[Inliner] Preserve alignment of byval arguments #137455
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thank you for submitting a Pull Request (PR) to the LLVM Project! This PR will be automatically labeled and the relevant teams will be notified. If you wish to, you can add reviewers by using the "Reviewers" section on this page. If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers. If you have further questions, they may be answered by the LLVM GitHub User Guide. You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums. |
|
@llvm/pr-subscribers-llvm-transforms Author: None (sallto) ChangesPreviously the inliner always produced a memcpy with alignment 1 for src and destination, leading to potentially suboptimal Codegen. Since the Src ptr alignment is only available through the CallBase it has to be passed to HandleByValArgumentInit. Dst Alignment is already known so it doesn't have to be passed along. If there is no specified Src Alignment my changes cause the ptr to have no align data attached instead of align 1 as before (see inline-tail.ll), I believe this is fine but since I'm a first time contributor, please confirm. My changes are already covered by 4 existing regression tests, so I did not add any additional ones. The example from #45778 now results in: opt -S -passes=inline,instcombine,sroa,instcombine test.ll
define dso_local i32 @<!-- -->test(ptr %t) {
entry:
%.sroa.0.0.copyload = load ptr, ptr %t, align 8 # this used to be align 1 in the original issue
%arrayidx.i = getelementptr inbounds nuw i8, ptr %.sroa.0.0.copyload, i64 24
%0 = load i32, ptr %arrayidx.i, align 4
ret i32 %0
}Full diff: https://github.com/llvm/llvm-project/pull/137455.diff 6 Files Affected:
diff --git a/llvm/lib/Transforms/Utils/InlineFunction.cpp b/llvm/lib/Transforms/Utils/InlineFunction.cpp
index 295518fef687d..7e785c0e019e7 100644
--- a/llvm/lib/Transforms/Utils/InlineFunction.cpp
+++ b/llvm/lib/Transforms/Utils/InlineFunction.cpp
@@ -1703,7 +1703,8 @@ static void AddAlignmentAssumptions(CallBase &CB, InlineFunctionInfo &IFI) {
}
static void HandleByValArgumentInit(Type *ByValType, Value *Dst, Value *Src,
- Module *M, BasicBlock *InsertBlock,
+ MaybeAlign SrcAlign, Module *M,
+ BasicBlock *InsertBlock,
InlineFunctionInfo &IFI,
Function *CalledFunc) {
IRBuilder<> Builder(InsertBlock, InsertBlock->begin());
@@ -1711,11 +1712,12 @@ static void HandleByValArgumentInit(Type *ByValType, Value *Dst, Value *Src,
Value *Size =
Builder.getInt64(M->getDataLayout().getTypeStoreSize(ByValType));
- // Always generate a memcpy of alignment 1 here because we don't know
- // the alignment of the src pointer. Other optimizations can infer
- // better alignment.
- CallInst *CI = Builder.CreateMemCpy(Dst, /*DstAlign*/ Align(1), Src,
- /*SrcAlign*/ Align(1), Size);
+ Align DstAlign = Dst->getPointerAlignment(M->getDataLayout());
+
+ // Generate a mempcpy with the correct alignments. At this point, the
+ // alignment of the src pointer is unknown. Therefore its alignment has to be
+ // passed in from the creation of the src ptr.
+ CallInst *CI = Builder.CreateMemCpy(Dst, DstAlign, Src, SrcAlign, Size);
// The verifier requires that all calls of debug-info-bearing functions
// from debug-info-bearing functions have a debug location (for inlining
@@ -2629,9 +2631,12 @@ llvm::InlineResult llvm::InlineFunction(CallBase &CB, InlineFunctionInfo &IFI,
struct ByValInit {
Value *Dst;
Value *Src;
+ MaybeAlign SrcAlign;
Type *Ty;
};
- // Keep a list of pair (dst, src) to emit byval initializations.
+ // Keep a list of tuples (dst, src, src_align) to emit byval
+ // initializations. Src Alignment is only available though the callbase,
+ // therefore has to be saved.
SmallVector<ByValInit, 4> ByValInits;
// When inlining a function that contains noalias scope metadata,
@@ -2661,8 +2666,9 @@ llvm::InlineResult llvm::InlineFunction(CallBase &CB, InlineFunctionInfo &IFI,
&CB, CalledFunc, IFI,
CalledFunc->getParamAlign(ArgNo));
if (ActualArg != *AI)
- ByValInits.push_back(
- {ActualArg, (Value *)*AI, CB.getParamByValType(ArgNo)});
+ ByValInits.push_back({ActualArg, (Value *)*AI,
+ CB.getParamAlign(ArgNo),
+ CB.getParamByValType(ArgNo)});
}
VMap[&*I] = ActualArg;
@@ -2712,8 +2718,9 @@ llvm::InlineResult llvm::InlineFunction(CallBase &CB, InlineFunctionInfo &IFI,
// Inject byval arguments initialization.
for (ByValInit &Init : ByValInits)
- HandleByValArgumentInit(Init.Ty, Init.Dst, Init.Src, Caller->getParent(),
- &*FirstNewBlock, IFI, CalledFunc);
+ HandleByValArgumentInit(Init.Ty, Init.Dst, Init.Src, Init.SrcAlign,
+ Caller->getParent(), &*FirstNewBlock, IFI,
+ CalledFunc);
std::optional<OperandBundleUse> ParentDeopt =
CB.getOperandBundle(LLVMContext::OB_deopt);
diff --git a/llvm/test/Transforms/Inline/byval-align.ll b/llvm/test/Transforms/Inline/byval-align.ll
index 766094f05be0c..0b135aa570a72 100644
--- a/llvm/test/Transforms/Inline/byval-align.ll
+++ b/llvm/test/Transforms/Inline/byval-align.ll
@@ -29,7 +29,7 @@ define void @byval_caller(ptr nocapture align 64 %a, ptr %b) #0 {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[A1:%.*]] = alloca float, align 128
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 4, ptr [[A1]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[A1]], ptr align 1 [[A]], i64 4, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 128 [[A1]], ptr align 128 [[A]], i64 4, i1 false)
; CHECK-NEXT: [[LOAD_I:%.*]] = load float, ptr [[A1]], align 4
; CHECK-NEXT: [[B_IDX_I:%.*]] = getelementptr inbounds float, ptr [[B]], i64 8
; CHECK-NEXT: [[ADD_I:%.*]] = fadd float [[LOAD_I]], 2.000000e+00
diff --git a/llvm/test/Transforms/Inline/byval-with-non-alloca-addrspace.ll b/llvm/test/Transforms/Inline/byval-with-non-alloca-addrspace.ll
index 42ec0f2bf5699..1d1cb459d53b6 100644
--- a/llvm/test/Transforms/Inline/byval-with-non-alloca-addrspace.ll
+++ b/llvm/test/Transforms/Inline/byval-with-non-alloca-addrspace.ll
@@ -27,7 +27,7 @@ define i64 @foo(ptr %arg) {
; CHECK-SAME: ptr [[ARG:%.*]]) {
; CHECK-NEXT: [[ARG1:%.*]] = alloca [[STRUCT:%.*]], align 8
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 16, ptr [[ARG1]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[ARG1]], ptr align 1 [[ARG]], i64 16, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[ARG1]], ptr align 8 [[ARG]], i64 16, i1 false)
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [[STRUCT]], ptr [[ARG1]], i64 0, i32 1
; CHECK-NEXT: [[TMP2:%.*]] = load i64, ptr [[TMP1]], align 4
; CHECK-NEXT: call void @llvm.lifetime.end.p0(i64 16, ptr [[ARG1]])
diff --git a/llvm/test/Transforms/Inline/byval.ll b/llvm/test/Transforms/Inline/byval.ll
index d98f6e8efa05d..b4a19c55c20a0 100644
--- a/llvm/test/Transforms/Inline/byval.ll
+++ b/llvm/test/Transforms/Inline/byval.ll
@@ -36,7 +36,7 @@ define i32 @test1() nounwind {
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr [[STRUCT_SS]], ptr [[S]], i32 0, i32 1
; CHECK-NEXT: store i64 2, ptr [[TMP4]], align 4
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 12, ptr [[S1]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[S1]], ptr align 1 [[S]], i64 12, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[S1]], ptr [[S]], i64 12, i1 false)
; CHECK-NEXT: [[TMP1_I:%.*]] = load i32, ptr [[S1]], align 4
; CHECK-NEXT: [[TMP2_I:%.*]] = add i32 [[TMP1_I]], 1
; CHECK-NEXT: store i32 [[TMP2_I]], ptr [[S1]], align 4
@@ -105,7 +105,7 @@ define void @test3() nounwind {
; CHECK-NEXT: [[S1:%.*]] = alloca [[STRUCT_SS:%.*]], align 64
; CHECK-NEXT: [[S:%.*]] = alloca [[STRUCT_SS]], align 1
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 12, ptr [[S1]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[S1]], ptr align 1 [[S]], i64 12, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 64 [[S1]], ptr align 64 [[S]], i64 12, i1 false)
; CHECK-NEXT: call void @g3(ptr align 64 [[S1]]) #[[ATTR0]]
; CHECK-NEXT: call void @llvm.lifetime.end.p0(i64 12, ptr [[S1]])
; CHECK-NEXT: ret void
@@ -158,7 +158,7 @@ define i32 @test5() {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[B:%.*]] = alloca [[STRUCT_S0:%.*]], align 8
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 4, ptr [[B]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[B]], ptr align 1 @b, i64 4, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[B]], ptr align 4 @b, i64 4, i1 false)
; CHECK-NEXT: store i32 0, ptr @b, align 4
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[B]], align 4
; CHECK-NEXT: store i32 [[TMP0]], ptr @a, align 4
diff --git a/llvm/test/Transforms/Inline/inline-deferred-instsimplify.ll b/llvm/test/Transforms/Inline/inline-deferred-instsimplify.ll
index f02d03688f039..c74351b300399 100644
--- a/llvm/test/Transforms/Inline/inline-deferred-instsimplify.ll
+++ b/llvm/test/Transforms/Inline/inline-deferred-instsimplify.ll
@@ -53,7 +53,7 @@ define i32 @main() {
; CHECK-LABEL: define i32 @main() {
; CHECK-NEXT: [[G_VAR:%.*]] = alloca [[STRUCT_A:%.*]], align 8
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 20, ptr [[G_VAR]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[G_VAR]], ptr align 1 @g_var, i64 20, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[G_VAR]], ptr align 8 @g_var, i64 20, i1 false)
; CHECK-NEXT: [[VAL_I:%.*]] = load i32, ptr [[G_VAR]], align 8
; CHECK-NEXT: [[DOTNOT_I:%.*]] = icmp eq i32 [[VAL_I]], 0
; CHECK-NEXT: br i1 [[DOTNOT_I]], label [[CHECK_POINTERS_ARE_EQUAL_I:%.*]], label [[STORE_PTR_IN_GVAR_I:%.*]]
diff --git a/llvm/test/Transforms/Inline/inline-tail.ll b/llvm/test/Transforms/Inline/inline-tail.ll
index cbee89c87f17b..0bfd0565eef57 100644
--- a/llvm/test/Transforms/Inline/inline-tail.ll
+++ b/llvm/test/Transforms/Inline/inline-tail.ll
@@ -65,7 +65,7 @@ define void @test_byval_a(ptr byval(i32) %p) {
; CHECK-SAME: (ptr byval(i32) [[P:%.*]]) {
; CHECK-NEXT: [[P1:%.*]] = alloca i32, align 4
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 4, ptr [[P1]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[P1]], ptr align 1 [[P]], i64 4, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[P1]], ptr [[P]], i64 4, i1 false)
; CHECK-NEXT: musttail call void @test_byval_c(ptr byval(i32) [[P1]])
; CHECK-NEXT: ret void
;
@@ -88,7 +88,7 @@ define void @test_dynalloca_a(ptr byval(i32) %p, i32 %n) {
; CHECK-NEXT: [[P1:%.*]] = alloca i32, align 4
; CHECK-NEXT: [[SAVEDSTACK:%.*]] = call ptr @llvm.stacksave.p0()
; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 4, ptr [[P1]])
-; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 1 [[P1]], ptr align 1 [[P]], i64 4, i1 false)
+; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[P1]], ptr [[P]], i64 4, i1 false)
; CHECK-NEXT: [[BUF_I:%.*]] = alloca i8, i32 [[N]], align 1
; CHECK-NEXT: call void @escape(ptr [[BUF_I]])
; CHECK-NEXT: musttail call void @test_dynalloca_c(ptr byval(i32) [[P1]], i32 [[N]])
|
nikic
approved these changes
Apr 26, 2025
| /*SrcAlign*/ Align(1), Size); | ||
| Align DstAlign = Dst->getPointerAlignment(M->getDataLayout()); | ||
|
|
||
| // Generate a mempcpy with the correct alignments. At this point, the |
There was a problem hiding this comment.
Suggested change
| // Generate a mempcpy with the correct alignments. At this point, the | |
| // Generate a memcpy with the correct alignments. At this point, the |
|
|
||
| // Generate a mempcpy with the correct alignments. At this point, the | ||
| // alignment of the src pointer is unknown. Therefore its alignment has to be | ||
| // passed in from the creation of the src ptr. |
There was a problem hiding this comment.
TBH I'd just drop the last two sentences. The alignment isn't unknown if you pass it in...
… parameters passed by value
nikic
changed the title
|
@sallto Congratulations on having your first Pull Request (PR) merged into the LLVM Project! Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR. Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues. How to do this, and the rest of the post-merge process, is covered in detail here. If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again. If you don't get any reports, no action is required from you. Your changes are working as expected, well done! |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/23/builds/9765 Here is the relevant piece of the build log for the reference |
jyli0116
pushed a commit
to jyli0116/llvm-project
that referenced
this pull request
Apr 28, 2025
Previously the inliner always produced a memcpy with alignment 1 for src and destination, leading to potentially suboptimal Codegen. Since the Src ptr alignment is only available through the CallBase it has to be passed to HandleByValArgumentInit. Dst Alignment is already known so it doesn't have to be passed along. If there is no specified Src Alignment my changes cause the ptr to have no align data attached instead of align 1 as before (see inline-tail.ll), I believe this is fine but since I'm a first time contributor, please confirm. My changes are already covered by 4 existing regression tests, so I did not add any additional ones. The example from llvm#45778 now results in: ```C opt -S -passes=inline,instcombine,sroa,instcombine test.ll define dso_local i32 @test(ptr %t) { entry: %.sroa.0.0.copyload = load ptr, ptr %t, align 8 # this used to be align 1 in the original issue %arrayidx.i = getelementptr inbounds nuw i8, ptr %.sroa.0.0.copyload, i64 24 %0 = load i32, ptr %arrayidx.i, align 4 ret i32 %0 } ``` Fixes llvm#45778.
IanWood1
pushed a commit
to IanWood1/llvm-project
that referenced
this pull request
May 6, 2025
Previously the inliner always produced a memcpy with alignment 1 for src and destination, leading to potentially suboptimal Codegen. Since the Src ptr alignment is only available through the CallBase it has to be passed to HandleByValArgumentInit. Dst Alignment is already known so it doesn't have to be passed along. If there is no specified Src Alignment my changes cause the ptr to have no align data attached instead of align 1 as before (see inline-tail.ll), I believe this is fine but since I'm a first time contributor, please confirm. My changes are already covered by 4 existing regression tests, so I did not add any additional ones. The example from llvm#45778 now results in: ```C opt -S -passes=inline,instcombine,sroa,instcombine test.ll define dso_local i32 @test(ptr %t) { entry: %.sroa.0.0.copyload = load ptr, ptr %t, align 8 # this used to be align 1 in the original issue %arrayidx.i = getelementptr inbounds nuw i8, ptr %.sroa.0.0.copyload, i64 24 %0 = load i32, ptr %arrayidx.i, align 4 ret i32 %0 } ``` Fixes llvm#45778.
IanWood1
pushed a commit
to IanWood1/llvm-project
that referenced
this pull request
May 6, 2025
Previously the inliner always produced a memcpy with alignment 1 for src and destination, leading to potentially suboptimal Codegen. Since the Src ptr alignment is only available through the CallBase it has to be passed to HandleByValArgumentInit. Dst Alignment is already known so it doesn't have to be passed along. If there is no specified Src Alignment my changes cause the ptr to have no align data attached instead of align 1 as before (see inline-tail.ll), I believe this is fine but since I'm a first time contributor, please confirm. My changes are already covered by 4 existing regression tests, so I did not add any additional ones. The example from llvm#45778 now results in: ```C opt -S -passes=inline,instcombine,sroa,instcombine test.ll define dso_local i32 @test(ptr %t) { entry: %.sroa.0.0.copyload = load ptr, ptr %t, align 8 # this used to be align 1 in the original issue %arrayidx.i = getelementptr inbounds nuw i8, ptr %.sroa.0.0.copyload, i64 24 %0 = load i32, ptr %arrayidx.i, align 4 ret i32 %0 } ``` Fixes llvm#45778.
IanWood1
pushed a commit
to IanWood1/llvm-project
that referenced
this pull request
May 6, 2025
Previously the inliner always produced a memcpy with alignment 1 for src and destination, leading to potentially suboptimal Codegen. Since the Src ptr alignment is only available through the CallBase it has to be passed to HandleByValArgumentInit. Dst Alignment is already known so it doesn't have to be passed along. If there is no specified Src Alignment my changes cause the ptr to have no align data attached instead of align 1 as before (see inline-tail.ll), I believe this is fine but since I'm a first time contributor, please confirm. My changes are already covered by 4 existing regression tests, so I did not add any additional ones. The example from llvm#45778 now results in: ```C opt -S -passes=inline,instcombine,sroa,instcombine test.ll define dso_local i32 @test(ptr %t) { entry: %.sroa.0.0.copyload = load ptr, ptr %t, align 8 # this used to be align 1 in the original issue %arrayidx.i = getelementptr inbounds nuw i8, ptr %.sroa.0.0.copyload, i64 24 %0 = load i32, ptr %arrayidx.i, align 4 ret i32 %0 } ``` Fixes llvm#45778.
Ankur-0429
pushed a commit
to Ankur-0429/llvm-project
that referenced
this pull request
May 9, 2025
Previously the inliner always produced a memcpy with alignment 1 for src and destination, leading to potentially suboptimal Codegen. Since the Src ptr alignment is only available through the CallBase it has to be passed to HandleByValArgumentInit. Dst Alignment is already known so it doesn't have to be passed along. If there is no specified Src Alignment my changes cause the ptr to have no align data attached instead of align 1 as before (see inline-tail.ll), I believe this is fine but since I'm a first time contributor, please confirm. My changes are already covered by 4 existing regression tests, so I did not add any additional ones. The example from llvm#45778 now results in: ```C opt -S -passes=inline,instcombine,sroa,instcombine test.ll define dso_local i32 @test(ptr %t) { entry: %.sroa.0.0.copyload = load ptr, ptr %t, align 8 # this used to be align 1 in the original issue %arrayidx.i = getelementptr inbounds nuw i8, ptr %.sroa.0.0.copyload, i64 24 %0 = load i32, ptr %arrayidx.i, align 4 ret i32 %0 } ``` Fixes llvm#45778.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Previously the inliner always produced a memcpy with alignment 1 for src and destination, leading to potentially suboptimal Codegen.
Since the Src ptr alignment is only available through the CallBase it has to be passed to HandleByValArgumentInit. Dst Alignment is already known so it doesn't have to be passed along.
If there is no specified Src Alignment my changes cause the ptr to have no align data attached instead of align 1 as before (see inline-tail.ll), I believe this is fine but since I'm a first time contributor, please confirm.
My changes are already covered by 4 existing regression tests, so I did not add any additional ones.
The example from #45778 now results in:
Fixes #45778.