts of this loop and
// any of its parents, so we must invalidate the entire subtree if we've made
@@ -747,6 +746,7 @@ namespace {
void getAnalysisUsage(AnalysisUsage &AU) const override {
AU.addRequired<AssumptionCacheTracker>();
+ AU.addRequired<TargetTransformInfoWrapperPass>();
// We need loop information to identify the loops...
AU.addRequired<DominatorTreeWrapperPass>();
@@ -777,6 +777,7 @@ INITIALIZE_PASS_BEGIN(LoopSimplify, "loop-simplify",
INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass)
+INITIALIZE_PASS_DEPENDENCY(TargetTransformInfoWrapperPass)
INITIALIZE_PASS_END(LoopSimplify, "loop-simplify", "Canonicalize natural loops",
false, false)
@@ -791,6 +792,7 @@ bool LoopSimplify::runOnFunction(Function &F) {
bool Changed = false;
LoopInfo *LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
DominatorTree *DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
+ auto &TTI = getAnalysis<TargetTransformInfoWrapperPass>().getTTI(F);
auto *SEWP = getAnalysisIfAvailable<ScalarEvolutionWrapperPass>();
ScalarEvolution *SE = SEWP ? &SEWP->getSE() : nullptr;
AssumptionCache *AC =
@@ -807,7 +809,7 @@ bool LoopSimplify::runOnFunction(Function &F) {
// Simplify each loop nest in the function.
for (auto *L : *LI)
- Changed |= simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), PreserveLCSSA);
+ Changed |= simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), &TTI, PreserveLCSSA);
#ifndef NDEBUG
if (PreserveLCSSA) {
@@ -832,13 +834,14 @@ PreservedAnalyses LoopSimplifyPass::run(Function &F,
auto *MSSA = &MSSAAnalysis->getMSSA();
MSSAU = std::make_unique<MemorySSAUpdater>(MSSA);
}
+ auto &TTI = AM.getResult<TargetIRAnalysis>(F);
// Note that we don't preserve LCSSA in the new PM, if you need it run LCSSA
// after simplifying the loops. MemorySSA is preserved if it exists.
for (auto *L : *LI)
Changed |=
- simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), /*PreserveLCSSA*/ false);
+ simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), &TTI, /*PreserveLCSSA*/ false);
if (!Changed)
return PreservedAnalyses::all();
diff --git a/llvm/lib/Transforms/Utils/LoopUnroll.cpp b/llvm/lib/Transforms/Utils/LoopUnroll.cpp
index 86b268de43cf6..44b9da2c0da5a 100644
--- a/llvm/lib/Transforms/Utils/LoopUnroll.cpp
+++ b/llvm/lib/Transforms/Utils/LoopUnroll.cpp
@@ -1078,11 +1078,11 @@ llvm::UnrollLoop(Loop *L, UnrollLoopOptions ULO, LoopInfo *LI,
// TODO: That potentially might be compile-time expensive. We should try
// to fix the loop-simplified form incrementally.
- simplifyLoop(OuterL, DT, LI, SE, AC, nullptr, PreserveLCSSA);
+ simplifyLoop(OuterL, DT, LI, SE, AC, nullptr, TTI, PreserveLCSSA);
} else {
// Simplify loops for which we might've broken loop-simplify form.
for (Loop *SubLoop : LoopsToSimplify)
- simplifyLoop(SubLoop, DT, LI, SE, AC, nullptr, PreserveLCSSA);
+ simplifyLoop(SubLoop, DT, LI, SE, AC, nullptr, TTI, PreserveLCSSA);
}
return CompletelyUnroll ? LoopUnrollResult::FullyUnrolled
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 907839711a39c..bb0586d2b215e 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -10426,7 +10426,7 @@ LoopVectorizeResult LoopVectorizePass::runImpl(Function &F) {
// vectorized.
for (const auto &L : *LI)
Changed |= CFGChanged |=
- simplifyLoop(L, DT, LI, SE, AC, nullptr, false /* PreserveLCSSA */);
+ simplifyLoop(L, DT, LI, SE, AC, nullptr, TTI, false /* PreserveLCSSA */);
// Build up a worklist of inner-loops to vectorize. This is necessary as
// the act of vectorizing or partially unrolling a loop creates new loops
diff --git a/llvm/test/Transforms/LoopSimplify/merge-exits.ll b/llvm/test/Transforms/LoopSimplify/merge-exits.ll
index 10e634fd47a1a..4eef7e71cc9c0 100644
--- a/llvm/test/Transforms/LoopSimplify/merge-exits.ll
+++ b/llvm/test/Transforms/LoopSimplify/merge-exits.ll
@@ -84,33 +84,28 @@ define float @merge_branches_profile_metadata(ptr %pTmp1, ptr %peakWeight, i32 %
; CHECK-NEXT: entry:
; CHECK-NEXT: [[T0:%.*]] = load float, ptr [[PEAKWEIGHT:%.*]], align 4
; CHECK-NEXT: [[T11:%.*]] = add i32 [[BANDEDGEINDEX:%.*]], -1
-; CHECK-NEXT: [[T121:%.*]] = icmp sgt i32 [[T11]], 0
-; CHECK-NEXT: br i1 [[T121]], label [[BB_LR_PH:%.*]], label [[BB3:%.*]], !prof [[PROF0:![0-9]+]]
-; CHECK: bb.lr.ph:
-; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[T11]] to i64
+; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[T11]], i32 0)
+; CHECK-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[SMAX]] to i64
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb:
-; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[BB_LR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[BB]] ]
-; CHECK-NEXT: [[DISTERBHI_04:%.*]] = phi float [ 0.000000e+00, [[BB_LR_PH]] ], [ [[T4:%.*]], [[BB]] ]
-; CHECK-NEXT: [[PEAKCOUNT_02:%.*]] = phi float [ [[T0]], [[BB_LR_PH]] ], [ [[T9:%.*]], [[BB]] ]
-; CHECK-NEXT: [[T2:%.*]] = getelementptr float, ptr [[PTMP1:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[T2:%.*]] = getelementptr float, ptr [[PTMP1:%.*]], i64 [[INDVARS_IV:%.*]]
; CHECK-NEXT: [[T3:%.*]] = load float, ptr [[T2]], align 4
-; CHECK-NEXT: [[T4]] = fadd float [[T3]], [[DISTERBHI_04]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-NEXT: [[T4:%.*]] = fadd float [[T3]], [[DISTERBHI_04:%.*]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[T7:%.*]] = getelementptr float, ptr [[PEAKWEIGHT]], i64 [[INDVARS_IV_NEXT]]
; CHECK-NEXT: [[T8:%.*]] = load float, ptr [[T7]], align 4
-; CHECK-NEXT: [[T9]] = fadd float [[T8]], [[PEAKCOUNT_02]]
-; CHECK-NEXT: [[T10:%.*]] = fcmp olt float [[T4]], 2.500000e+00
-; CHECK-NEXT: [[T12:%.*]] = icmp sgt i64 [[TMP0]], [[INDVARS_IV_NEXT]]
-; CHECK-NEXT: [[OR_COND:%.*]] = and i1 [[T10]], [[T12]]
-; CHECK-NEXT: br i1 [[OR_COND]], label [[BB]], label [[BB1_BB3_CRIT_EDGE:%.*]], !prof [[PROF1:![0-9]+]]
-; CHECK: bb1.bb3_crit_edge:
-; CHECK-NEXT: [[T4_LCSSA:%.*]] = phi float [ [[T4]], [[BB]] ]
-; CHECK-NEXT: [[T9_LCSSA:%.*]] = phi float [ [[T9]], [[BB]] ]
-; CHECK-NEXT: br label [[BB3]]
+; CHECK-NEXT: [[T9:%.*]] = fadd float [[T8]], [[PEAKCOUNT_02:%.*]]
+; CHECK-NEXT: [[T10:%.*]] = fcmp uge float [[T4]], 2.500000e+00
+; CHECK-NEXT: br i1 [[T10]], label [[BB3:%.*]], label [[BB]], !prof [[PROF0:![0-9]+]]
+; CHECK: bb2:
+; CHECK-NEXT: [[INDVARS_IV]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT]], [[BB1:%.*]] ]
+; CHECK-NEXT: [[DISTERBHI_04]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[T4]], [[BB1]] ]
+; CHECK-NEXT: [[PEAKCOUNT_02]] = phi float [ [[T0]], [[ENTRY]] ], [ [[T9]], [[BB1]] ]
+; CHECK-NEXT: [[EXITCOND:%.*]] = icmp ne i64 [[INDVARS_IV]], [[WIDE_TRIP_COUNT]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[BB1]], label [[BB3]]
; CHECK: bb3:
-; CHECK-NEXT: [[PEAKCOUNT_0_LCSSA:%.*]] = phi float [ [[T9_LCSSA]], [[BB1_BB3_CRIT_EDGE]] ], [ [[T0]], [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[DISTERBHI_0_LCSSA:%.*]] = phi float [ [[T4_LCSSA]], [[BB1_BB3_CRIT_EDGE]] ], [ 0.000000e+00, [[ENTRY]] ]
+; CHECK-NEXT: [[PEAKCOUNT_0_LCSSA:%.*]] = phi float [ [[PEAKCOUNT_02]], [[BB]] ], [ [[T9]], [[BB1]] ]
+; CHECK-NEXT: [[DISTERBHI_0_LCSSA:%.*]] = phi float [ [[DISTERBHI_04]], [[BB]] ], [ [[T4]], [[BB1]] ]
; CHECK-NEXT: [[T13:%.*]] = fdiv float [[PEAKCOUNT_0_LCSSA]], [[DISTERBHI_0_LCSSA]]
; CHECK-NEXT: ret float [[T13]]
;
diff --git a/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll b/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll
index 227c45fa80a1d..e2a9984a0cda0 100644
--- a/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll
+++ b/llvm/test/Transforms/LoopUnroll/runtime-loop-multiple-exits.ll
@@ -2967,9 +2967,14 @@ define void @non_loop(i32 %arg) {
; CHECK: preheader:
; CHECK-NEXT: br label %header
; CHECK: header:
+; CHECK-NEXT: br i1 true, label %latchExit, label %latch
+; CHECK: latch:
+; CHECK-NEXT: br label %latchExit
+; CHECK: latchExit:
+; CHECK-NEXT: %i2.ph = phi i32 [ %arg, %header ], [ -1, %latch ]
; CHECK-NEXT: br label %returnblock
; CHECK: returnblock:
-; CHECK-NEXT: %i2 = phi i32 [ -1, %entry ], [ %arg, %header ]
+; CHECK-NEXT: %i2 = phi i32 [ -1, %entry ], [ %i2.ph, %latchExit ]
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LoopUnroll/scevunroll.ll b/llvm/test/Transforms/LoopUnroll/scevunroll.ll
index aa33b979bf17e..b6b14e365cc1d 100644
--- a/llvm/test/Transforms/LoopUnroll/scevunroll.ll
+++ b/llvm/test/Transforms/LoopUnroll/scevunroll.ll
@@ -310,16 +310,16 @@ define void @nsw_latch(ptr %a) nounwind {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: [[B_03:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TOBOOL:%.*]] = icmp eq i32 [[B_03]], 0
-; CHECK-NEXT: [[ADD]] = add nuw nsw i32 [[B_03]], 8
-; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = xor i1 [[TOBOOL]], true
-; CHECK-NEXT: [[TMP0:%.*]] = select i1 [[TOBOOL]], i32 [[B_03]], i32 [[B_03]]
-; CHECK-NEXT: [[TMP1:%.*]] = select i1 [[TOBOOL]], i32 0, i32 1
-; CHECK-NEXT: [[OR_COND:%.*]] = select i1 [[TOBOOL_NOT]], i1 true, i1 false
-; CHECK-NEXT: br i1 [[OR_COND]], label [[RETURN:%.*]], label [[FOR_BODY]]
+; CHECK-NEXT: br label [[FOR_COND:%.*]]
+; CHECK: for.cond:
+; CHECK-NEXT: br i1 false, label [[RETURN:%.*]], label [[FOR_BODY_1:%.*]]
+; CHECK: for.body.1:
+; CHECK-NEXT: br i1 false, label [[FOR_COND_1:%.*]], label [[RETURN]]
+; CHECK: for.cond.1:
+; CHECK-NEXT: br label [[RETURN]]
; CHECK: return:
-; CHECK-NEXT: [[B_03_LCSSA:%.*]] = phi i32 [ [[TMP0]], [[FOR_BODY]] ]
+; CHECK-NEXT: [[B_03_LCSSA:%.*]] = phi i32 [ 0, [[FOR_COND]] ], [ 8, [[FOR_BODY_1]] ], [ 0, [[FOR_COND_1]] ]
+; CHECK-NEXT: [[RETVAL_0:%.*]] = phi i32 [ 0, [[FOR_COND]] ], [ 1, [[FOR_BODY_1]] ], [ 0, [[FOR_COND_1]] ]
; CHECK-NEXT: store i32 [[B_03_LCSSA]], ptr [[A:%.*]], align 4
; CHECK-NEXT: ret void
;
>From 27e113a8d329430db0bbe65bfde82ad599326b31 Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Fri, 4 Jul 2025 16:40:56 -0700
Subject: [PATCH 6/8] format
---
.../Scalar/InductiveRangeCheckElimination.cpp | 9 +++++----
llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp | 3 ++-
llvm/lib/Transforms/Scalar/LoopFuse.cpp | 4 ++--
.../Transforms/Scalar/LoopLoadElimination.cpp | 16 ++++++++--------
llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp | 8 ++++----
llvm/lib/Transforms/Utils/LoopConstrainer.cpp | 9 +++++----
llvm/lib/Transforms/Utils/LoopPeel.cpp | 5 +++--
llvm/lib/Transforms/Utils/LoopSimplify.cpp | 15 +++++++++------
llvm/lib/Transforms/Vectorize/LoopVectorize.cpp | 4 ++--
9 files changed, 40 insertions(+), 33 deletions(-)
diff --git a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
index b56646af58313..892bfc8c09ed6 100644
--- a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
+++ b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
@@ -249,7 +249,8 @@ class InductiveRangeCheckElimination {
public:
InductiveRangeCheckElimination(ScalarEvolution &SE,
- BranchProbabilityInfo *BPI, TargetTransformInfo &TTI, DominatorTree &DT,
+ BranchProbabilityInfo *BPI,
+ TargetTransformInfo &TTI, DominatorTree &DT,
LoopInfo &LI, GetBFIFunc GetBFI = std::nullopt)
: SE(SE), BPI(BPI), TTI(TTI), DT(DT), LI(LI), GetBFI(GetBFI) {}
@@ -915,14 +916,14 @@ PreservedAnalyses IRCEPass::run(Function &F, FunctionAnalysisManager &AM) {
auto getBFI = [&F, &AM ]()->BlockFrequencyInfo & {
return AM.getResult<BlockFrequencyAnalysis>(F);
};
- InductiveRangeCheckElimination IRCE(SE, &BPI, TTI, DT, LI, { getBFI });
+ InductiveRangeCheckElimination IRCE(SE, &BPI, TTI, DT, LI, {getBFI});
bool Changed = false;
{
bool CFGChanged = false;
for (const auto &L : LI) {
- CFGChanged |= simplifyLoop(L, &DT, &LI, &SE, nullptr, nullptr,
- &TTI, /*PreserveLCSSA=*/false);
+ CFGChanged |= simplifyLoop(L, &DT, &LI, &SE, nullptr, nullptr, &TTI,
+ /*PreserveLCSSA=*/false);
Changed |= formLCSSARecursively(*L, DT, &LI, &SE);
}
Changed |= CFGChanged;
diff --git a/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp b/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp
index 756a6d81f666a..25d77563e5887 100644
--- a/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopBoundSplit.cpp
@@ -285,7 +285,8 @@ static BranchInst *findSplitCandidate(const Loop &L, ScalarEvolution &SE,
}
static bool splitLoopBound(Loop &L, DominatorTree &DT, LoopInfo &LI,
- ScalarEvolution &SE, LPMUpdater &U, const TargetTransformInfo *TTI) {
+ ScalarEvolution &SE, LPMUpdater &U,
+ const TargetTransformInfo *TTI) {
ConditionInfo SplitCandidateCond;
ConditionInfo ExitingCond;
diff --git a/llvm/lib/Transforms/Scalar/LoopFuse.cpp b/llvm/lib/Transforms/Scalar/LoopFuse.cpp
index c92accd143d70..3a48d8328c027 100644
--- a/llvm/lib/Transforms/Scalar/LoopFuse.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopFuse.cpp
@@ -2083,8 +2083,8 @@ PreservedAnalyses LoopFusePass::run(Function &F, FunctionAnalysisManager &AM) {
// LoopSimplify pass as a dependency.
bool Changed = false;
for (auto &L : LI) {
- Changed |=
- simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, &TTI, false /* PreserveLCSSA */);
+ Changed |= simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, &TTI,
+ false /* PreserveLCSSA */);
}
if (Changed)
PDT.recalculate(F);
diff --git a/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp b/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp
index ec89f55b1b78c..8df48f0f95a0b 100644
--- a/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp
@@ -640,12 +640,10 @@ class LoadEliminationForLoop {
} // end anonymous namespace
-static bool eliminateLoadsAcrossLoops(Function &F, LoopInfo &LI,
- DominatorTree &DT,
- BlockFrequencyInfo *BFI,
- ProfileSummaryInfo *PSI,
- ScalarEvolution *SE, AssumptionCache *AC,
- LoopAccessInfoManager &LAIs, const TargetTransformInfo *TTI) {
+static bool eliminateLoadsAcrossLoops(
+ Function &F, LoopInfo &LI, DominatorTree &DT, BlockFrequencyInfo *BFI,
+ ProfileSummaryInfo *PSI, ScalarEvolution *SE, AssumptionCache *AC,
+ LoopAccessInfoManager &LAIs, const TargetTransformInfo *TTI) {
// Build up a worklist of inner-loops to transform to avoid iterator
// invalidation.
// FIXME: This logic comes from other passes that actually change the loop
@@ -657,7 +655,8 @@ static bool eliminateLoadsAcrossLoops(Function &F, LoopInfo &LI,
for (Loop *TopLevelLoop : LI)
for (Loop *L : depth_first(TopLevelLoop)) {
- Changed |= simplifyLoop(L, &DT, &LI, SE, AC, /*MSSAU*/ nullptr, TTI, false);
+ Changed |=
+ simplifyLoop(L, &DT, &LI, SE, AC, /*MSSAU*/ nullptr, TTI, false);
// We only handle inner-most loops.
if (L->isInnermost())
Worklist.push_back(L);
@@ -694,7 +693,8 @@ PreservedAnalyses LoopLoadEliminationPass::run(Function &F,
LoopAccessInfoManager &LAIs = AM.getResult<LoopAccessAnalysis>(F);
auto &TTI = AM.getResult<TargetIRAnalysis>(F);
- bool Changed = eliminateLoadsAcrossLoops(F, LI, DT, BFI, PSI, &SE, &AC, LAIs, &TTI);
+ bool Changed =
+ eliminateLoadsAcrossLoops(F, LI, DT, BFI, PSI, &SE, &AC, LAIs, &TTI);
if (!Changed)
return PreservedAnalyses::all();
diff --git a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
index 8882acf63f11d..3808fbba9f9e2 100644
--- a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp
@@ -1314,8 +1314,8 @@ tryToUnrollLoop(Loop *L, DominatorTree &DT, LoopInfo *LI, ScalarEvolution &SE,
});
ValueToValueMapTy VMap;
- if (peelLoop(L, PP.PeelCount, PP.PeelLast, LI, &SE, &TTI, DT, &AC, PreserveLCSSA,
- VMap)) {
+ if (peelLoop(L, PP.PeelCount, PP.PeelLast, LI, &SE, &TTI, DT, &AC,
+ PreserveLCSSA, VMap)) {
simplifyLoopAfterUnroll(L, true, LI, &SE, &DT, &AC, &TTI, nullptr);
// If the loop was peeled, we already "used up" the profile information
// we had, so we don't want to unroll or peel again.
@@ -1623,8 +1623,8 @@ PreservedAnalyses LoopUnrollPass::run(Function &F,
// will simplify all loops, regardless of whether anything end up being
// unrolled.
for (const auto &L : LI) {
- Changed |=
- simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, &TTI, false /* PreserveLCSSA */);
+ Changed |= simplifyLoop(L, &DT, &LI, &SE, &AC, nullptr, &TTI,
+ false /* PreserveLCSSA */);
Changed |= formLCSSARecursively(*L, DT, &LI, &SE);
}
diff --git a/llvm/lib/Transforms/Utils/LoopConstrainer.cpp b/llvm/lib/Transforms/Utils/LoopConstrainer.cpp
index 5afb2a719dc65..f633ec1e47dbe 100644
--- a/llvm/lib/Transforms/Utils/LoopConstrainer.cpp
+++ b/llvm/lib/Transforms/Utils/LoopConstrainer.cpp
@@ -460,10 +460,11 @@ static void DisableAllLoopOptsOnLoop(Loop &L) {
LoopConstrainer::LoopConstrainer(Loop &L, LoopInfo &LI,
function_ref<void(Loop *, bool)> LPMAddNewLoop,
const LoopStructure &LS, ScalarEvolution &SE,
- TargetTransformInfo &TTI, DominatorTree &DT, Type *T, SubRanges SR)
- : F(*L.getHeader()->getParent()), Ctx(L.getHeader()->getContext()), SE(SE), TTI(TTI),
- DT(DT), LI(LI), LPMAddNewLoop(LPMAddNewLoop), OriginalLoop(L), RangeTy(T),
- MainLoopStructure(LS), SR(SR) {}
+ TargetTransformInfo &TTI, DominatorTree &DT,
+ Type *T, SubRanges SR)
+ : F(*L.getHeader()->getParent()), Ctx(L.getHeader()->getContext()), SE(SE),
+ TTI(TTI), DT(DT), LI(LI), LPMAddNewLoop(LPMAddNewLoop), OriginalLoop(L),
+ RangeTy(T), MainLoopStructure(LS), SR(SR) {}
void LoopConstrainer::cloneLoop(LoopConstrainer::ClonedLoop &Result,
const char *Tag) const {
diff --git a/llvm/lib/Transforms/Utils/LoopPeel.cpp b/llvm/lib/Transforms/Utils/LoopPeel.cpp
index 7082b4e58bb2d..8af7911476d97 100644
--- a/llvm/lib/Transforms/Utils/LoopPeel.cpp
+++ b/llvm/lib/Transforms/Utils/LoopPeel.cpp
@@ -1030,8 +1030,9 @@ llvm::gatherPeelingPreferences(Loop *L, ScalarEvolution &SE,
/// for the bulk of dynamic execution, can be further simplified by scalar
/// optimizations.
bool llvm::peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
- ScalarEvolution *SE, const TargetTransformInfo *TTI, DominatorTree &DT, AssumptionCache *AC,
- bool PreserveLCSSA, ValueToValueMapTy &LVMap) {
+ ScalarEvolution *SE, const TargetTransformInfo *TTI,
+ DominatorTree &DT, AssumptionCache *AC, bool PreserveLCSSA,
+ ValueToValueMapTy &LVMap) {
assert(PeelCount > 0 && "Attempt to peel out zero iterations?");
assert(canPeel(L) && "Attempt to peel a loop which is not peelable?");
assert((!PeelLast || (canPeelLastIteration(*L, *SE) && PeelCount == 1)) &&
diff --git a/llvm/lib/Transforms/Utils/LoopSimplify.cpp b/llvm/lib/Transforms/Utils/LoopSimplify.cpp
index 77013d36185bd..5ddc646326996 100644
--- a/llvm/lib/Transforms/Utils/LoopSimplify.cpp
+++ b/llvm/lib/Transforms/Utils/LoopSimplify.cpp
@@ -474,7 +474,9 @@ static BasicBlock *insertUniqueBackedgeBlock(Loop *L, BasicBlock *Preheader,
static bool simplifyOneLoop(Loop *L, SmallVectorImpl<Loop *> &Worklist,
DominatorTree *DT, LoopInfo *LI,
ScalarEvolution *SE, AssumptionCache *AC,
- MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI, bool PreserveLCSSA) {
+ MemorySSAUpdater *MSSAU,
+ const TargetTransformInfo *TTI,
+ bool PreserveLCSSA) {
bool Changed = false;
if (MSSAU && VerifyMemorySSA)
MSSAU->getMemorySSA()->verifyMemorySSA();
@@ -695,7 +697,8 @@ static bool simplifyOneLoop(Loop *L, SmallVectorImpl<Loop *> &Worklist,
bool llvm::simplifyLoop(Loop *L, DominatorTree *DT, LoopInfo *LI,
ScalarEvolution *SE, AssumptionCache *AC,
- MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI, bool PreserveLCSSA) {
+ MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI,
+ bool PreserveLCSSA) {
bool Changed = false;
#ifndef NDEBUG
@@ -809,7 +812,8 @@ bool LoopSimplify::runOnFunction(Function &F) {
// Simplify each loop nest in the function.
for (auto *L : *LI)
- Changed |= simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), &TTI, PreserveLCSSA);
+ Changed |=
+ simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), &TTI, PreserveLCSSA);
#ifndef NDEBUG
if (PreserveLCSSA) {
@@ -836,12 +840,11 @@ PreservedAnalyses LoopSimplifyPass::run(Function &F,
}
auto &TTI = AM.getResult<TargetIRAnalysis>(F);
-
// Note that we don't preserve LCSSA in the new PM, if you need it run LCSSA
// after simplifying the loops. MemorySSA is preserved if it exists.
for (auto *L : *LI)
- Changed |=
- simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), &TTI, /*PreserveLCSSA*/ false);
+ Changed |= simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), &TTI,
+ /*PreserveLCSSA*/ false);
if (!Changed)
return PreservedAnalyses::all();
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index bb0586d2b215e..41f43e16e2a10 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -10425,8 +10425,8 @@ LoopVectorizeResult LoopVectorizePass::runImpl(Function &F) {
// will simplify all loops, regardless of whether anything end up being
// vectorized.
for (const auto &L : *LI)
- Changed |= CFGChanged |=
- simplifyLoop(L, DT, LI, SE, AC, nullptr, TTI, false /* PreserveLCSSA */);
+ Changed |= CFGChanged |= simplifyLoop(L, DT, LI, SE, AC, nullptr, TTI,
+ false /* PreserveLCSSA */);
// Build up a worklist of inner-loops to vectorize. This is necessary as
// the act of vectorizing or partially unrolling a loop creates new loops
>From f3f60e6a99c3ba9c693dc67550eca5b54ba5766d Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Fri, 4 Jul 2025 17:20:44 -0700
Subject: [PATCH 7/8] fixup unittest
---
llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll b/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll
index 1414c403d4aad..fdc0b0b911d3d 100644
--- a/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll
+++ b/llvm/test/Transforms/LoopVectorize/unsupported_early_exit.ll
@@ -85,8 +85,8 @@ define i64 @multiple_uncountable_exits() {
; CHECK-NEXT: [[INDEX_NEXT3]] = add nuw i64 [[INDEX1]], 4
; CHECK-NEXT: [[TMP8:%.*]] = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> [[TMP7]])
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT3]], 64
-; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
; CHECK-NEXT: [[TMP10:%.*]] = or i1 [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_SPLIT:%.*]], label [[SEARCH1]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.split:
; CHECK-NEXT: br i1 [[TMP8]], label [[VECTOR_EARLY_EXIT:%.*]], label [[MIDDLE_BLOCK:%.*]]
>From d9cdc2f051c59a99bd3fee91065994d3a9c47061 Mon Sep 17 00:00:00 2001
From: h2h <h2h at meta.com>
Date: Fri, 4 Jul 2025 19:07:55 -0700
Subject: [PATCH 8/8] format
---
llvm/include/llvm/Transforms/Utils/LoopConstrainer.h | 3 ++-
llvm/include/llvm/Transforms/Utils/LoopPeel.h | 5 +++--
llvm/include/llvm/Transforms/Utils/LoopSimplify.h | 3 ++-
3 files changed, 7 insertions(+), 4 deletions(-)
diff --git a/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h b/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h
index 1e7e9adedc98d..f73cb0a4c7bec 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopConstrainer.h
@@ -218,7 +218,8 @@ class LoopConstrainer {
LoopConstrainer(Loop &L, LoopInfo &LI,
function_ref<void(Loop *, bool)> LPMAddNewLoop,
const LoopStructure &LS, ScalarEvolution &SE,
- TargetTransformInfo &TTI, DominatorTree &DT, Type *T, SubRanges SR);
+ TargetTransformInfo &TTI, DominatorTree &DT, Type *T,
+ SubRanges SR);
// Entry point for the algorithm. Returns true on success.
bool run();
diff --git a/llvm/include/llvm/Transforms/Utils/LoopPeel.h b/llvm/include/llvm/Transforms/Utils/LoopPeel.h
index 73d3b84b36899..6da0ce9d6820a 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopPeel.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopPeel.h
@@ -31,8 +31,9 @@ bool canPeelLastIteration(const Loop &L, ScalarEvolution &SE);
/// off the last \p PeelCount iterations from \p L (canPeelLastIteration must be
/// true for \p L), otherwise peel off the first \p PeelCount iterations.
bool peelLoop(Loop *L, unsigned PeelCount, bool PeelLast, LoopInfo *LI,
- ScalarEvolution *SE, const TargetTransformInfo *TTI, DominatorTree &DT, AssumptionCache *AC,
- bool PreserveLCSSA, ValueToValueMapTy &VMap);
+ ScalarEvolution *SE, const TargetTransformInfo *TTI,
+ DominatorTree &DT, AssumptionCache *AC, bool PreserveLCSSA,
+ ValueToValueMapTy &VMap);
TargetTransformInfo::PeelingPreferences
gatherPeelingPreferences(Loop *L, ScalarEvolution &SE,
diff --git a/llvm/include/llvm/Transforms/Utils/LoopSimplify.h b/llvm/include/llvm/Transforms/Utils/LoopSimplify.h
index 230fac2f8a0ac..66c8482c468b0 100644
--- a/llvm/include/llvm/Transforms/Utils/LoopSimplify.h
+++ b/llvm/include/llvm/Transforms/Utils/LoopSimplify.h
@@ -65,7 +65,8 @@ class LoopSimplifyPass : public PassInfoMixin<LoopSimplifyPass> {
/// analyses if they're non-null, and LCSSA if \c PreserveLCSSA is true.
LLVM_ABI bool simplifyLoop(Loop *L, DominatorTree *DT, LoopInfo *LI,
ScalarEvolution *SE, AssumptionCache *AC,
- MemorySSAUpdater *MSSAU, const TargetTransformInfo *TTI, bool PreserveLCSSA);
+ MemorySSAUpdater *MSSAU,
+ const TargetTransformInfo *TTI, bool PreserveLCSSA);
} // end namespace llvm
More information about the llvm-commits
mailing list