[llvm-branch-commits] [clang] [llvm] [TySan] A Type Sanitizer (Clang) (PR #76260)
@@ -1441,8 +1441,10 @@ collectSanitizerRuntimes(const ToolChain &TC, const
ArgList &Args,
if (SanArgs.needsScudoRt()) {
SharedRuntimes.push_back("scudo_standalone");
}
-if (SanArgs.needsTsanRt())
+if (SanArgs.needsTsanRt() && SanArgs.linkRuntimes())
fhahn wrote:
Nope, I think this was left over from a merge, removed, thanks!
https://github.com/llvm/llvm-project/pull/76260
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [llvm] [TySan] A Type Sanitizer (Clang) (PR #76260)
https://github.com/fhahn commented: > LGTM with comments addressed. (I haven't looked carefully at the non-clang > parts; this is just approval for the clang changes.) Thanks! Any chance you would be able to help with the review of the LLVM side as well? https://github.com/llvm/llvm-project/pull/76259 https://github.com/llvm/llvm-project/pull/76260 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [llvm] [TySan] A Type Sanitizer (Clang) (PR #76260)
@@ -1441,8 +1441,10 @@ collectSanitizerRuntimes(const ToolChain &TC, const
ArgList &Args,
if (SanArgs.needsScudoRt()) {
SharedRuntimes.push_back("scudo_standalone");
}
-if (SanArgs.needsTsanRt())
+if (SanArgs.needsTsanRt() && SanArgs.linkRuntimes())
SharedRuntimes.push_back("tsan");
+if (SanArgs.needsTysanRt())
+ StaticRuntimes.push_back("tysan");
fhahn wrote:
Fixed, thanks!
https://github.com/llvm/llvm-project/pull/76260
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [llvm] [TySan] A Type Sanitizer (Clang) (PR #76260)
https://github.com/fhahn edited https://github.com/llvm/llvm-project/pull/76260 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [CodeGen][PM] Initialize analyses with isAnalysis=true (PR #118779)
https://github.com/optimisan updated
https://github.com/llvm/llvm-project/pull/118779
>From d48e889915d463cc32f6c5d3e6ce788fb7f44cea Mon Sep 17 00:00:00 2001
From: Akshat Oke
Date: Thu, 5 Dec 2024 10:05:30 +
Subject: [PATCH 1/2] [CodeGen][PM] Initialize analyses with isAnalysis=true
Analyses should be marked as analyses else they are prone to be ignored
by the legacy analysis cache mechanism and get scheduled redundantly.
---
llvm/lib/CodeGen/GCMetadata.cpp| 2 +-
llvm/lib/CodeGen/LiveDebugVariables.cpp| 2 +-
llvm/lib/CodeGen/LiveIntervals.cpp | 2 +-
llvm/lib/CodeGen/LiveRegMatrix.cpp | 2 +-
llvm/lib/CodeGen/LiveStacks.cpp| 2 +-
llvm/lib/CodeGen/SlotIndexes.cpp | 2 +-
llvm/lib/CodeGen/VirtRegMap.cpp| 2 +-
llvm/lib/Transforms/Utils/LoopSimplify.cpp | 4 ++--
8 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/llvm/lib/CodeGen/GCMetadata.cpp b/llvm/lib/CodeGen/GCMetadata.cpp
index 6d1cc1a58e27df..f33008c9e0f232 100644
--- a/llvm/lib/CodeGen/GCMetadata.cpp
+++ b/llvm/lib/CodeGen/GCMetadata.cpp
@@ -66,7 +66,7 @@ GCFunctionAnalysis::run(Function &F, FunctionAnalysisManager
&FAM) {
}
INITIALIZE_PASS(GCModuleInfo, "collector-metadata",
-"Create Garbage Collector Module Metadata", false, false)
+"Create Garbage Collector Module Metadata", false, true)
//
-
diff --git a/llvm/lib/CodeGen/LiveDebugVariables.cpp
b/llvm/lib/CodeGen/LiveDebugVariables.cpp
index 317d3401f000a4..79085e587ebc45 100644
--- a/llvm/lib/CodeGen/LiveDebugVariables.cpp
+++ b/llvm/lib/CodeGen/LiveDebugVariables.cpp
@@ -82,7 +82,7 @@ INITIALIZE_PASS_BEGIN(LiveDebugVariablesWrapperLegacy,
DEBUG_TYPE,
INITIALIZE_PASS_DEPENDENCY(MachineDominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(LiveIntervalsWrapperPass)
INITIALIZE_PASS_END(LiveDebugVariablesWrapperLegacy, DEBUG_TYPE,
-"Debug Variable Analysis", false, false)
+"Debug Variable Analysis", false, true)
void LiveDebugVariablesWrapperLegacy::getAnalysisUsage(
AnalysisUsage &AU) const {
diff --git a/llvm/lib/CodeGen/LiveIntervals.cpp
b/llvm/lib/CodeGen/LiveIntervals.cpp
index f9ee6e4563f8d6..f38527a3ce6a31 100644
--- a/llvm/lib/CodeGen/LiveIntervals.cpp
+++ b/llvm/lib/CodeGen/LiveIntervals.cpp
@@ -83,7 +83,7 @@ INITIALIZE_PASS_BEGIN(LiveIntervalsWrapperPass,
"liveintervals",
INITIALIZE_PASS_DEPENDENCY(MachineDominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(SlotIndexesWrapperPass)
INITIALIZE_PASS_END(LiveIntervalsWrapperPass, "liveintervals",
-"Live Interval Analysis", false, false)
+"Live Interval Analysis", false, true)
bool LiveIntervalsWrapperPass::runOnMachineFunction(MachineFunction &MF) {
LIS.Indexes = &getAnalysis().getSI();
diff --git a/llvm/lib/CodeGen/LiveRegMatrix.cpp
b/llvm/lib/CodeGen/LiveRegMatrix.cpp
index bc8c59381a40e1..9744c47d5a8510 100644
--- a/llvm/lib/CodeGen/LiveRegMatrix.cpp
+++ b/llvm/lib/CodeGen/LiveRegMatrix.cpp
@@ -41,7 +41,7 @@ INITIALIZE_PASS_BEGIN(LiveRegMatrixWrapperLegacy,
"liveregmatrix",
INITIALIZE_PASS_DEPENDENCY(LiveIntervalsWrapperPass)
INITIALIZE_PASS_DEPENDENCY(VirtRegMapWrapperLegacy)
INITIALIZE_PASS_END(LiveRegMatrixWrapperLegacy, "liveregmatrix",
-"Live Register Matrix", false, false)
+"Live Register Matrix", false, true)
void LiveRegMatrixWrapperLegacy::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesAll();
diff --git a/llvm/lib/CodeGen/LiveStacks.cpp b/llvm/lib/CodeGen/LiveStacks.cpp
index 92cc6699f2d331..d615caf48c0ad3 100644
--- a/llvm/lib/CodeGen/LiveStacks.cpp
+++ b/llvm/lib/CodeGen/LiveStacks.cpp
@@ -25,7 +25,7 @@ INITIALIZE_PASS_BEGIN(LiveStacksWrapperLegacy, DEBUG_TYPE,
"Live Stack Slot Analysis", false, false)
INITIALIZE_PASS_DEPENDENCY(SlotIndexesWrapperPass)
INITIALIZE_PASS_END(LiveStacksWrapperLegacy, DEBUG_TYPE,
-"Live Stack Slot Analysis", false, false)
+"Live Stack Slot Analysis", false, true)
char &llvm::LiveStacksID = LiveStacksWrapperLegacy::ID;
diff --git a/llvm/lib/CodeGen/SlotIndexes.cpp b/llvm/lib/CodeGen/SlotIndexes.cpp
index 1b92a5aa59d18c..8d2832b3fdd56e 100644
--- a/llvm/lib/CodeGen/SlotIndexes.cpp
+++ b/llvm/lib/CodeGen/SlotIndexes.cpp
@@ -45,7 +45,7 @@ SlotIndexes::~SlotIndexes() {
}
INITIALIZE_PASS(SlotIndexesWrapperPass, DEBUG_TYPE, "Slot index numbering",
-false, false)
+false, true)
STATISTIC(NumLocalRenum, "Number of local renumberings");
diff --git a/llvm/lib/CodeGen/VirtRegMap.cpp b/llvm/lib/CodeGen/VirtRegMap.cpp
index 1352102a93d01b..b28c74600e7a29 100644
--- a/llvm/lib/CodeGen/VirtRegMap.cpp
+++ b/llvm/lib/CodeGen/VirtRegMap.cpp
@@ -60,7 +60,7 @@ STATISTIC(NumIdCopies, "Number of identity moves eliminated
after rewriting");
char
[llvm-branch-commits] [clang] [Serialization] Code cleanups and polish 83233 (PR #83237)
ilya-biryukov wrote: > So the meaning of stacked PR in this series of patches is pretty > questionable. So I feel it is better to merge these PRs into a single commit > so that it is easier to be cherry-picked and reverted in my experience +100, this is definitely the right call. ... and I wish we had a better way to do stacked PRs, this is definitely one of the downgrades compared to Phabricator. https://github.com/llvm/llvm-project/pull/83237 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [CodeGen][PM] Initialize analyses with isAnalysis=true (PR #118779)
https://github.com/optimisan updated
https://github.com/llvm/llvm-project/pull/118779
>From d48e889915d463cc32f6c5d3e6ce788fb7f44cea Mon Sep 17 00:00:00 2001
From: Akshat Oke
Date: Thu, 5 Dec 2024 10:05:30 +
Subject: [PATCH 1/2] [CodeGen][PM] Initialize analyses with isAnalysis=true
Analyses should be marked as analyses else they are prone to be ignored
by the legacy analysis cache mechanism and get scheduled redundantly.
---
llvm/lib/CodeGen/GCMetadata.cpp| 2 +-
llvm/lib/CodeGen/LiveDebugVariables.cpp| 2 +-
llvm/lib/CodeGen/LiveIntervals.cpp | 2 +-
llvm/lib/CodeGen/LiveRegMatrix.cpp | 2 +-
llvm/lib/CodeGen/LiveStacks.cpp| 2 +-
llvm/lib/CodeGen/SlotIndexes.cpp | 2 +-
llvm/lib/CodeGen/VirtRegMap.cpp| 2 +-
llvm/lib/Transforms/Utils/LoopSimplify.cpp | 4 ++--
8 files changed, 9 insertions(+), 9 deletions(-)
diff --git a/llvm/lib/CodeGen/GCMetadata.cpp b/llvm/lib/CodeGen/GCMetadata.cpp
index 6d1cc1a58e27df..f33008c9e0f232 100644
--- a/llvm/lib/CodeGen/GCMetadata.cpp
+++ b/llvm/lib/CodeGen/GCMetadata.cpp
@@ -66,7 +66,7 @@ GCFunctionAnalysis::run(Function &F, FunctionAnalysisManager
&FAM) {
}
INITIALIZE_PASS(GCModuleInfo, "collector-metadata",
-"Create Garbage Collector Module Metadata", false, false)
+"Create Garbage Collector Module Metadata", false, true)
//
-
diff --git a/llvm/lib/CodeGen/LiveDebugVariables.cpp
b/llvm/lib/CodeGen/LiveDebugVariables.cpp
index 317d3401f000a4..79085e587ebc45 100644
--- a/llvm/lib/CodeGen/LiveDebugVariables.cpp
+++ b/llvm/lib/CodeGen/LiveDebugVariables.cpp
@@ -82,7 +82,7 @@ INITIALIZE_PASS_BEGIN(LiveDebugVariablesWrapperLegacy,
DEBUG_TYPE,
INITIALIZE_PASS_DEPENDENCY(MachineDominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(LiveIntervalsWrapperPass)
INITIALIZE_PASS_END(LiveDebugVariablesWrapperLegacy, DEBUG_TYPE,
-"Debug Variable Analysis", false, false)
+"Debug Variable Analysis", false, true)
void LiveDebugVariablesWrapperLegacy::getAnalysisUsage(
AnalysisUsage &AU) const {
diff --git a/llvm/lib/CodeGen/LiveIntervals.cpp
b/llvm/lib/CodeGen/LiveIntervals.cpp
index f9ee6e4563f8d6..f38527a3ce6a31 100644
--- a/llvm/lib/CodeGen/LiveIntervals.cpp
+++ b/llvm/lib/CodeGen/LiveIntervals.cpp
@@ -83,7 +83,7 @@ INITIALIZE_PASS_BEGIN(LiveIntervalsWrapperPass,
"liveintervals",
INITIALIZE_PASS_DEPENDENCY(MachineDominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(SlotIndexesWrapperPass)
INITIALIZE_PASS_END(LiveIntervalsWrapperPass, "liveintervals",
-"Live Interval Analysis", false, false)
+"Live Interval Analysis", false, true)
bool LiveIntervalsWrapperPass::runOnMachineFunction(MachineFunction &MF) {
LIS.Indexes = &getAnalysis().getSI();
diff --git a/llvm/lib/CodeGen/LiveRegMatrix.cpp
b/llvm/lib/CodeGen/LiveRegMatrix.cpp
index bc8c59381a40e1..9744c47d5a8510 100644
--- a/llvm/lib/CodeGen/LiveRegMatrix.cpp
+++ b/llvm/lib/CodeGen/LiveRegMatrix.cpp
@@ -41,7 +41,7 @@ INITIALIZE_PASS_BEGIN(LiveRegMatrixWrapperLegacy,
"liveregmatrix",
INITIALIZE_PASS_DEPENDENCY(LiveIntervalsWrapperPass)
INITIALIZE_PASS_DEPENDENCY(VirtRegMapWrapperLegacy)
INITIALIZE_PASS_END(LiveRegMatrixWrapperLegacy, "liveregmatrix",
-"Live Register Matrix", false, false)
+"Live Register Matrix", false, true)
void LiveRegMatrixWrapperLegacy::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesAll();
diff --git a/llvm/lib/CodeGen/LiveStacks.cpp b/llvm/lib/CodeGen/LiveStacks.cpp
index 92cc6699f2d331..d615caf48c0ad3 100644
--- a/llvm/lib/CodeGen/LiveStacks.cpp
+++ b/llvm/lib/CodeGen/LiveStacks.cpp
@@ -25,7 +25,7 @@ INITIALIZE_PASS_BEGIN(LiveStacksWrapperLegacy, DEBUG_TYPE,
"Live Stack Slot Analysis", false, false)
INITIALIZE_PASS_DEPENDENCY(SlotIndexesWrapperPass)
INITIALIZE_PASS_END(LiveStacksWrapperLegacy, DEBUG_TYPE,
-"Live Stack Slot Analysis", false, false)
+"Live Stack Slot Analysis", false, true)
char &llvm::LiveStacksID = LiveStacksWrapperLegacy::ID;
diff --git a/llvm/lib/CodeGen/SlotIndexes.cpp b/llvm/lib/CodeGen/SlotIndexes.cpp
index 1b92a5aa59d18c..8d2832b3fdd56e 100644
--- a/llvm/lib/CodeGen/SlotIndexes.cpp
+++ b/llvm/lib/CodeGen/SlotIndexes.cpp
@@ -45,7 +45,7 @@ SlotIndexes::~SlotIndexes() {
}
INITIALIZE_PASS(SlotIndexesWrapperPass, DEBUG_TYPE, "Slot index numbering",
-false, false)
+false, true)
STATISTIC(NumLocalRenum, "Number of local renumberings");
diff --git a/llvm/lib/CodeGen/VirtRegMap.cpp b/llvm/lib/CodeGen/VirtRegMap.cpp
index 1352102a93d01b..b28c74600e7a29 100644
--- a/llvm/lib/CodeGen/VirtRegMap.cpp
+++ b/llvm/lib/CodeGen/VirtRegMap.cpp
@@ -60,7 +60,7 @@ STATISTIC(NumIdCopies, "Number of identity moves eliminated
after rewriting");
char
[llvm-branch-commits] [llvm] [CodeGen][PM] Initialize analyses with isAnalysis=true (PR #118779)
optimisan wrote: ### Merge activity * **Dec 6, 4:43 AM EST**: A user started a stack merge that includes this pull request via [Graphite](https://app.graphite.dev/github/pr/llvm/llvm-project/118779). https://github.com/llvm/llvm-project/pull/118779 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [flang] [lld] [llvm] [Flang] LLVM_ENABLE_RUNTIMES=flang-rt (PR #110217)
Meinersbur wrote: As @jeanPerier correctly pointed out to me, some buildbots would start failing with this patch, at least those running the Test-Suite using Flang will fail since the runtime is needed to correctly exectute applications, but also two builders that build the runtime out-of-tree will fail. I prepared a llvm-zorg patch to adapt the build configurations: https://github.com/llvm/llvm-zorg/pull/333. Once this PR are accepted, I would wait for feedback from buildbots before applying the llvm-zorg patch in order to have a chance to revert if any `check-flang` steps are failing. After https://github.com/llvm/llvm-zorg/pull/333 @gkistanova needs to restart the [buildbot](https://lab.llvm.org/buildbot), so some intermittent buildbot failures seem unavoidable anyway. I also cannot test https://github.com/llvm/llvm-zorg/pull/333 in advance, so some more updates may be required, possibly by the worker's owners. https://github.com/llvm/llvm-project/pull/110217 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][Utils] Extract BuildDebugInfoMDMap from CloneFunctionInto (PR #118622)
@@ -79,14 +79,16 @@ struct ValueMapConfig {
static mutex_type *getMutex(const ExtraDataT &/*Data*/) { return nullptr; }
};
+/// This type stores Metadata. Used in ValueMap.
+using MDMapT = DenseMap;
artempyanykh wrote:
@felipepiovezan good call! Let me just spell out the type completely.
https://github.com/llvm/llvm-project/pull/118622
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][Utils] Extract BuildDebugInfoMDMap from CloneFunctionInto (PR #118622)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118622 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][Utils] Extract CloneFunctionMetadataInto from CloneFunctionInto (PR #118623)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118623 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][Utils] Eliminate DISubprogram set from BuildDebugInfoMDMap (PR #118625)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118625 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [Utils] Identity map global debug info on first use in CloneFunction* (PR #118627)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118627 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [Coro] Prebuild a global debug info set and share it between all coroutine clones (PR #118628)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118628 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][Utils] Extract CloneFunctionBodyInto from CloneFunctionInto (PR #118624)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118624 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [Coro] Use DebugInfoCache to speed up cloning in CoroSplitPass (PR #118630)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118630 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [Analysis] Add DebugInfoCache analysis (PR #118629)
https://github.com/artempyanykh edited https://github.com/llvm/llvm-project/pull/118629 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [Serialization] Code cleanups and polish 83233 (PR #83237)
slydiman wrote: [Commit b5bd192](https://github.com/llvm/llvm-project/commit/b5bd1928c6d43bc525a4e3fb65d2c750d61e) break buildbots https://lab.llvm.org/buildbot/#/builders/195/builds/1985 https://lab.llvm.org/staging/#/builders/197/builds/1161 https://github.com/llvm/llvm-project/pull/83237 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] 0320d20 - Revert "[Clang] Deleting an incomplete enum type is not an error (#118455)"
Author: cor3ntin
Date: 2024-12-06T15:16:34+01:00
New Revision: 0320d20dbdcafec2a090006c9674f642e207db27
URL:
https://github.com/llvm/llvm-project/commit/0320d20dbdcafec2a090006c9674f642e207db27
DIFF:
https://github.com/llvm/llvm-project/commit/0320d20dbdcafec2a090006c9674f642e207db27.diff
LOG: Revert "[Clang] Deleting an incomplete enum type is not an error (#118455)"
This reverts commit 8271195de05742ed7079d7882fbebc2daecbd7e2.
Added:
Modified:
clang/docs/ReleaseNotes.rst
clang/lib/Sema/SemaExprCXX.cpp
clang/test/SemaCXX/new-delete.cpp
Removed:
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 3f58e64cf0ccbc..59e3a6609123d2 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -796,7 +796,6 @@ Bug Fixes to C++ Support
- Fixed an assertion failure caused by mangled names with invalid identifiers.
(#GH112205)
- Fixed an incorrect lambda scope of generic lambdas that caused Clang to
crash when computing potential lambda
captures at the end of a full expression. (#GH115931)
-- Clang no longer rejects deleting a pointer of incomplete enumeration type.
(#GH99278)
Bug Fixes to AST Handling
^
diff --git a/clang/lib/Sema/SemaExprCXX.cpp b/clang/lib/Sema/SemaExprCXX.cpp
index db9ea7fb66e05a..f58c0fa21e8380 100644
--- a/clang/lib/Sema/SemaExprCXX.cpp
+++ b/clang/lib/Sema/SemaExprCXX.cpp
@@ -3747,8 +3747,7 @@ Sema::ActOnCXXDelete(SourceLocation StartLoc, bool
UseGlobal,
} else if (!Pointee->isDependentType()) {
// FIXME: This can result in errors if the definition was imported from a
// module but is hidden.
- if (!Pointee->isStructureOrClassType() ||
- !RequireCompleteType(StartLoc, Pointee,
+ if (!RequireCompleteType(StartLoc, Pointee,
LangOpts.CPlusPlus26
? diag::err_delete_incomplete
: diag::warn_delete_incomplete,
diff --git a/clang/test/SemaCXX/new-delete.cpp
b/clang/test/SemaCXX/new-delete.cpp
index 98b168d9df1f37..595bdc689d694b 100644
--- a/clang/test/SemaCXX/new-delete.cpp
+++ b/clang/test/SemaCXX/new-delete.cpp
@@ -540,13 +540,6 @@ namespace PR10504 {
void f(A *x) { delete x; } // expected-warning {{delete called on
'PR10504::A' that is abstract but has non-virtual destructor}}
}
-#if __cplusplus >= 201103L
-enum GH99278_1 {
-zero = decltype(delete static_cast(nullptr), 0){}
-// expected-warning@-1 {{expression with side effects has no effect in an
unevaluated context}}
-};
-#endif
-
struct PlacementArg {};
inline void *operator new[](size_t, const PlacementArg &) throw () {
return 0;
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC] Remove adhoc definition of MDMapT in IRMover (PR #118626)
artempyanykh wrote: Abandoned due to a change in #118622. https://github.com/llvm/llvm-project/pull/118626 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC] Remove adhoc definition of MDMapT in IRMover (PR #118626)
https://github.com/artempyanykh closed https://github.com/llvm/llvm-project/pull/118626 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [flang] [flang][cuda] Handle gpu.return in AbstractResult pass (PR #119035)
llvmbot wrote:
@llvm/pr-subscribers-flang-fir-hlfir
Author: Valentin Clement (バレンタイン クレメン) (clementval)
Changes
Process `gpu.return` in AbstractResult pass when the function is a `gpu.func`.
---
Full diff: https://github.com/llvm/llvm-project/pull/119035.diff
2 Files Affected:
- (modified) flang/lib/Optimizer/Transforms/AbstractResult.cpp (+74-49)
- (added) flang/test/Fir/CUDA/cuda-abstract-result.mlir (+37)
``diff
diff --git a/flang/lib/Optimizer/Transforms/AbstractResult.cpp
b/flang/lib/Optimizer/Transforms/AbstractResult.cpp
index 2ed66cc83eefb5..b0327cc10e9de6 100644
--- a/flang/lib/Optimizer/Transforms/AbstractResult.cpp
+++ b/flang/lib/Optimizer/Transforms/AbstractResult.cpp
@@ -234,6 +234,60 @@ class SaveResultOpConversion
}
};
+template
+static mlir::LogicalResult
+processReturnLikeOp(OpTy ret, mlir::Value newArg,
+mlir::PatternRewriter &rewriter) {
+ auto loc = ret.getLoc();
+ rewriter.setInsertionPoint(ret);
+ mlir::Value resultValue = ret.getOperand(0);
+ fir::LoadOp resultLoad;
+ mlir::Value resultStorage;
+ // Identify result local storage.
+ if (auto load = resultValue.getDefiningOp()) {
+resultLoad = load;
+resultStorage = load.getMemref();
+// The result alloca may be behind a fir.declare, if any.
+if (auto declare = resultStorage.getDefiningOp())
+ resultStorage = declare.getMemref();
+ }
+ // Replace old local storage with new storage argument, unless
+ // the derived type is C_PTR/C_FUN_PTR, in which case the return
+ // type is updated to return void* (no new argument is passed).
+ if (fir::isa_builtin_cptr_type(resultValue.getType())) {
+auto module = ret->template getParentOfType();
+FirOpBuilder builder(rewriter, module);
+mlir::Value cptr = resultValue;
+if (resultLoad) {
+ // Replace whole derived type load by component load.
+ cptr = resultLoad.getMemref();
+ rewriter.setInsertionPoint(resultLoad);
+}
+mlir::Value newResultValue =
+fir::factory::genCPtrOrCFunptrValue(builder, loc, cptr);
+newResultValue = builder.createConvert(
+loc, getVoidPtrType(ret.getContext()), newResultValue);
+rewriter.setInsertionPoint(ret);
+rewriter.replaceOpWithNewOp(ret, mlir::ValueRange{newResultValue});
+ } else if (resultStorage) {
+resultStorage.replaceAllUsesWith(newArg);
+rewriter.replaceOpWithNewOp(ret);
+ } else {
+// The result storage may have been optimized out by a memory to
+// register pass, this is possible for fir.box results, or fir.record
+// with no length parameters. Simply store the result in the result
+// storage. at the return point.
+rewriter.create(loc, resultValue, newArg);
+rewriter.replaceOpWithNewOp(ret);
+ }
+ // Delete result old local storage if unused.
+ if (resultStorage)
+if (auto alloc = resultStorage.getDefiningOp())
+ if (alloc->use_empty())
+rewriter.eraseOp(alloc);
+ return mlir::success();
+}
+
class ReturnOpConversion : public mlir::OpRewritePattern
{
public:
using OpRewritePattern::OpRewritePattern;
@@ -242,55 +296,23 @@ class ReturnOpConversion : public
mlir::OpRewritePattern {
llvm::LogicalResult
matchAndRewrite(mlir::func::ReturnOp ret,
mlir::PatternRewriter &rewriter) const override {
-auto loc = ret.getLoc();
-rewriter.setInsertionPoint(ret);
-mlir::Value resultValue = ret.getOperand(0);
-fir::LoadOp resultLoad;
-mlir::Value resultStorage;
-// Identify result local storage.
-if (auto load = resultValue.getDefiningOp()) {
- resultLoad = load;
- resultStorage = load.getMemref();
- // The result alloca may be behind a fir.declare, if any.
- if (auto declare = resultStorage.getDefiningOp())
-resultStorage = declare.getMemref();
-}
-// Replace old local storage with new storage argument, unless
-// the derived type is C_PTR/C_FUN_PTR, in which case the return
-// type is updated to return void* (no new argument is passed).
-if (fir::isa_builtin_cptr_type(resultValue.getType())) {
- auto module = ret->getParentOfType();
- FirOpBuilder builder(rewriter, module);
- mlir::Value cptr = resultValue;
- if (resultLoad) {
-// Replace whole derived type load by component load.
-cptr = resultLoad.getMemref();
-rewriter.setInsertionPoint(resultLoad);
- }
- mlir::Value newResultValue =
- fir::factory::genCPtrOrCFunptrValue(builder, loc, cptr);
- newResultValue = builder.createConvert(
- loc, getVoidPtrType(ret.getContext()), newResultValue);
- rewriter.setInsertionPoint(ret);
- rewriter.replaceOpWithNewOp(
- ret, mlir::ValueRange{newResultValue});
-} else if (resultStorage) {
- resultStorage.replaceAllUsesWith(newArg);
- rewriter.replaceOpWithNewOp(ret);
-} else {
- // The result storage may have been optimized out by a memory to
- // reg
[llvm-branch-commits] [clang] [flang] [lld] [llvm] [Flang] LLVM_ENABLE_RUNTIMES=flang-rt (PR #110217)
Meinersbur wrote: @h-vetinari I completed rebase and cleanup. Now would be a great time to have another look. https://github.com/llvm/llvm-project/pull/110217 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [flang] [flang][cuda] Handle gpu.return in AbstractResult pass (PR #119035)
https://github.com/clementval created
https://github.com/llvm/llvm-project/pull/119035
Process `gpu.return` in AbstractResult pass when the function is a `gpu.func`.
>From 72a892909a7ce898423cea709f674956dc2121d1 Mon Sep 17 00:00:00 2001
From: Valentin Clement
Date: Fri, 6 Dec 2024 13:42:06 -0800
Subject: [PATCH] [flang][cuda] Handle gpu.return in AbstractResult pass
---
.../Optimizer/Transforms/AbstractResult.cpp | 123 +++---
flang/test/Fir/CUDA/cuda-abstract-result.mlir | 37 ++
2 files changed, 111 insertions(+), 49 deletions(-)
create mode 100644 flang/test/Fir/CUDA/cuda-abstract-result.mlir
diff --git a/flang/lib/Optimizer/Transforms/AbstractResult.cpp
b/flang/lib/Optimizer/Transforms/AbstractResult.cpp
index 2ed66cc83eefb5..b0327cc10e9de6 100644
--- a/flang/lib/Optimizer/Transforms/AbstractResult.cpp
+++ b/flang/lib/Optimizer/Transforms/AbstractResult.cpp
@@ -234,6 +234,60 @@ class SaveResultOpConversion
}
};
+template
+static mlir::LogicalResult
+processReturnLikeOp(OpTy ret, mlir::Value newArg,
+mlir::PatternRewriter &rewriter) {
+ auto loc = ret.getLoc();
+ rewriter.setInsertionPoint(ret);
+ mlir::Value resultValue = ret.getOperand(0);
+ fir::LoadOp resultLoad;
+ mlir::Value resultStorage;
+ // Identify result local storage.
+ if (auto load = resultValue.getDefiningOp()) {
+resultLoad = load;
+resultStorage = load.getMemref();
+// The result alloca may be behind a fir.declare, if any.
+if (auto declare = resultStorage.getDefiningOp())
+ resultStorage = declare.getMemref();
+ }
+ // Replace old local storage with new storage argument, unless
+ // the derived type is C_PTR/C_FUN_PTR, in which case the return
+ // type is updated to return void* (no new argument is passed).
+ if (fir::isa_builtin_cptr_type(resultValue.getType())) {
+auto module = ret->template getParentOfType();
+FirOpBuilder builder(rewriter, module);
+mlir::Value cptr = resultValue;
+if (resultLoad) {
+ // Replace whole derived type load by component load.
+ cptr = resultLoad.getMemref();
+ rewriter.setInsertionPoint(resultLoad);
+}
+mlir::Value newResultValue =
+fir::factory::genCPtrOrCFunptrValue(builder, loc, cptr);
+newResultValue = builder.createConvert(
+loc, getVoidPtrType(ret.getContext()), newResultValue);
+rewriter.setInsertionPoint(ret);
+rewriter.replaceOpWithNewOp(ret, mlir::ValueRange{newResultValue});
+ } else if (resultStorage) {
+resultStorage.replaceAllUsesWith(newArg);
+rewriter.replaceOpWithNewOp(ret);
+ } else {
+// The result storage may have been optimized out by a memory to
+// register pass, this is possible for fir.box results, or fir.record
+// with no length parameters. Simply store the result in the result
+// storage. at the return point.
+rewriter.create(loc, resultValue, newArg);
+rewriter.replaceOpWithNewOp(ret);
+ }
+ // Delete result old local storage if unused.
+ if (resultStorage)
+if (auto alloc = resultStorage.getDefiningOp())
+ if (alloc->use_empty())
+rewriter.eraseOp(alloc);
+ return mlir::success();
+}
+
class ReturnOpConversion : public mlir::OpRewritePattern
{
public:
using OpRewritePattern::OpRewritePattern;
@@ -242,55 +296,23 @@ class ReturnOpConversion : public
mlir::OpRewritePattern {
llvm::LogicalResult
matchAndRewrite(mlir::func::ReturnOp ret,
mlir::PatternRewriter &rewriter) const override {
-auto loc = ret.getLoc();
-rewriter.setInsertionPoint(ret);
-mlir::Value resultValue = ret.getOperand(0);
-fir::LoadOp resultLoad;
-mlir::Value resultStorage;
-// Identify result local storage.
-if (auto load = resultValue.getDefiningOp()) {
- resultLoad = load;
- resultStorage = load.getMemref();
- // The result alloca may be behind a fir.declare, if any.
- if (auto declare = resultStorage.getDefiningOp())
-resultStorage = declare.getMemref();
-}
-// Replace old local storage with new storage argument, unless
-// the derived type is C_PTR/C_FUN_PTR, in which case the return
-// type is updated to return void* (no new argument is passed).
-if (fir::isa_builtin_cptr_type(resultValue.getType())) {
- auto module = ret->getParentOfType();
- FirOpBuilder builder(rewriter, module);
- mlir::Value cptr = resultValue;
- if (resultLoad) {
-// Replace whole derived type load by component load.
-cptr = resultLoad.getMemref();
-rewriter.setInsertionPoint(resultLoad);
- }
- mlir::Value newResultValue =
- fir::factory::genCPtrOrCFunptrValue(builder, loc, cptr);
- newResultValue = builder.createConvert(
- loc, getVoidPtrType(ret.getContext()), newResultValue);
- rewriter.setInsertionPoint(ret);
- rewriter.replaceOpWithNewOp(
- ret, mlir::ValueRange{newResultValue});
-} else if (resultStorage) {
-
[llvm-branch-commits] [flang] [flang][cuda] Handle gpu.return in AbstractResult pass (PR #119035)
https://github.com/wangzpgi approved this pull request. https://github.com/llvm/llvm-project/pull/119035 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang-tools-extra] [clang-tidy] fix false positive in lambda expr for return-const-ref-from-parameter (PR #118990)
https://github.com/5chmidti approved this pull request. Nice catch, LGTM https://github.com/llvm/llvm-project/pull/118990 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang-tools-extra] [clang-tidy] fix false positive in lambda expr for return-const-ref-from-parameter (PR #118990)
https://github.com/HerrCai0907 created
https://github.com/llvm/llvm-project/pull/118990
None
>From 2fb6e21f63a16b5bde5401a4e1cb68af1405a258 Mon Sep 17 00:00:00 2001
From: Congcong Cai
Date: Fri, 6 Dec 2024 23:45:56 +0800
Subject: [PATCH] [clang-tidy] fix false positive in lambda expr for
return-const-ref-from-parameter
---
.../ReturnConstRefFromParameterCheck.cpp | 14 +--
clang-tools-extra/docs/ReleaseNotes.rst | 4 ++--
.../return-const-ref-from-parameter.cpp | 23 +++
3 files changed, 31 insertions(+), 10 deletions(-)
diff --git
a/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
b/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
index a35fcd99d494af..295955a971d7e8 100644
--- a/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
+++ b/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
@@ -31,22 +31,20 @@ void
ReturnConstRefFromParameterCheck::registerMatchers(MatchFinder *Finder) {
qualType(lValueReferenceType(pointee(
qualType(isConstQualified()
.bind("type"))),
- hasDeclContext(functionDecl().bind("owner")),
+ hasDeclContext(functionDecl(
+ equalsBoundNode("func"),
+ hasReturnTypeLoc(loc(qualType(
+
hasCanonicalType(equalsBoundNode("type"))),
unless(hasLifetimeBoundAttr()))
.bind("param")))
.bind("dref"));
- const auto Func =
- functionDecl(equalsBoundNode("owner"),
- hasReturnTypeLoc(loc(
- qualType(hasCanonicalType(equalsBoundNode("type"))
- .bind("func");
Finder->addMatcher(
returnStmt(
+ hasAncestor(functionDecl().bind("func")),
hasReturnValue(anyOf(
DRef, ignoringParens(conditionalOperator(eachOf(
-hasTrueExpression(DRef), hasFalseExpression(DRef)),
- hasAncestor(Func)),
+hasTrueExpression(DRef),
hasFalseExpression(DRef))),
this);
}
diff --git a/clang-tools-extra/docs/ReleaseNotes.rst
b/clang-tools-extra/docs/ReleaseNotes.rst
index e00f86f7d01447..b2b66dca6ccf85 100644
--- a/clang-tools-extra/docs/ReleaseNotes.rst
+++ b/clang-tools-extra/docs/ReleaseNotes.rst
@@ -183,8 +183,8 @@ Changes in existing checks
- Improved :doc:`bugprone-return-const-ref-from-parameter
` check to
diagnose potential dangling references when returning a ``const &`` parameter
- by using the conditional operator ``cond ? var1 : var2`` and no longer giving
- false positives for functions which contain lambda and ignore parameters
+ by using the conditional operator ``cond ? var1 : var2`` and fixing false
+ positives for functions which contain lambda and ignore parameters
with ``[[clang::lifetimebound]]`` attribute.
- Improved :doc:`bugprone-sizeof-expression
diff --git
a/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
b/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
index 46cb9063beda97..a3297ca0f8084e 100644
---
a/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
+++
b/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
@@ -203,3 +203,26 @@ namespace use_lifetime_bound_attr {
int const &f(int const &a [[clang::lifetimebound]]) { return a; }
} // namespace use_lifetime_bound_attr
} // namespace gh117696
+
+
+namespace lambda {
+using T = const int &;
+using K = const float &;
+T inner_valid_lambda(T a) {
+ [&]() -> T { return a; };
+ return a;
+ // CHECK-MESSAGES: :[[@LINE-1]]:10: warning: returning a constant reference
parameter
+}
+T inner_invalid_lambda(T a) {
+ [&](T a) -> T { return a; };
+ // CHECK-MESSAGES: :[[@LINE-1]]:26: warning: returning a constant reference
parameter
+ return a;
+ // CHECK-MESSAGES: :[[@LINE-1]]:10: warning: returning a constant reference
parameter
+}
+T inner_invalid_lambda2(T a) {
+ [&](K a) -> K { return a; };
+ // CHECK-MESSAGES: :[[@LINE-1]]:26: warning: returning a constant reference
parameter
+ return a;
+ // CHECK-MESSAGES: :[[@LINE-1]]:10: warning: returning a constant reference
parameter
+}
+} // namespace lambda
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang-tools-extra] [clang-tidy] fix false positive in lambda expr for return-const-ref-from-parameter (PR #118990)
llvmbot wrote:
@llvm/pr-subscribers-clang-tidy
Author: Congcong Cai (HerrCai0907)
Changes
---
Full diff: https://github.com/llvm/llvm-project/pull/118990.diff
3 Files Affected:
- (modified)
clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
(+6-8)
- (modified) clang-tools-extra/docs/ReleaseNotes.rst (+2-2)
- (modified)
clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
(+23)
``diff
diff --git
a/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
b/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
index a35fcd99d494af..295955a971d7e8 100644
--- a/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
+++ b/clang-tools-extra/clang-tidy/bugprone/ReturnConstRefFromParameterCheck.cpp
@@ -31,22 +31,20 @@ void
ReturnConstRefFromParameterCheck::registerMatchers(MatchFinder *Finder) {
qualType(lValueReferenceType(pointee(
qualType(isConstQualified()
.bind("type"))),
- hasDeclContext(functionDecl().bind("owner")),
+ hasDeclContext(functionDecl(
+ equalsBoundNode("func"),
+ hasReturnTypeLoc(loc(qualType(
+
hasCanonicalType(equalsBoundNode("type"))),
unless(hasLifetimeBoundAttr()))
.bind("param")))
.bind("dref"));
- const auto Func =
- functionDecl(equalsBoundNode("owner"),
- hasReturnTypeLoc(loc(
- qualType(hasCanonicalType(equalsBoundNode("type"))
- .bind("func");
Finder->addMatcher(
returnStmt(
+ hasAncestor(functionDecl().bind("func")),
hasReturnValue(anyOf(
DRef, ignoringParens(conditionalOperator(eachOf(
-hasTrueExpression(DRef), hasFalseExpression(DRef)),
- hasAncestor(Func)),
+hasTrueExpression(DRef),
hasFalseExpression(DRef))),
this);
}
diff --git a/clang-tools-extra/docs/ReleaseNotes.rst
b/clang-tools-extra/docs/ReleaseNotes.rst
index e00f86f7d01447..b2b66dca6ccf85 100644
--- a/clang-tools-extra/docs/ReleaseNotes.rst
+++ b/clang-tools-extra/docs/ReleaseNotes.rst
@@ -183,8 +183,8 @@ Changes in existing checks
- Improved :doc:`bugprone-return-const-ref-from-parameter
` check to
diagnose potential dangling references when returning a ``const &`` parameter
- by using the conditional operator ``cond ? var1 : var2`` and no longer giving
- false positives for functions which contain lambda and ignore parameters
+ by using the conditional operator ``cond ? var1 : var2`` and fixing false
+ positives for functions which contain lambda and ignore parameters
with ``[[clang::lifetimebound]]`` attribute.
- Improved :doc:`bugprone-sizeof-expression
diff --git
a/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
b/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
index 46cb9063beda97..a3297ca0f8084e 100644
---
a/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
+++
b/clang-tools-extra/test/clang-tidy/checkers/bugprone/return-const-ref-from-parameter.cpp
@@ -203,3 +203,26 @@ namespace use_lifetime_bound_attr {
int const &f(int const &a [[clang::lifetimebound]]) { return a; }
} // namespace use_lifetime_bound_attr
} // namespace gh117696
+
+
+namespace lambda {
+using T = const int &;
+using K = const float &;
+T inner_valid_lambda(T a) {
+ [&]() -> T { return a; };
+ return a;
+ // CHECK-MESSAGES: :[[@LINE-1]]:10: warning: returning a constant reference
parameter
+}
+T inner_invalid_lambda(T a) {
+ [&](T a) -> T { return a; };
+ // CHECK-MESSAGES: :[[@LINE-1]]:26: warning: returning a constant reference
parameter
+ return a;
+ // CHECK-MESSAGES: :[[@LINE-1]]:10: warning: returning a constant reference
parameter
+}
+T inner_invalid_lambda2(T a) {
+ [&](K a) -> K { return a; };
+ // CHECK-MESSAGES: :[[@LINE-1]]:26: warning: returning a constant reference
parameter
+ return a;
+ // CHECK-MESSAGES: :[[@LINE-1]]:10: warning: returning a constant reference
parameter
+}
+} // namespace lambda
``
https://github.com/llvm/llvm-project/pull/118990
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang-tools-extra] [clang-tidy] fix false positive in lambda expr for return-const-ref-from-parameter (PR #118990)
https://github.com/HerrCai0907 edited https://github.com/llvm/llvm-project/pull/118990 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [Serialization] Code cleanups and polish 83233 (PR #83237)
luporl wrote: This also broke the build of https://lab.llvm.org/buildbot/#/builders/141/builds/4432. https://github.com/llvm/llvm-project/pull/83237 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [ubsan] Improve lowering of @llvm.allow.ubsan.check (PR #119013)
https://github.com/vitalybuka created
https://github.com/llvm/llvm-project/pull/119013
This fix the case, when single hot inlined callsite, prevent
checks for all other. This helps to reduce number of removed checks up to 50%
(deppedes on `cutoff-hot` value) .
`ScalarOptimizerLateEPCallback` was happening during
CGSCC walk, after each inlining, but this is effectively
after inlining.
Example, order in comments:
```
static void overflow() {
// 1. Inline get/set if possible
// 2. Simplify
// 3. LowerAllowCheckPass
set(get() + get());
}
void test() {
// 4. Inline
// 5. Nothing for LowerAllowCheckPass
overflow();
}
```
With this patch it will look like:
```
static void overflow() {
// 1. Inline get/set if possible
// 2. Simplify
set(get() + get());
}
void test() {
// 3. Inline
// 4. Simplify
overflow();
}
// Later, after inliner CGSCC walk complete:
// 5. LowerAllowCheckPass for `overflow`
// 6. LowerAllowCheckPass for `test`
```
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [ubsan] Improve lowering of @llvm.allow.ubsan.check (PR #119013)
llvmbot wrote:
@llvm/pr-subscribers-clang
Author: Vitaly Buka (vitalybuka)
Changes
This fix the case, when single hot inlined callsite, prevent
checks for all other. This helps to reduce number of removed checks up to 50%
(deppedes on `cutoff-hot` value) .
`ScalarOptimizerLateEPCallback` was happening during
CGSCC walk, after each inlining, but this is effectively
after inlining.
Example, order in comments:
```
static void overflow() {
// 1. Inline get/set if possible
// 2. Simplify
// 3. LowerAllowCheckPass
set(get() + get());
}
void test() {
// 4. Inline
// 5. Nothing for LowerAllowCheckPass
overflow();
}
```
With this patch it will look like:
```
static void overflow() {
// 1. Inline get/set if possible
// 2. Simplify
set(get() + get());
}
void test() {
// 3. Inline
// 4. Simplify
overflow();
}
// Later, after inliner CGSCC walk complete:
// 5. LowerAllowCheckPass for `overflow`
// 6. LowerAllowCheckPass for `test`
```
---
Full diff: https://github.com/llvm/llvm-project/pull/119013.diff
2 Files Affected:
- (modified) clang/lib/CodeGen/BackendUtil.cpp (+6-7)
- (modified) clang/test/CodeGen/allow-ubsan-check-inline.c (+2-2)
``diff
diff --git a/clang/lib/CodeGen/BackendUtil.cpp
b/clang/lib/CodeGen/BackendUtil.cpp
index bf9b04f02e9f44..1f0e760e8ebd29 100644
--- a/clang/lib/CodeGen/BackendUtil.cpp
+++ b/clang/lib/CodeGen/BackendUtil.cpp
@@ -789,13 +789,12 @@ static void addSanitizers(const Triple &TargetTriple,
}
if (LowerAllowCheckPass::IsRequested()) {
-// We can optimize after inliner, and PGO profile matching. The hook below
-// is called at the end `buildFunctionSimplificationPipeline`, which called
-// from `buildInlinerPipeline`, which called after profile matching.
-PB.registerScalarOptimizerLateEPCallback(
-[](FunctionPassManager &FPM, OptimizationLevel Level) {
- FPM.addPass(LowerAllowCheckPass());
-});
+// We want to call it after inline, which is about
OptimizerEarlyEPCallback.
+PB.registerOptimizerEarlyEPCallback([](ModulePassManager &MPM,
+ OptimizationLevel Level,
+ ThinOrFullLTOPhase Phase) {
+ MPM.addPass(createModuleToFunctionPassAdaptor(LowerAllowCheckPass()));
+});
}
}
diff --git a/clang/test/CodeGen/allow-ubsan-check-inline.c
b/clang/test/CodeGen/allow-ubsan-check-inline.c
index cabe76d8034d77..1de24ab90dac0e 100644
--- a/clang/test/CodeGen/allow-ubsan-check-inline.c
+++ b/clang/test/CodeGen/allow-ubsan-check-inline.c
@@ -7,8 +7,8 @@ void set(int x);
// We will only make decision in the `overflow` function.
// NOINL-COUNT-1: remark: Allowed check:
-// FIXME: We will make decision on every inline.
-// INLINE-COUNT-1: remark: Allowed check:
+// We will make decision on every inline.
+// INLINE-COUNT-5: remark: Allowed check:
static void overflow() {
set(get() + get());
``
https://github.com/llvm/llvm-project/pull/119013
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [Serialization] Code cleanups and polish 83233 (PR #83237)
JDevlieghere wrote: This also broke the Darwin LLDB bots: https://ci.swift.org/view/all/job/llvm.org/view/LLDB/job/lldb-cmake/ https://ci.swift.org/view/all/job/llvm.org/view/LLDB/job/as-lldb-cmake/ Given they've been red for 15h and broke various other bots, can you please revert while you investigate? https://github.com/llvm/llvm-project/pull/83237 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [Serialization] Code cleanups and polish 83233 (PR #83237)
zeroomega wrote: Change b5bd1928c6d43bc525a4e3fb65d2c750d61e broke llvm's windows bots. See : https://lab.llvm.org/buildbot/#/builders/63/builds/2944 and this patch is the only commit in the blame list. I reverted this patch in 12bdeba76eef1c7adf004a280036a7fb690ba573. Please reland the change after fixing the underlying issues. https://github.com/llvm/llvm-project/pull/83237 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [ubsan] Improve lowering of @llvm.allow.ubsan.check (PR #119013)
https://github.com/kstoimenov approved this pull request. https://github.com/llvm/llvm-project/pull/119013 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] release/19.x: [Clang][perf-training] Fix clean command in perf-helper.py (#118978) (PR #119051)
https://github.com/llvmbot milestoned https://github.com/llvm/llvm-project/pull/119051 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] release/19.x: [Clang][perf-training] Fix clean command in perf-helper.py (#118978) (PR #119051)
https://github.com/llvmbot created https://github.com/llvm/llvm-project/pull/119051 Backport 18af3fc1bf8855e1e166e64a9210ed07d610aa54 Requested by: @tstellar >From 0b1ed9168731f0bee35542988ec9867ebd3cfc47 Mon Sep 17 00:00:00 2001 From: Tom Stellard Date: Fri, 6 Dec 2024 18:34:59 -0800 Subject: [PATCH] [Clang][perf-training] Fix clean command in perf-helper.py (#118978) The first path argument was always being ignored, and since most calls to this command only passed one path, it wasn't actually doing anything in most cases. This bug was introduced by dd0356d741aefa25ece973d6cc4b55dcb73b84b4. (cherry picked from commit 18af3fc1bf8855e1e166e64a9210ed07d610aa54) --- clang/utils/perf-training/perf-helper.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/clang/utils/perf-training/perf-helper.py b/clang/utils/perf-training/perf-helper.py index 3ed42a187fd803..d76c6ede3fe5a2 100644 --- a/clang/utils/perf-training/perf-helper.py +++ b/clang/utils/perf-training/perf-helper.py @@ -36,7 +36,7 @@ def clean(args): + "\tRemoves all files with extension from ." ) return 1 -for path in args[1:-1]: +for path in args[0:-1]: for filename in findFilesWithExtension(path, args[-1]): os.remove(filename) return 0 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] release/19.x: [Clang][perf-training] Fix clean command in perf-helper.py (#118978) (PR #119051)
llvmbot wrote: @llvm/pr-subscribers-clang Author: None (llvmbot) Changes Backport 18af3fc1bf8855e1e166e64a9210ed07d610aa54 Requested by: @tstellar --- Full diff: https://github.com/llvm/llvm-project/pull/119051.diff 1 Files Affected: - (modified) clang/utils/perf-training/perf-helper.py (+1-1) ``diff diff --git a/clang/utils/perf-training/perf-helper.py b/clang/utils/perf-training/perf-helper.py index 3ed42a187fd803..d76c6ede3fe5a2 100644 --- a/clang/utils/perf-training/perf-helper.py +++ b/clang/utils/perf-training/perf-helper.py @@ -36,7 +36,7 @@ def clean(args): + "\tRemoves all files with extension from ." ) return 1 -for path in args[1:-1]: +for path in args[0:-1]: for filename in findFilesWithExtension(path, args[-1]): os.remove(filename) return 0 `` https://github.com/llvm/llvm-project/pull/119051 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] release/19.x: [Clang][perf-training] Fix clean command in perf-helper.py (#118978) (PR #119051)
llvmbot wrote: @petrhosek What do you think about merging this PR to the release branch? https://github.com/llvm/llvm-project/pull/119051 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
