[GlobalMerge]Prefer use global-merge-max-offset instead of the target-specific constant offset. #165591
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
…stant offset. In the Dhrystone benchmark, I find some adjacent global not be merged, on the contrary the GCC's anchor optimize is work. Use global-merge-max-offset to set the max offset can yield similar results (still slightly different)
davemgreen
approved these changes
Nov 13, 2025
Collaborator
davemgreen
left a comment
davemgreen
left a comment
There was a problem hiding this comment.
Sounds OK to me. Could we add a test to show the option overriding the target offset?
llvm/lib/CodeGen/GlobalMerge.cpp
Outdated
| ? GlobalMergeAllConst | ||
| : MergeConstAggressiveByDefault; | ||
| return new GlobalMerge(TM, Offset, OnlyOptimizeForSize, MergeExternal, | ||
| unsigned PreferOffset = GlobalMergeMaxOffset ? GlobalMergeMaxOffset : Offset; |
Collaborator
There was a problem hiding this comment.
Could use GlobalMergeOffset.getNumOccurrences() > 0
davemgreen
reviewed
Nov 13, 2025
Collaborator
davemgreen
left a comment
davemgreen
left a comment
There was a problem hiding this comment.
Actually, could we delete this code too? I don't think it is ever used.
explicit GlobalMerge() : FunctionPass(ID) {
Opt.MaxOffset = GlobalMergeMaxOffset;
Opt.MergeConstantGlobals = EnableGlobalMergeOnConst;
Opt.MergeConstAggressive = GlobalMergeAllConst;
initializeGlobalMergePass(*PassRegistry::getPassRegistry());
}
Member
|
@llvm/pr-subscribers-backend-arm @llvm/pr-subscribers-backend-aarch64 Author: None (hstk30-hw) ChangesIn the Dhrystone benchmark, I find some adjacent global not be merged, on the contrary the GCC's anchor optimize is work. Use global-merge-max-offset to set the max offset can yield similar results (still slightly different, at least we can control the offset). Full diff: https://github.com/llvm/llvm-project/pull/165591.diff 3 Files Affected:
diff --git a/llvm/lib/CodeGen/GlobalMerge.cpp b/llvm/lib/CodeGen/GlobalMerge.cpp
index e58d7e344c28b..0f398b375d799 100644
--- a/llvm/lib/CodeGen/GlobalMerge.cpp
+++ b/llvm/lib/CodeGen/GlobalMerge.cpp
@@ -196,13 +196,6 @@ class GlobalMerge : public FunctionPass {
public:
static char ID; // Pass identification, replacement for typeid.
- explicit GlobalMerge() : FunctionPass(ID) {
- Opt.MaxOffset = GlobalMergeMaxOffset;
- Opt.MergeConstantGlobals = EnableGlobalMergeOnConst;
- Opt.MergeConstAggressive = GlobalMergeAllConst;
- initializeGlobalMergePass(*PassRegistry::getPassRegistry());
- }
-
explicit GlobalMerge(const TargetMachine *TM, unsigned MaximalOffset,
bool OnlyOptimizeForSize, bool MergeExternalGlobals,
bool MergeConstantGlobals, bool MergeConstAggressive)
@@ -772,6 +765,8 @@ Pass *llvm::createGlobalMergePass(const TargetMachine *TM, unsigned Offset,
bool MergeConstAggressive = GlobalMergeAllConst.getNumOccurrences() > 0
? GlobalMergeAllConst
: MergeConstAggressiveByDefault;
- return new GlobalMerge(TM, Offset, OnlyOptimizeForSize, MergeExternal,
+ unsigned PreferOffset = GlobalMergeMaxOffset.getNumOccurrences() > 0 ?
+ GlobalMergeMaxOffset : Offset;
+ return new GlobalMerge(TM, PreferOffset, OnlyOptimizeForSize, MergeExternal,
MergeConstant, MergeConstAggressive);
}
diff --git a/llvm/test/CodeGen/AArch64/global-merge.ll b/llvm/test/CodeGen/AArch64/global-merge.ll
index f2826e4cb00cb..723b033720937 100644
--- a/llvm/test/CodeGen/AArch64/global-merge.ll
+++ b/llvm/test/CodeGen/AArch64/global-merge.ll
@@ -1,17 +1,22 @@
; RUN: llc < %s -mtriple=aarch64-none-linux-gnu -O0 | FileCheck --check-prefix=NO-MERGE %s
+; RUN: llc < %s -mtriple=aarch64-none-linux-gnu -O1 | FileCheck --check-prefix=NO-MERGE %s
+; RUN: llc < %s -mtriple=aarch64-none-linux-gnu -O2 | FileCheck --check-prefix=NO-MERGE %s
+; RUN: llc < %s -mtriple=aarch64-none-linux-gnu -O3 | FileCheck %s
+; RUN: llc < %s -mtriple=aarch64-none-linux-gnu -O3 -global-merge-max-offset=0 | FileCheck %s --check-prefix=NO-MERGE
; RUN: llc < %s -mtriple=aarch64-none-linux-gnu -O0 -global-merge-on-external=true | FileCheck --check-prefix=NO-MERGE %s
; RUN: llc < %s -mtriple=aarch64-apple-ios -O0 | FileCheck %s --check-prefix=CHECK-APPLE-IOS-NO-MERGE
+; RUN: llc < %s -mtriple=aarch64-apple-ios -O1 | FileCheck %s --check-prefix=CHECK-APPLE-IOS-NO-MERGE
+; RUN: llc < %s -mtriple=aarch64-apple-ios -O2 | FileCheck %s --check-prefix=CHECK-APPLE-IOS-NO-MERGE
+; RUN: llc < %s -mtriple=aarch64-apple-ios -O3 | FileCheck %s --check-prefix=CHECK-APPLE-IOS
; RUN: llc < %s -mtriple=aarch64-apple-ios -O0 -global-merge-on-external=true | FileCheck %s --check-prefix=CHECK-APPLE-IOS-NO-MERGE
-; FIXME: add O1/O2 test for aarch64-none-linux-gnu and aarch64-apple-ios
-
@m = internal global i32 0, align 4
@n = internal global i32 0, align 4
define void @f1(i32 %a1, i32 %a2) {
; CHECK-LABEL: f1:
-; CHECK: adrp x{{[0-9]+}}, _MergedGlobals
+; CHECK: adrp x{{[0-9]+}}, .L_MergedGlobals
; CHECK-NOT: adrp
; CHECK-APPLE-IOS-LABEL: f1:
@@ -22,9 +27,9 @@ define void @f1(i32 %a1, i32 %a2) {
ret void
}
-; CHECK: .local _MergedGlobals
-; CHECK: .comm _MergedGlobals,8,8
+; CHECK: .local .L_MergedGlobals
+; CHECK: .comm .L_MergedGlobals,8,4
; NO-MERGE-NOT: .local _MergedGlobals
-; CHECK-APPLE-IOS: .zerofill __DATA,__bss,__MergedGlobals,8,3
-; CHECK-APPLE-IOS-NO-MERGE-NOT: .zerofill __DATA,__bss,__MergedGlobals,8,3
+; CHECK-APPLE-IOS: .zerofill __DATA,__bss,__MergedGlobals,8,2
+; CHECK-APPLE-IOS-NO-MERGE-NOT: .zerofill __DATA,__bss,__MergedGlobals,8,2
diff --git a/llvm/test/CodeGen/ARM/global-merge-1.ll b/llvm/test/CodeGen/ARM/global-merge-1.ll
index 05719ae4eb37d..fe42f7e54b192 100644
--- a/llvm/test/CodeGen/ARM/global-merge-1.ll
+++ b/llvm/test/CodeGen/ARM/global-merge-1.ll
@@ -7,6 +7,7 @@
; RUN: llc %s -O3 -o - | FileCheck -check-prefix=MERGE %s
; RUN: llc %s -O3 -o - -arm-global-merge=false | FileCheck -check-prefix=NO-MERGE %s
; RUN: llc %s -O3 -o - -arm-global-merge=true | FileCheck -check-prefix=MERGE %s
+; RUN: llc %s -O3 -o - -arm-global-merge=true -global-merge-max-offse=0 | FileCheck -check-prefix=NO-MERGE %s
; MERGE-NOT: .zerofill __DATA,__bss,_bar,20,2
; MERGE-NOT: .zerofill __DATA,__bss,_baz,20,2
|
|
✅ With the latest revision this PR passed the C/C++ code formatter. |
Contributor
Author
This is for |
Collaborator
I see - perhaps something we can remove when it moves to the NPM. |
davemgreen
approved these changes
Nov 16, 2025
Collaborator
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/174/builds/27572 Here is the relevant piece of the build log for the reference |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
In the Dhrystone benchmark, I find some adjacent global not be merged, on the contrary the GCC's anchor optimize is work. Use global-merge-max-offset to set the max offset can yield similar results (still slightly different, at least we can control the offset).