[llvm] [InstComb] Allow more user for (add (ptrtoint %B), %O) to GEP transform. (PR #153566)
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Thu Aug 14 03:36:31 PDT 2025
https://github.com/fhahn created https://github.com/llvm/llvm-project/pull/153566
Generalize the logic from
https://github.com/llvm/llvm-project/pull/153421 to support additional
cases where the pointer is only used as integer.
Alive2 Proof: https://alive2.llvm.org/ce/z/po58pP
This enables vectorizing std::find for some cases, if additional
assumptions are provided: https://godbolt.org/z/94oq3576E
Depends on https://github.com/llvm/llvm-project/pull/153421 (included in
the PR).
>From 1a6529d8d2248a560a8919d62aec7f3da1c0d967 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Wed, 13 Aug 2025 15:09:35 +0100
Subject: [PATCH 1/5] [InstCombine] Add tests for (inttoptr (add (ptrtoint
%Base), %Offset)).
---
.../InstCombine/fold-bin-operand.ll | 158 +++++++++++++++++-
1 file changed, 157 insertions(+), 1 deletion(-)
diff --git a/llvm/test/Transforms/InstCombine/fold-bin-operand.ll b/llvm/test/Transforms/InstCombine/fold-bin-operand.ll
index f28262b2a77e0..38baf237d9223 100644
--- a/llvm/test/Transforms/InstCombine/fold-bin-operand.ll
+++ b/llvm/test/Transforms/InstCombine/fold-bin-operand.ll
@@ -1,6 +1,6 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes=instcombine -S | FileCheck %s
-target datalayout = "E-p:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128"
+target datalayout = "E-p:64:64:64-p1:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128"
define i1 @f(i1 %x) {
; CHECK-LABEL: @f(
@@ -30,6 +30,162 @@ define i32 @g(i32 %x) {
ret i32 %b
}
+define i1 @inttoptr_add_ptrtoint_used_by_single_icmp(ptr %src, ptr %p2) {
+; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_single_icmp(
+; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
+; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: ret i1 [[C]]
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ %c = icmp eq ptr %p, %p2
+ ret i1 %c
+}
+
+define i1 @inttoptr_add_ptrtoint_used_by_single_icmp_different_src_address_spaces(ptr addrspace(1) %src, ptr %p2) {
+; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_single_icmp_different_src_address_spaces(
+; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr addrspace(1) [[SRC:%.*]] to i64
+; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: ret i1 [[C]]
+;
+ %i = ptrtoint ptr addrspace(1) %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ %c = icmp eq ptr %p, %p2
+ ret i1 %c
+}
+
+define i1 @inttoptr_add_ptrtoint_used_by_single_icmp_different_dst_address_spaces(ptr %src, ptr addrspace(1) %p2) {
+; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_single_icmp_different_dst_address_spaces(
+; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
+; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr addrspace(1)
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr addrspace(1) [[P2:%.*]], [[P]]
+; CHECK-NEXT: ret i1 [[C]]
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr addrspace(1)
+ %c = icmp eq ptr addrspace(1) %p, %p2
+ ret i1 %c
+}
+
+define i1 @inttoptr_add_ptrtoint_used_by_single_icmp_int_type_does_not_match_ptr_ty(ptr %src, ptr %p2) {
+; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_single_icmp_int_type_does_not_match_ptr_ty(
+; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
+; CHECK-NEXT: [[I:%.*]] = trunc i64 [[TMP1]] to i8
+; CHECK-NEXT: [[A:%.*]] = add i8 [[I]], 10
+; CHECK-NEXT: [[TMP2:%.*]] = zext i8 [[A]] to i64
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[TMP2]] to ptr
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: ret i1 [[C]]
+;
+ %i = ptrtoint ptr %src to i8
+ %a = add i8 %i, 10
+ %p = inttoptr i8 %a to ptr
+ %c = icmp eq ptr %p, %p2
+ ret i1 %c
+}
+
+define i1 @multiple_inttoptr_add_ptrtoint_used_by_single_icmp(ptr %src) {
+; CHECK-LABEL: @multiple_inttoptr_add_ptrtoint_used_by_single_icmp(
+; CHECK-NEXT: ret i1 false
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ %a.2 = add i64 %i, 11
+ %p.2 = inttoptr i64 %a.2 to ptr
+ %c = icmp eq ptr %p, %p.2
+ ret i1 %c
+}
+
+define i1 @multiple_inttoptr_add_ptrtoint_used_by_single_icmp_non_constant_offset(ptr %src, i64 %off.1, i64 %off.2) {
+; CHECK-LABEL: @multiple_inttoptr_add_ptrtoint_used_by_single_icmp_non_constant_offset(
+; CHECK-NEXT: [[C:%.*]] = icmp eq i64 [[OFF_1:%.*]], [[OFF_2:%.*]]
+; CHECK-NEXT: ret i1 [[C]]
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, %off.1
+ %p = inttoptr i64 %a to ptr
+ %a.2 = add i64 %i, %off.2
+ %p.2 = inttoptr i64 %a.2 to ptr
+ %c = icmp eq ptr %p, %p.2
+ ret i1 %c
+}
+
+define i1 @inttoptr_add_ptrtoint_used_by_single_icmp_in_different_bb(i1 %bc, ptr %src, ptr %p2) {
+; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_single_icmp_in_different_bb(
+; CHECK-NEXT: br i1 [[BC:%.*]], label [[THEN:%.*]], label [[ELSE:%.*]]
+; CHECK: then:
+; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
+; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: ret i1 [[C]]
+; CHECK: else:
+; CHECK-NEXT: ret i1 false
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ br i1 %bc, label %then, label %else
+
+then:
+ %c = icmp eq ptr %p, %p2
+ ret i1 %c
+
+else:
+ ret i1 false
+}
+
+define i1 @inttoptr_add_ptrtoint_used_by_multiple_icmps(ptr %src, ptr %p2, ptr %p3) {
+; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_multiple_icmps(
+; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
+; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
+; CHECK-NEXT: [[C_1:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: [[C_2:%.*]] = icmp eq ptr [[P3:%.*]], [[P]]
+; CHECK-NEXT: [[XOR:%.*]] = xor i1 [[C_1]], [[C_2]]
+; CHECK-NEXT: ret i1 [[XOR]]
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ %c.1 = icmp eq ptr %p, %p2
+ %c.2 = icmp eq ptr %p, %p3
+ %xor = xor i1 %c.1, %c.2
+ ret i1 %xor
+}
+
+declare void @foo(ptr)
+
+define i1 @inttoptr_add_ptrtoint_used_by_multiple_icmps_and_other_user(ptr %src, ptr %p2, ptr %p3) {
+; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_multiple_icmps_and_other_user(
+; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
+; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[TMP1]], 10
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[TMP2]] to ptr
+; CHECK-NEXT: [[C_1:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: [[C_2:%.*]] = icmp eq ptr [[P3:%.*]], [[P]]
+; CHECK-NEXT: [[XOR:%.*]] = xor i1 [[C_1]], [[C_2]]
+; CHECK-NEXT: call void @foo(ptr [[P]])
+; CHECK-NEXT: ret i1 [[XOR]]
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ %c.1 = icmp eq ptr %p, %p2
+ %c.2 = icmp eq ptr %p, %p3
+ %xor = xor i1 %c.1, %c.2
+ call void @foo(ptr %p)
+ ret i1 %xor
+}
+
define i32 @h(i1 %A, i32 %B) {
; CHECK-LABEL: @h(
; CHECK-NEXT: EntryBlock:
>From 2758a5edb5b3059d81c12982a127e19c35bfcaf7 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Wed, 13 Aug 2025 15:11:44 +0100
Subject: [PATCH 2/5] [InstComb] Try to convert inttoptr (add (ptrtoint %B),
%O) to GEP.
Replace inttoptr (add (ptrtoint %B), %O) with (getelementptr i8, %B, %o)
if all users are ICmp instruction, which in turn means only the address
value is compared. We should be able to do this, if the src pointer,
the integer type and the destination pointer types have the same
bitwidth and address space.
A common source of such (inttoptr (add (ptrtoint %B), %O)) is from
various iterations in libc++.
In practice this triggers in a number of files in Clang and various open
source projects, including cppcheck, diamond, llama and more.
Alive2 Proof with constant offset: https://alive2.llvm.org/ce/z/K_5N_B
---
.../InstCombine/InstCombineCasts.cpp | 16 +++++++++++++++
.../InstCombine/fold-bin-operand.ll | 20 +++++++------------
2 files changed, 23 insertions(+), 13 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
index a43a6ee1f58b0..2edabcec2aecf 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
@@ -2073,6 +2073,22 @@ Instruction *InstCombinerImpl::visitIntToPtr(IntToPtrInst &CI) {
return new IntToPtrInst(P, CI.getType());
}
+ // Replace (inttoptr (add (ptrtoint %Base), %Offset)) with
+ //(getelementptr i8, %Base, %Offset) if all users are ICmps.
+ Value *Base;
+ Value *Offset;
+ if (match(CI.getOperand(0),
+ m_Add(m_PtrToInt(m_Value(Base)), m_Value(Offset))) &&
+ all_of(CI.users(), IsaPred<ICmpInst>)) {
+ Type *BasePtrTy = Base->getType();
+ if (CI.getType()->getPointerAddressSpace() ==
+ BasePtrTy->getPointerAddressSpace() &&
+ DL.getTypeSizeInBits(BasePtrTy) ==
+ DL.getTypeSizeInBits(CI.getSrcTy())) {
+ return GetElementPtrInst::Create(Builder.getInt8Ty(), Base, Offset);
+ }
+ }
+
if (Instruction *I = commonCastTransforms(CI))
return I;
diff --git a/llvm/test/Transforms/InstCombine/fold-bin-operand.ll b/llvm/test/Transforms/InstCombine/fold-bin-operand.ll
index 38baf237d9223..f596bc5d226be 100644
--- a/llvm/test/Transforms/InstCombine/fold-bin-operand.ll
+++ b/llvm/test/Transforms/InstCombine/fold-bin-operand.ll
@@ -32,10 +32,8 @@ define i32 @g(i32 %x) {
define i1 @inttoptr_add_ptrtoint_used_by_single_icmp(ptr %src, ptr %p2) {
; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_single_icmp(
-; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
-; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
-; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[SRC:%.*]], i64 10
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2]], [[P:%.*]]
; CHECK-NEXT: ret i1 [[C]]
;
%i = ptrtoint ptr %src to i64
@@ -123,10 +121,8 @@ define i1 @inttoptr_add_ptrtoint_used_by_single_icmp_in_different_bb(i1 %bc, ptr
; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_single_icmp_in_different_bb(
; CHECK-NEXT: br i1 [[BC:%.*]], label [[THEN:%.*]], label [[ELSE:%.*]]
; CHECK: then:
-; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
-; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
-; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[SRC:%.*]], i64 10
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[P2]], [[P:%.*]]
; CHECK-NEXT: ret i1 [[C]]
; CHECK: else:
; CHECK-NEXT: ret i1 false
@@ -146,11 +142,9 @@ else:
define i1 @inttoptr_add_ptrtoint_used_by_multiple_icmps(ptr %src, ptr %p2, ptr %p3) {
; CHECK-LABEL: @inttoptr_add_ptrtoint_used_by_multiple_icmps(
-; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
-; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
-; CHECK-NEXT: [[C_1:%.*]] = icmp eq ptr [[P2:%.*]], [[P]]
-; CHECK-NEXT: [[C_2:%.*]] = icmp eq ptr [[P3:%.*]], [[P]]
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[SRC:%.*]], i64 10
+; CHECK-NEXT: [[C_1:%.*]] = icmp eq ptr [[P2]], [[P:%.*]]
+; CHECK-NEXT: [[C_2:%.*]] = icmp eq ptr [[P2]], [[P3:%.*]]
; CHECK-NEXT: [[XOR:%.*]] = xor i1 [[C_1]], [[C_2]]
; CHECK-NEXT: ret i1 [[XOR]]
;
>From 8e2ac0e12c48703522e1851ef3810079b2e1c8b6 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Wed, 13 Aug 2025 21:27:31 +0100
Subject: [PATCH 3/5] !fixup use m_PtrToIntSameSize
---
llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp | 12 ++++--------
1 file changed, 4 insertions(+), 8 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
index 2edabcec2aecf..184d573f52fe7 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
@@ -2078,15 +2078,11 @@ Instruction *InstCombinerImpl::visitIntToPtr(IntToPtrInst &CI) {
Value *Base;
Value *Offset;
if (match(CI.getOperand(0),
- m_Add(m_PtrToInt(m_Value(Base)), m_Value(Offset))) &&
+ m_Add(m_PtrToIntSameSize(DL, m_Value(Base)), m_Value(Offset))) &&
+ CI.getType()->getPointerAddressSpace() ==
+ Base->getType()->getPointerAddressSpace() &&
all_of(CI.users(), IsaPred<ICmpInst>)) {
- Type *BasePtrTy = Base->getType();
- if (CI.getType()->getPointerAddressSpace() ==
- BasePtrTy->getPointerAddressSpace() &&
- DL.getTypeSizeInBits(BasePtrTy) ==
- DL.getTypeSizeInBits(CI.getSrcTy())) {
- return GetElementPtrInst::Create(Builder.getInt8Ty(), Base, Offset);
- }
+ return GetElementPtrInst::Create(Builder.getInt8Ty(), Base, Offset);
}
if (Instruction *I = commonCastTransforms(CI))
>From eb8899e5bb5065edbb2254c7b92c2a48981db414 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Thu, 14 Aug 2025 10:39:29 +0100
Subject: [PATCH 4/5] [InstComb] Add tests with phi users of inttoptr.
---
.../InstCombine/inttoptr-add-phi.ll | 70 +++++++++++++++++++
1 file changed, 70 insertions(+)
create mode 100644 llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll
diff --git a/llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll b/llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll
new file mode 100644
index 0000000000000..e9fd6e5a07dd0
--- /dev/null
+++ b/llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll
@@ -0,0 +1,70 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -passes=instcombine -S %s | FileCheck %s
+target datalayout = "E-p:64:64:64-p1:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128"
+
+define i64 @inttoptr_used_by_phi_with_ptrtoint(i1 %c, ptr %src, ptr %p2) {
+; CHECK-LABEL: define i64 @inttoptr_used_by_phi_with_ptrtoint(
+; CHECK-SAME: i1 [[C:%.*]], ptr [[SRC:%.*]], ptr [[P2:%.*]]) {
+; CHECK-NEXT: br i1 [[C]], label %[[THEN:.*]], label %[[ELSE:.*]]
+; CHECK: [[THEN]]:
+; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC]] to i64
+; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: br label %[[EXIT:.*]]
+; CHECK: [[ELSE]]:
+; CHECK-NEXT: br label %[[EXIT]]
+; CHECK: [[EXIT]]:
+; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[A]], %[[THEN]] ], [ 0, %[[ELSE]] ]
+; CHECK-NEXT: ret i64 [[PHI]]
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ br i1 %c, label %then, label %else
+
+then:
+ br label %exit
+
+else:
+ br label %exit
+
+exit:
+ %phi = phi ptr [ %p, %then ], [ null, %else ]
+ %i.2 = ptrtoint ptr %phi to i64
+ ret i64 %i.2
+}
+
+declare void @foo(ptr)
+
+define i64 @inttoptr_used_by_phi_with_other_users(i1 %c, ptr %src, ptr %p2) {
+; CHECK-LABEL: define i64 @inttoptr_used_by_phi_with_other_users(
+; CHECK-SAME: i1 [[C:%.*]], ptr [[SRC:%.*]], ptr [[P2:%.*]]) {
+; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC]] to i64
+; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: br i1 [[C]], label %[[THEN:.*]], label %[[ELSE:.*]]
+; CHECK: [[THEN]]:
+; CHECK-NEXT: br label %[[EXIT:.*]]
+; CHECK: [[ELSE]]:
+; CHECK-NEXT: br label %[[EXIT]]
+; CHECK: [[EXIT]]:
+; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[A]], %[[THEN]] ], [ 0, %[[ELSE]] ]
+; CHECK-NEXT: [[P:%.*]] = inttoptr i64 [[A]] to ptr
+; CHECK-NEXT: call void @foo(ptr [[P]])
+; CHECK-NEXT: ret i64 [[PHI]]
+;
+ %i = ptrtoint ptr %src to i64
+ %a = add i64 %i, 10
+ %p = inttoptr i64 %a to ptr
+ br i1 %c, label %then, label %else
+
+then:
+ br label %exit
+
+else:
+ br label %exit
+
+exit:
+ %phi = phi ptr [ %p, %then ], [ null, %else ]
+ call void @foo(ptr %p)
+ %i.2 = ptrtoint ptr %phi to i64
+ ret i64 %i.2
+}
>From 94a5c2767f9c89637fbb1f6d78297dd41462dae9 Mon Sep 17 00:00:00 2001
From: Florian Hahn <flo at fhahn.com>
Date: Thu, 14 Aug 2025 10:59:54 +0100
Subject: [PATCH 5/5] [InstComb] Allow more user for (add (ptrtoint %B), %O) to
GEP transform.
Generalize the logic from
https://github.com/llvm/llvm-project/pull/153421 to support additional
cases where the pointer is only used as integer.
Alive2 Proof: https://alive2.llvm.org/ce/z/po58pP
This enables vectorizing std::find for some cases, if additional
assumptions are provided: https://godbolt.org/z/94oq3576E
Depends on https://github.com/llvm/llvm-project/pull/153421 (included in
the PR).
---
.../InstCombine/InstCombineCasts.cpp | 12 +++-
.../InstCombine/inttoptr-add-phi.ll | 6 +-
.../PhaseOrdering/AArch64/std-find.ll | 70 ++++++++++---------
3 files changed, 51 insertions(+), 37 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
index 184d573f52fe7..dcab043c14fa5 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp
@@ -2074,14 +2074,22 @@ Instruction *InstCombinerImpl::visitIntToPtr(IntToPtrInst &CI) {
}
// Replace (inttoptr (add (ptrtoint %Base), %Offset)) with
- //(getelementptr i8, %Base, %Offset) if all users are ICmps.
+ // (getelementptr i8, %Base, %Offset) if the pointer is only used as integer
+ // value.
Value *Base;
Value *Offset;
+ auto UsesPointerAsInt = [](User *U) {
+ if (isa<ICmpInst, PtrToIntInst>(U))
+ return true;
+ if (auto *P = dyn_cast<PHINode>(U))
+ return P->hasOneUse() && isa<ICmpInst, PtrToIntInst>(*P->user_begin());
+ return false;
+ };
if (match(CI.getOperand(0),
m_Add(m_PtrToIntSameSize(DL, m_Value(Base)), m_Value(Offset))) &&
CI.getType()->getPointerAddressSpace() ==
Base->getType()->getPointerAddressSpace() &&
- all_of(CI.users(), IsaPred<ICmpInst>)) {
+ all_of(CI.users(), UsesPointerAsInt)) {
return GetElementPtrInst::Create(Builder.getInt8Ty(), Base, Offset);
}
diff --git a/llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll b/llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll
index e9fd6e5a07dd0..233ab328d7bbb 100644
--- a/llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll
+++ b/llvm/test/Transforms/InstCombine/inttoptr-add-phi.ll
@@ -7,13 +7,13 @@ define i64 @inttoptr_used_by_phi_with_ptrtoint(i1 %c, ptr %src, ptr %p2) {
; CHECK-SAME: i1 [[C:%.*]], ptr [[SRC:%.*]], ptr [[P2:%.*]]) {
; CHECK-NEXT: br i1 [[C]], label %[[THEN:.*]], label %[[ELSE:.*]]
; CHECK: [[THEN]]:
-; CHECK-NEXT: [[I:%.*]] = ptrtoint ptr [[SRC]] to i64
-; CHECK-NEXT: [[A:%.*]] = add i64 [[I]], 10
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr [[SRC]], i64 10
+; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[P]] to i64
; CHECK-NEXT: br label %[[EXIT:.*]]
; CHECK: [[ELSE]]:
; CHECK-NEXT: br label %[[EXIT]]
; CHECK: [[EXIT]]:
-; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[A]], %[[THEN]] ], [ 0, %[[ELSE]] ]
+; CHECK-NEXT: [[PHI:%.*]] = phi i64 [ [[TMP1]], %[[THEN]] ], [ 0, %[[ELSE]] ]
; CHECK-NEXT: ret i64 [[PHI]]
;
%i = ptrtoint ptr %src to i64
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/std-find.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/std-find.ll
index fa25af5f3467d..0ef21adff74a6 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/std-find.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/std-find.ll
@@ -9,27 +9,34 @@ define i64 @std_find_i16_constant_offset_with_assumptions(ptr %first.coerce, i16
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: call void @llvm.assume(i1 true) [ "align"(ptr [[FIRST_COERCE]], i64 2) ]
; CHECK-NEXT: call void @llvm.assume(i1 true) [ "dereferenceable"(ptr [[FIRST_COERCE]], i64 256) ]
-; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[FIRST_COERCE]] to i64
-; CHECK-NEXT: [[COERCE_VAL_PI_I:%.*]] = add i64 [[TMP0]], 256
-; CHECK-NEXT: [[COERCE_VAL_IP:%.*]] = inttoptr i64 [[COERCE_VAL_PI_I]] to ptr
-; CHECK-NEXT: [[CMP_NOT6_I_I:%.*]] = icmp eq ptr [[FIRST_COERCE]], [[COERCE_VAL_IP]]
-; CHECK-NEXT: br i1 [[CMP_NOT6_I_I]], label %[[RETURN:.*]], label %[[LOOP_HEADER:.*]]
-; CHECK: [[LOOP_HEADER]]:
-; CHECK-NEXT: [[PTR_IV:%.*]] = phi ptr [ [[PTR_IV_NEXT:%.*]], %[[LOOP_LATCH:.*]] ], [ [[FIRST_COERCE]], %[[ENTRY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = load i16, ptr [[PTR_IV]], align 2
-; CHECK-NEXT: [[CMP2_I_I:%.*]] = icmp eq i16 [[TMP1]], [[S]]
-; CHECK-NEXT: br i1 [[CMP2_I_I]], label %[[RETURN_LOOPEXIT:.*]], label %[[LOOP_LATCH]]
-; CHECK: [[LOOP_LATCH]]:
-; CHECK-NEXT: [[PTR_IV_NEXT]] = getelementptr inbounds nuw i8, ptr [[PTR_IV]], i64 2
-; CHECK-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq ptr [[PTR_IV_NEXT]], [[COERCE_VAL_IP]]
-; CHECK-NEXT: br i1 [[CMP_NOT_I_I]], label %[[RETURN_LOOPEXIT]], label %[[LOOP_HEADER]]
-; CHECK: [[RETURN_LOOPEXIT]]:
-; CHECK-NEXT: [[MERGE_PH:%.*]] = phi ptr [ [[COERCE_VAL_IP]], %[[LOOP_LATCH]] ], [ [[PTR_IV]], %[[LOOP_HEADER]] ]
-; CHECK-NEXT: [[DOTPRE:%.*]] = ptrtoint ptr [[MERGE_PH]] to i64
+; CHECK-NEXT: [[COERCE_VAL_IP:%.*]] = getelementptr i8, ptr [[FIRST_COERCE]], i64 256
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <8 x i16> poison, i16 [[S]], i64 0
+; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <8 x i16> [[BROADCAST_SPLATINSERT]], <8 x i16> poison, <8 x i32> zeroinitializer
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[FIRST_COERCE]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[NEXT_GEP]], align 2
+; CHECK-NEXT: [[TMP0:%.*]] = icmp eq <8 x i16> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
+; CHECK-NEXT: [[TMP1:%.*]] = bitcast <8 x i1> [[TMP0]] to i8
+; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i8 [[TMP1]], 0
+; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 128
+; CHECK-NEXT: [[TMP4:%.*]] = or i1 [[TMP2]], [[TMP3]]
+; CHECK-NEXT: br i1 [[TMP4]], label %[[MIDDLE_SPLIT:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: [[MIDDLE_SPLIT]]:
+; CHECK-NEXT: br i1 [[TMP2]], label %[[VECTOR_EARLY_EXIT:.*]], label %[[RETURN:.*]]
+; CHECK: [[VECTOR_EARLY_EXIT]]:
+; CHECK-NEXT: [[TMP5:%.*]] = tail call i64 @llvm.experimental.cttz.elts.i64.v8i1(<8 x i1> [[TMP0]], i1 true)
+; CHECK-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], [[TMP5]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl i64 [[TMP6]], 1
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[FIRST_COERCE]], i64 [[TMP7]]
; CHECK-NEXT: br label %[[RETURN]]
; CHECK: [[RETURN]]:
-; CHECK-NEXT: [[RES_PRE_PHI:%.*]] = phi i64 [ [[DOTPRE]], %[[RETURN_LOOPEXIT]] ], [ [[TMP0]], %[[ENTRY]] ]
-; CHECK-NEXT: ret i64 [[RES_PRE_PHI]]
+; CHECK-NEXT: [[__FIRST_ADDR_0_LCSSA_I_I_PH:%.*]] = phi ptr [ [[TMP8]], %[[VECTOR_EARLY_EXIT]] ], [ [[COERCE_VAL_IP]], %[[MIDDLE_SPLIT]] ]
+; CHECK-NEXT: [[DOTPRE:%.*]] = ptrtoint ptr [[__FIRST_ADDR_0_LCSSA_I_I_PH]] to i64
+; CHECK-NEXT: ret i64 [[DOTPRE]]
;
entry:
%first = alloca { ptr }, align 8
@@ -71,27 +78,21 @@ define i64 @std_find_i16_constant_offset_no_assumptions(ptr %first.coerce, i16 n
; CHECK-LABEL: define i64 @std_find_i16_constant_offset_no_assumptions(
; CHECK-SAME: ptr [[FIRST_COERCE:%.*]], i16 noundef signext [[S:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] {
; CHECK-NEXT: [[ENTRY:.*]]:
-; CHECK-NEXT: [[TMP0:%.*]] = ptrtoint ptr [[FIRST_COERCE]] to i64
-; CHECK-NEXT: [[COERCE_VAL_PI_I:%.*]] = add i64 [[TMP0]], 256
-; CHECK-NEXT: [[COERCE_VAL_IP:%.*]] = inttoptr i64 [[COERCE_VAL_PI_I]] to ptr
-; CHECK-NEXT: [[CMP_NOT6_I_I:%.*]] = icmp eq ptr [[FIRST_COERCE]], [[COERCE_VAL_IP]]
-; CHECK-NEXT: br i1 [[CMP_NOT6_I_I]], label %[[RETURN:.*]], label %[[LOOP_HEADER:.*]]
+; CHECK-NEXT: [[COERCE_VAL_IP:%.*]] = getelementptr i8, ptr [[FIRST_COERCE]], i64 256
+; CHECK-NEXT: br label %[[LOOP_HEADER:.*]]
; CHECK: [[LOOP_HEADER]]:
-; CHECK-NEXT: [[PTR_IV:%.*]] = phi ptr [ [[PTR_IV_NEXT:%.*]], %[[LOOP_LATCH:.*]] ], [ [[FIRST_COERCE]], %[[ENTRY]] ]
+; CHECK-NEXT: [[PTR_IV:%.*]] = phi ptr [ [[FIRST_COERCE]], %[[ENTRY]] ], [ [[PTR_IV_NEXT:%.*]], %[[LOOP_LATCH:.*]] ]
; CHECK-NEXT: [[TMP1:%.*]] = load i16, ptr [[PTR_IV]], align 2
; CHECK-NEXT: [[CMP2_I_I:%.*]] = icmp eq i16 [[TMP1]], [[S]]
-; CHECK-NEXT: br i1 [[CMP2_I_I]], label %[[RETURN_LOOPEXIT:.*]], label %[[LOOP_LATCH]]
+; CHECK-NEXT: br i1 [[CMP2_I_I]], label %[[RETURN:.*]], label %[[LOOP_LATCH]]
; CHECK: [[LOOP_LATCH]]:
; CHECK-NEXT: [[PTR_IV_NEXT]] = getelementptr inbounds nuw i8, ptr [[PTR_IV]], i64 2
; CHECK-NEXT: [[CMP_NOT_I_I:%.*]] = icmp eq ptr [[PTR_IV_NEXT]], [[COERCE_VAL_IP]]
-; CHECK-NEXT: br i1 [[CMP_NOT_I_I]], label %[[RETURN_LOOPEXIT]], label %[[LOOP_HEADER]]
-; CHECK: [[RETURN_LOOPEXIT]]:
+; CHECK-NEXT: br i1 [[CMP_NOT_I_I]], label %[[RETURN]], label %[[LOOP_HEADER]]
+; CHECK: [[RETURN]]:
; CHECK-NEXT: [[MERGE_PH:%.*]] = phi ptr [ [[COERCE_VAL_IP]], %[[LOOP_LATCH]] ], [ [[PTR_IV]], %[[LOOP_HEADER]] ]
; CHECK-NEXT: [[DOTPRE:%.*]] = ptrtoint ptr [[MERGE_PH]] to i64
-; CHECK-NEXT: br label %[[RETURN]]
-; CHECK: [[RETURN]]:
-; CHECK-NEXT: [[RES_PRE_PHI:%.*]] = phi i64 [ [[DOTPRE]], %[[RETURN_LOOPEXIT]] ], [ [[TMP0]], %[[ENTRY]] ]
-; CHECK-NEXT: ret i64 [[RES_PRE_PHI]]
+; CHECK-NEXT: ret i64 [[DOTPRE]]
;
entry:
%first = alloca { ptr }, align 8
@@ -128,3 +129,8 @@ return:
}
declare void @llvm.assume(i1 noundef)
+;.
+; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
+; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
+; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
+;.
More information about the llvm-commits
mailing list