Skip to content

[MemProf] Add remarks for matched allocs and calls #170379

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
teresajohnson wants to merge 1 commit into
base: main
Choose a base branch
from
+32 −4
Open

[MemProf] Add remarks for matched allocs and calls #170379

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 17 additions & 2 deletions llvm/lib/Transforms/Instrumentation/MemProfUse.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -454,6 +454,15 @@ handleAllocSite(Instruction &I, CallBase *CI,
InlinedCallStack.size())] = {
AllocInfo->Info.getTotalSize(), AllocType};
}
ORE.emit(
OptimizationRemark(DEBUG_TYPE, "MemprofContext", CI)
Copy link

@snehasish snehasish Dec 5, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What is the significance of "MemprofContext" here?

<< ore::NV("AllocationCall", CI) << " in function "
<< ore::NV("Caller", CI->getFunction())
<< " matched alloc context with alloc type "
<< ore::NV("Attribute", getAllocTypeAttributeString(AllocType))
<< " total size " << ore::NV("Size", AllocInfo->Info.getTotalSize())
<< " full context id " << ore::NV("Context", FullStackId)
<< " frame count " << ore::NV("Frames", InlinedCallStack.size()));
}
}
// If the threshold for the percent of cold bytes is less than 100%,
Expand Down Expand Up @@ -516,7 +525,8 @@ static void handleCallSite(
Instruction &I, const Function *CalledFunction,
ArrayRef<uint64_t> InlinedCallStack,
const std::unordered_set<CallSiteEntry, CallSiteEntryHash> &CallSiteEntries,
Module &M, std::set<std::vector<uint64_t>> &MatchedCallSites) {
Module &M, std::set<std::vector<uint64_t>> &MatchedCallSites,
OptimizationRemarkEmitter &ORE) {
auto &Ctx = M.getContext();
for (const auto &CallSiteEntry : CallSiteEntries) {
// If we found and thus matched all frames on the call, create and
Expand All @@ -539,6 +549,11 @@ static void handleCallSite(
append_range(CallStack, InlinedCallStack);
MatchedCallSites.insert(std::move(CallStack));
}
ORE.emit(OptimizationRemark(DEBUG_TYPE, "MemprofContext", &I)
<< ore::NV("CallSite", &I) << " in function "
<< ore::NV("Caller", I.getFunction())
<< " matched callsite with frame count "
<< ore::NV("Frames", InlinedCallStack.size()));
break;
}
}
Expand Down Expand Up @@ -719,7 +734,7 @@ static void readMemprof(Module &M, Function &F,
// instruction's leaf location in the callsites map and not the
// allocation map.
handleCallSite(I, CalledFunction, InlinedCallStack,
CallSitesIter->second, M, MatchedCallSites);
CallSitesIter->second, M, MatchedCallSites, ORE);
}
}
}
Expand Down
9 changes: 8 additions & 1 deletion llvm/test/Transforms/PGOProfile/memprof-dump-matched-alloc-site.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@
; REQUIRES: x86_64-linux
; RUN: split-file %s %t
; RUN: llvm-profdata merge %t/memprof-dump-matched-alloc-site.yaml -o %t/memprof-dump-matched-alloc-site.memprofdata
; RUN: opt < %t/memprof-dump-matched-alloc-site.ll -passes='memprof-use<profile-filename=%t/memprof-dump-matched-alloc-site.memprofdata>' -memprof-print-match-info -S 2>&1 | FileCheck %s
; RUN: opt < %t/memprof-dump-matched-alloc-site.ll -passes='memprof-use<profile-filename=%t/memprof-dump-matched-alloc-site.memprofdata>' -memprof-print-match-info -S -pass-remarks=memprof 2>&1 | FileCheck %s

;--- memprof-dump-matched-alloc-site.yaml
---
Expand Down Expand Up @@ -77,6 +77,13 @@ HeapProfileRecords:
# Kept empty here because this section is irrelevant for this test.
...
;--- memprof-dump-matched-alloc-site.ll

;; From -pass-remarks=memprof
; CHECK: remark: memprof-dump-matched-alloc-site.cc:1:21: call in function _Z2f1v matched alloc context with alloc type notcold total size 3 full context id 5736731103568718490 frame count 1
; CHECK: remark: memprof-dump-matched-alloc-site.cc:1:21: call in function _Z2f2v matched alloc context with alloc type notcold total size 3 full context id 5736731103568718490 frame count 2
; CHECK: remark: memprof-dump-matched-alloc-site.cc:1:21: call in function _Z2f3v matched alloc context with alloc type notcold total size 3 full context id 5736731103568718490 frame count 3

;; From -memprof-print-match-info
; CHECK: MemProf notcold context with id 5736731103568718490 has total profiled size 3 is matched with 1 frames
; CHECK: MemProf notcold context with id 5736731103568718490 has total profiled size 3 is matched with 2 frames
; CHECK: MemProf notcold context with id 5736731103568718490 has total profiled size 3 is matched with 3 frames
Expand Down
8 changes: 7 additions & 1 deletion llvm/test/Transforms/PGOProfile/memprof-dump-matched-call-sites.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@
; REQUIRES: x86_64-linux
; RUN: split-file %s %t
; RUN: llvm-profdata merge %t/memprof-dump-matched-call-site.yaml -o %t/memprof-dump-matched-call-site.memprofdata
; RUN: opt < %t/memprof-dump-matched-call-site.ll -passes='memprof-use<profile-filename=%t/memprof-dump-matched-call-site.memprofdata>' -memprof-print-match-info -S 2>&1 | FileCheck %s
; RUN: opt < %t/memprof-dump-matched-call-site.ll -passes='memprof-use<profile-filename=%t/memprof-dump-matched-call-site.memprofdata>' -memprof-print-match-info -S -pass-remarks=memprof 2>&1 | FileCheck %s

;--- memprof-dump-matched-call-site.yaml
---
Expand Down Expand Up @@ -71,6 +71,12 @@ HeapProfileRecords:
CallSites: []
...
;--- memprof-dump-matched-call-site.ll

;; From -pass-remarks=memprof
; CHECK: remark: match.cc:18:3: call in function main matched callsite with frame count 1
; CHECK: remark: match.cc:13:28: call in function _ZL2f1v matched callsite with frame count 2

;; From -memprof-print-match-info
; CHECK: MemProf notcold context with id 3894143216621363392 has total profiled size 4 is matched with 1 frames
; CHECK: MemProf callsite match for inline call stack 4745611964195289084 10616861955219347331
; CHECK: MemProf callsite match for inline call stack 5401059281181789382
Expand Down