[llvm] [Loads] Also use isAligned when checking assumptions. (PR #120916)
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Wed Jan 8 07:21:09 PST 2025
https://github.com/fhahn updated https://github.com/llvm/llvm-project/pull/120916
>From 9b33b6a2a6bbb28d812915624227180369467da1 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Sun, 22 Dec 2024 16:15:04 +0000
Subject: [PATCH 1/4] [Loads] Also use isAligned when checking assumptions.
Also use isAligned when trying to use alignment and dereferenceable
assumptions. This catches cases where dereferencable is known via the
assumption but alignment is known via isAligned (e.g. via argument
attribute or align of 1)
---
llvm/lib/Analysis/Loads.cpp | 10 +-
...able-info-from-assumption-constant-size.ll | 147 +++++++++++++-----
2 files changed, 111 insertions(+), 46 deletions(-)
diff --git a/llvm/lib/Analysis/Loads.cpp b/llvm/lib/Analysis/Loads.cpp
index 54b9521fda8fd2..c56a45aa32ae6f 100644
--- a/llvm/lib/Analysis/Loads.cpp
+++ b/llvm/lib/Analysis/Loads.cpp
@@ -173,9 +173,11 @@ static bool isDereferenceableAndAlignedPointer(
if (CtxI) {
/// Look through assumes to see if both dereferencability and alignment can
- /// be provent by an assume
+ /// be provent by an assume if needed.
RetainedKnowledge AlignRK;
RetainedKnowledge DerefRK;
+ APInt Offset(DL.getTypeStoreSizeInBits(V->getType()), 0);
+ bool IsAligned = isAligned(V, Offset, Alignment, DL);
if (getKnowledgeForValue(
V, {Attribute::Dereferenceable, Attribute::Alignment}, AC,
[&](RetainedKnowledge RK, Instruction *Assume, auto) {
@@ -185,8 +187,10 @@ static bool isDereferenceableAndAlignedPointer(
AlignRK = std::max(AlignRK, RK);
if (RK.AttrKind == Attribute::Dereferenceable)
DerefRK = std::max(DerefRK, RK);
- if (AlignRK && DerefRK && AlignRK.ArgValue >= Alignment.value() &&
- DerefRK.ArgValue >= Size.getZExtValue())
+ if (!IsAligned && AlignRK &&
+ AlignRK.ArgValue >= Alignment.value())
+ return false;
+ if (DerefRK && DerefRK.ArgValue >= Size.getZExtValue())
return true; // We have found what we needed so we stop looking
return false; // Other assumes may have better information. so
// keep looking
diff --git a/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll b/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll
index 1333451a0077e3..1636093a91a8e5 100644
--- a/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll
+++ b/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll
@@ -899,32 +899,15 @@ define void @deref_assumption_in_preheader_constant_trip_count_align_1(ptr noali
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_LOAD_CONTINUE2:.*]] ]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp sge <2 x i32> [[WIDE_LOAD]], zeroinitializer
-; CHECK-NEXT: [[TMP4:%.*]] = xor <2 x i1> [[TMP3]], splat (i1 true)
-; CHECK-NEXT: [[TMP5:%.*]] = extractelement <2 x i1> [[TMP4]], i32 0
-; CHECK-NEXT: br i1 [[TMP5]], label %[[PRED_LOAD_IF:.*]], label %[[PRED_LOAD_CONTINUE:.*]]
-; CHECK: [[PRED_LOAD_IF]]:
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 1
-; CHECK-NEXT: [[TMP8:%.*]] = insertelement <2 x i32> poison, i32 [[TMP7]], i32 0
-; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE]]
-; CHECK: [[PRED_LOAD_CONTINUE]]:
-; CHECK-NEXT: [[TMP9:%.*]] = phi <2 x i32> [ poison, %[[VECTOR_BODY]] ], [ [[TMP8]], %[[PRED_LOAD_IF]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = extractelement <2 x i1> [[TMP4]], i32 1
-; CHECK-NEXT: br i1 [[TMP10]], label %[[PRED_LOAD_IF1:.*]], label %[[PRED_LOAD_CONTINUE2]]
-; CHECK: [[PRED_LOAD_IF1]]:
-; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP11]]
-; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 1
-; CHECK-NEXT: [[TMP14:%.*]] = insertelement <2 x i32> [[TMP9]], i32 [[TMP13]], i32 1
-; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE2]]
-; CHECK: [[PRED_LOAD_CONTINUE2]]:
-; CHECK-NEXT: [[TMP15:%.*]] = phi <2 x i32> [ [[TMP9]], %[[PRED_LOAD_CONTINUE]] ], [ [[TMP14]], %[[PRED_LOAD_IF1]] ]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[TMP6]], i32 0
+; CHECK-NEXT: [[TMP15:%.*]] = load <2 x i32>, ptr [[TMP5]], align 1
; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP3]], <2 x i32> [[WIDE_LOAD]], <2 x i32> [[TMP15]]
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i32 0
@@ -1089,15 +1072,32 @@ define void @align_and_deref_assumption_in_preheader_constant_trip_count_align_4
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_LOAD_CONTINUE2:.*]] ]
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp sge <2 x i32> [[WIDE_LOAD]], zeroinitializer
+; CHECK-NEXT: [[TMP15:%.*]] = xor <2 x i1> [[TMP3]], splat (i1 true)
+; CHECK-NEXT: [[TMP5:%.*]] = extractelement <2 x i1> [[TMP15]], i32 0
+; CHECK-NEXT: br i1 [[TMP5]], label %[[PRED_LOAD_IF:.*]], label %[[PRED_LOAD_CONTINUE:.*]]
+; CHECK: [[PRED_LOAD_IF]]:
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[TMP4]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i32>, ptr [[TMP5]], align 4
+; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP4]], align 4
+; CHECK-NEXT: [[TMP17:%.*]] = insertelement <2 x i32> poison, i32 [[TMP16]], i32 0
+; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE]]
+; CHECK: [[PRED_LOAD_CONTINUE]]:
+; CHECK-NEXT: [[TMP9:%.*]] = phi <2 x i32> [ poison, %[[VECTOR_BODY]] ], [ [[TMP17]], %[[PRED_LOAD_IF]] ]
+; CHECK-NEXT: [[TMP10:%.*]] = extractelement <2 x i1> [[TMP15]], i32 1
+; CHECK-NEXT: br i1 [[TMP10]], label %[[PRED_LOAD_IF1:.*]], label %[[PRED_LOAD_CONTINUE2]]
+; CHECK: [[PRED_LOAD_IF1]]:
+; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
+; CHECK-NEXT: [[TMP14:%.*]] = insertelement <2 x i32> [[TMP9]], i32 [[TMP13]], i32 1
+; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE2]]
+; CHECK: [[PRED_LOAD_CONTINUE2]]:
+; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = phi <2 x i32> [ [[TMP9]], %[[PRED_LOAD_CONTINUE]] ], [ [[TMP14]], %[[PRED_LOAD_IF1]] ]
; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP3]], <2 x i32> [[WIDE_LOAD]], <2 x i32> [[WIDE_LOAD1]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 0
@@ -1168,32 +1168,93 @@ define void @deref_assumption_in_preheader_constant_trip_count_align_4_known_via
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_LOAD_CONTINUE2:.*]] ]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp sge <2 x i32> [[WIDE_LOAD]], zeroinitializer
-; CHECK-NEXT: [[TMP4:%.*]] = xor <2 x i1> [[TMP3]], splat (i1 true)
-; CHECK-NEXT: [[TMP5:%.*]] = extractelement <2 x i1> [[TMP4]], i32 0
-; CHECK-NEXT: br i1 [[TMP5]], label %[[PRED_LOAD_IF:.*]], label %[[PRED_LOAD_CONTINUE:.*]]
-; CHECK: [[PRED_LOAD_IF]]:
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
-; CHECK-NEXT: [[TMP8:%.*]] = insertelement <2 x i32> poison, i32 [[TMP7]], i32 0
-; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE]]
-; CHECK: [[PRED_LOAD_CONTINUE]]:
-; CHECK-NEXT: [[TMP9:%.*]] = phi <2 x i32> [ poison, %[[VECTOR_BODY]] ], [ [[TMP8]], %[[PRED_LOAD_IF]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = extractelement <2 x i1> [[TMP4]], i32 1
-; CHECK-NEXT: br i1 [[TMP10]], label %[[PRED_LOAD_IF1:.*]], label %[[PRED_LOAD_CONTINUE2]]
-; CHECK: [[PRED_LOAD_IF1]]:
-; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP11]]
-; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
-; CHECK-NEXT: [[TMP14:%.*]] = insertelement <2 x i32> [[TMP9]], i32 [[TMP13]], i32 1
-; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE2]]
-; CHECK: [[PRED_LOAD_CONTINUE2]]:
-; CHECK-NEXT: [[TMP15:%.*]] = phi <2 x i32> [ [[TMP9]], %[[PRED_LOAD_CONTINUE]] ], [ [[TMP14]], %[[PRED_LOAD_IF1]] ]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[TMP6]], i32 0
+; CHECK-NEXT: [[TMP15:%.*]] = load <2 x i32>, ptr [[TMP5]], align 4
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP3]], <2 x i32> [[WIDE_LOAD]], <2 x i32> [[TMP15]]
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i32 0
+; CHECK-NEXT: store <2 x i32> [[PREDPHI]], ptr [[TMP17]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; CHECK-NEXT: br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[SCALAR_PH]]
+; CHECK: [[SCALAR_PH]]:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT: br label %[[LOOP_HEADER:.*]]
+; CHECK: [[LOOP_HEADER]]:
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP_LATCH:.*]] ]
+; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; CHECK-NEXT: [[L_B:%.*]] = load i32, ptr [[GEP_B]], align 4
+; CHECK-NEXT: [[C_1:%.*]] = icmp sge i32 [[L_B]], 0
+; CHECK-NEXT: br i1 [[C_1]], label %[[LOOP_LATCH]], label %[[LOOP_THEN:.*]]
+; CHECK: [[LOOP_THEN]]:
+; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i32, ptr [[A]], i64 [[IV]]
+; CHECK-NEXT: [[L_A:%.*]] = load i32, ptr [[GEP_A]], align 4
+; CHECK-NEXT: br label %[[LOOP_LATCH]]
+; CHECK: [[LOOP_LATCH]]:
+; CHECK-NEXT: [[MERGE:%.*]] = phi i32 [ [[L_A]], %[[LOOP_THEN]] ], [ [[L_B]], %[[LOOP_HEADER]] ]
+; CHECK-NEXT: [[GEP_C:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
+; CHECK-NEXT: store i32 [[MERGE]], ptr [[GEP_C]], align 4
+; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP_HEADER]], !llvm.loop [[LOOP25:![0-9]+]]
+; CHECK: [[EXIT]]:
+; CHECK-NEXT: ret void
+;
+entry:
+ call void @llvm.assume(i1 true) [ "dereferenceable"(ptr %a, i64 4000) ]
+ br label %loop.header
+
+loop.header:
+ %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
+ %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
+ %l.b = load i32, ptr %gep.b, align 4
+ %c.1 = icmp sge i32 %l.b, 0
+ br i1 %c.1, label %loop.latch, label %loop.then
+
+loop.then:
+ %gep.a = getelementptr i32, ptr %a, i64 %iv
+ %l.a = load i32, ptr %gep.a, align 4
+ br label %loop.latch
+
+loop.latch:
+ %merge = phi i32 [ %l.a, %loop.then ], [ %l.b, %loop.header ]
+ %gep.c = getelementptr inbounds i32, ptr %c, i64 %iv
+ store i32 %merge, ptr %gep.c, align 4
+ %iv.next = add nuw nsw i64 %iv, 1
+ %ec = icmp eq i64 %iv.next, 1000
+ br i1 %ec, label %exit, label %loop.header
+
+exit:
+ ret void
+}
+
+define void @deref_assumption_in_preheader_constant_trip_count_align_4_not_known(ptr noalias %a, ptr noalias %b, ptr noalias %c) {
+; CHECK-LABEL: define void @deref_assumption_in_preheader_constant_trip_count_align_4_not_known(
+; CHECK-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]], ptr noalias [[C:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: call void @llvm.assume(i1 true) [ "dereferenceable"(ptr [[A]], i64 4000) ]
+; 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: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[TMP3:%.*]] = icmp sge <2 x i32> [[WIDE_LOAD]], zeroinitializer
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[TMP6]], i32 0
+; CHECK-NEXT: [[TMP15:%.*]] = load <2 x i32>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP3]], <2 x i32> [[WIDE_LOAD]], <2 x i32> [[TMP15]]
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i32 0
>From 4dbce03576e76198f73c05c05d51d700e54dfe17 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Wed, 8 Jan 2025 15:12:07 +0000
Subject: [PATCH 2/4] !fixup update tests
---
...able-info-from-assumption-constant-size.ll | 120 +++---------------
1 file changed, 19 insertions(+), 101 deletions(-)
diff --git a/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll b/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll
index 1636093a91a8e5..103bd0fb8a2313 100644
--- a/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll
+++ b/llvm/test/Transforms/LoopVectorize/dereferenceable-info-from-assumption-constant-size.ll
@@ -113,16 +113,29 @@ define void @deref_assumption_in_header_constant_trip_count_loop_invariant_ptr(p
; CHECK: [[VECTOR_PH]]:
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_LOAD_CONTINUE2:.*]] ]
; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp sge <2 x i32> [[WIDE_LOAD]], zeroinitializer
-; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[A]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = xor <2 x i1> [[TMP3]], splat (i1 true)
+; CHECK-NEXT: [[TMP6:%.*]] = extractelement <2 x i1> [[TMP4]], i32 0
+; CHECK-NEXT: br i1 [[TMP6]], label %[[PRED_LOAD_IF:.*]], label %[[PRED_LOAD_CONTINUE:.*]]
+; CHECK: [[PRED_LOAD_IF]]:
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[A]], align 4
-; CHECK-NEXT: [[TMP6:%.*]] = insertelement <2 x i32> poison, i32 [[TMP4]], i32 0
-; CHECK-NEXT: [[TMP7:%.*]] = insertelement <2 x i32> [[TMP6]], i32 [[TMP5]], i32 1
+; CHECK-NEXT: [[TMP12:%.*]] = insertelement <2 x i32> poison, i32 [[TMP5]], i32 0
+; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE]]
+; CHECK: [[PRED_LOAD_CONTINUE]]:
+; CHECK-NEXT: [[TMP13:%.*]] = phi <2 x i32> [ poison, %[[VECTOR_BODY]] ], [ [[TMP12]], %[[PRED_LOAD_IF]] ]
+; CHECK-NEXT: [[TMP14:%.*]] = extractelement <2 x i1> [[TMP4]], i32 1
+; CHECK-NEXT: br i1 [[TMP14]], label %[[PRED_LOAD_IF1:.*]], label %[[PRED_LOAD_CONTINUE2]]
+; CHECK: [[PRED_LOAD_IF1]]:
+; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[A]], align 4
+; CHECK-NEXT: [[TMP11:%.*]] = insertelement <2 x i32> [[TMP13]], i32 [[TMP15]], i32 1
+; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE2]]
+; CHECK: [[PRED_LOAD_CONTINUE2]]:
+; CHECK-NEXT: [[TMP7:%.*]] = phi <2 x i32> [ [[TMP13]], %[[PRED_LOAD_CONTINUE]] ], [ [[TMP11]], %[[PRED_LOAD_IF1]] ]
; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP3]], <2 x i32> [[WIDE_LOAD]], <2 x i32> [[TMP7]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP8]], i32 0
@@ -1183,7 +1196,7 @@ define void @deref_assumption_in_preheader_constant_trip_count_align_4_known_via
; CHECK-NEXT: store <2 x i32> [[PREDPHI]], ptr [[TMP17]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; CHECK-NEXT: br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[SCALAR_PH]]
; CHECK: [[SCALAR_PH]]:
@@ -1205,7 +1218,7 @@ define void @deref_assumption_in_preheader_constant_trip_count_align_4_known_via
; CHECK-NEXT: store i32 [[MERGE]], ptr [[GEP_C]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP_HEADER]], !llvm.loop [[LOOP25:![0-9]+]]
+; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP_HEADER]], !llvm.loop [[LOOP27:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -1261,101 +1274,6 @@ define void @deref_assumption_in_preheader_constant_trip_count_align_4_not_known
; CHECK-NEXT: store <2 x i32> [[PREDPHI]], ptr [[TMP17]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; CHECK-NEXT: br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
-; CHECK: [[MIDDLE_BLOCK]]:
-; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[SCALAR_PH]]
-; CHECK: [[SCALAR_PH]]:
-; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; CHECK-NEXT: br label %[[LOOP_HEADER:.*]]
-; CHECK: [[LOOP_HEADER]]:
-; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP_LATCH:.*]] ]
-; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT: [[L_B:%.*]] = load i32, ptr [[GEP_B]], align 4
-; CHECK-NEXT: [[C_1:%.*]] = icmp sge i32 [[L_B]], 0
-; CHECK-NEXT: br i1 [[C_1]], label %[[LOOP_LATCH]], label %[[LOOP_THEN:.*]]
-; CHECK: [[LOOP_THEN]]:
-; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i32, ptr [[A]], i64 [[IV]]
-; CHECK-NEXT: [[L_A:%.*]] = load i32, ptr [[GEP_A]], align 4
-; CHECK-NEXT: br label %[[LOOP_LATCH]]
-; CHECK: [[LOOP_LATCH]]:
-; CHECK-NEXT: [[MERGE:%.*]] = phi i32 [ [[L_A]], %[[LOOP_THEN]] ], [ [[L_B]], %[[LOOP_HEADER]] ]
-; CHECK-NEXT: [[GEP_C:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
-; CHECK-NEXT: store i32 [[MERGE]], ptr [[GEP_C]], align 4
-; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
-; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT: br i1 [[EC]], label %[[EXIT]], label %[[LOOP_HEADER]], !llvm.loop [[LOOP27:![0-9]+]]
-; CHECK: [[EXIT]]:
-; CHECK-NEXT: ret void
-;
-entry:
- call void @llvm.assume(i1 true) [ "dereferenceable"(ptr %a, i64 4000) ]
- br label %loop.header
-
-loop.header:
- %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
- %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
- %l.b = load i32, ptr %gep.b, align 4
- %c.1 = icmp sge i32 %l.b, 0
- br i1 %c.1, label %loop.latch, label %loop.then
-
-loop.then:
- %gep.a = getelementptr i32, ptr %a, i64 %iv
- %l.a = load i32, ptr %gep.a, align 4
- br label %loop.latch
-
-loop.latch:
- %merge = phi i32 [ %l.a, %loop.then ], [ %l.b, %loop.header ]
- %gep.c = getelementptr inbounds i32, ptr %c, i64 %iv
- store i32 %merge, ptr %gep.c, align 4
- %iv.next = add nuw nsw i64 %iv, 1
- %ec = icmp eq i64 %iv.next, 1000
- br i1 %ec, label %exit, label %loop.header
-
-exit:
- ret void
-}
-
-define void @deref_assumption_in_preheader_constant_trip_count_align_4_not_known(ptr noalias %a, ptr noalias %b, ptr noalias %c) {
-; CHECK-LABEL: define void @deref_assumption_in_preheader_constant_trip_count_align_4_not_known(
-; CHECK-SAME: ptr noalias [[A:%.*]], ptr noalias [[B:%.*]], ptr noalias [[C:%.*]]) {
-; CHECK-NEXT: [[ENTRY:.*]]:
-; CHECK-NEXT: call void @llvm.assume(i1 true) [ "dereferenceable"(ptr [[A]], i64 4000) ]
-; 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: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_LOAD_CONTINUE2:.*]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
-; CHECK-NEXT: [[TMP3:%.*]] = icmp sge <2 x i32> [[WIDE_LOAD]], zeroinitializer
-; CHECK-NEXT: [[TMP4:%.*]] = xor <2 x i1> [[TMP3]], splat (i1 true)
-; CHECK-NEXT: [[TMP5:%.*]] = extractelement <2 x i1> [[TMP4]], i32 0
-; CHECK-NEXT: br i1 [[TMP5]], label %[[PRED_LOAD_IF:.*]], label %[[PRED_LOAD_CONTINUE:.*]]
-; CHECK: [[PRED_LOAD_IF]]:
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
-; CHECK-NEXT: [[TMP8:%.*]] = insertelement <2 x i32> poison, i32 [[TMP7]], i32 0
-; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE]]
-; CHECK: [[PRED_LOAD_CONTINUE]]:
-; CHECK-NEXT: [[TMP9:%.*]] = phi <2 x i32> [ poison, %[[VECTOR_BODY]] ], [ [[TMP8]], %[[PRED_LOAD_IF]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = extractelement <2 x i1> [[TMP4]], i32 1
-; CHECK-NEXT: br i1 [[TMP10]], label %[[PRED_LOAD_IF1:.*]], label %[[PRED_LOAD_CONTINUE2]]
-; CHECK: [[PRED_LOAD_IF1]]:
-; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP11]]
-; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
-; CHECK-NEXT: [[TMP14:%.*]] = insertelement <2 x i32> [[TMP9]], i32 [[TMP13]], i32 1
-; CHECK-NEXT: br label %[[PRED_LOAD_CONTINUE2]]
-; CHECK: [[PRED_LOAD_CONTINUE2]]:
-; CHECK-NEXT: [[TMP15:%.*]] = phi <2 x i32> [ [[TMP9]], %[[PRED_LOAD_CONTINUE]] ], [ [[TMP14]], %[[PRED_LOAD_IF1]] ]
-; CHECK-NEXT: [[PREDPHI:%.*]] = select <2 x i1> [[TMP3]], <2 x i32> [[WIDE_LOAD]], <2 x i32> [[TMP15]]
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i32 0
-; CHECK-NEXT: store <2 x i32> [[PREDPHI]], ptr [[TMP17]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
; CHECK-NEXT: br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: br i1 true, label %[[EXIT:.*]], label %[[SCALAR_PH]]
>From e90d0fa566e46a7e0891e719842fc64d6231faa8 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Wed, 8 Jan 2025 15:19:44 +0000
Subject: [PATCH 3/4] !fixup restore changes to address comments
---
llvm/lib/Analysis/Loads.cpp | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/llvm/lib/Analysis/Loads.cpp b/llvm/lib/Analysis/Loads.cpp
index c56a45aa32ae6f..d322bc4175d21a 100644
--- a/llvm/lib/Analysis/Loads.cpp
+++ b/llvm/lib/Analysis/Loads.cpp
@@ -173,11 +173,11 @@ static bool isDereferenceableAndAlignedPointer(
if (CtxI) {
/// Look through assumes to see if both dereferencability and alignment can
- /// be provent by an assume if needed.
+ /// be proven by an assume if needed.
RetainedKnowledge AlignRK;
RetainedKnowledge DerefRK;
APInt Offset(DL.getTypeStoreSizeInBits(V->getType()), 0);
- bool IsAligned = isAligned(V, Offset, Alignment, DL);
+ bool IsAligned = V->getPointerAlignment(DL) >= Alignment;
if (getKnowledgeForValue(
V, {Attribute::Dereferenceable, Attribute::Alignment}, AC,
[&](RetainedKnowledge RK, Instruction *Assume, auto) {
>From 6f4c2f097fb36be4c35842fa42553083928ff890 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Wed, 8 Jan 2025 15:20:46 +0000
Subject: [PATCH 4/4] !fixup remove dead Offset variable.
---
llvm/lib/Analysis/Loads.cpp | 1 -
1 file changed, 1 deletion(-)
diff --git a/llvm/lib/Analysis/Loads.cpp b/llvm/lib/Analysis/Loads.cpp
index d322bc4175d21a..2e239df8751c3d 100644
--- a/llvm/lib/Analysis/Loads.cpp
+++ b/llvm/lib/Analysis/Loads.cpp
@@ -176,7 +176,6 @@ static bool isDereferenceableAndAlignedPointer(
/// be proven by an assume if needed.
RetainedKnowledge AlignRK;
RetainedKnowledge DerefRK;
- APInt Offset(DL.getTypeStoreSizeInBits(V->getType()), 0);
bool IsAligned = V->getPointerAlignment(DL) >= Alignment;
if (getKnowledgeForValue(
V, {Attribute::Dereferenceable, Attribute::Alignment}, AC,
More information about the llvm-commits
mailing list