https://github.com/fhahn updated https://github.com/llvm/llvm-project/pull/76260
>From ab8d005600b99fb62d991bc63c58136576429385 Mon Sep 17 00:00:00 2001
From: Florian Hahn
Date: Thu, 18 Apr 2024 23:01:03 +0100
Subject: [PATCH 1/3] [TySan] A Type Sanitizer (Clang)
---
clang/include/clang/Basic/
fhahn wrote:
> The clang changes are ok, but this needs some level of documentation/release
> notes, which I don't see in the clang release. As this is a part of a larger
> feature, do we intend to push that later?
>
> Also, the clang-format suggestion makes sense.
Sorry for the long delay! A
fhahn wrote:
Thanks for putting up the patch!
I just rebased the patches, hopefully we can get them in soon so it is easier
to submit bug-fixes iteratively.
I noticed that with this patch, I am seeing segfaults when running
`llvm-min-tblgen` when built with `-fsanitize=type`, but I wasn't ab
https://github.com/fhahn updated https://github.com/llvm/llvm-project/pull/76260
>From ab8d005600b99fb62d991bc63c58136576429385 Mon Sep 17 00:00:00 2001
From: Florian Hahn
Date: Thu, 18 Apr 2024 23:01:03 +0100
Subject: [PATCH 1/2] [TySan] A Type Sanitizer (Clang)
---
clang/include/clang/Basic/
https://github.com/fhahn closed https://github.com/llvm/llvm-project/pull/118020
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
https://github.com/fhahn edited https://github.com/llvm/llvm-project/pull/134035
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
@@ -55,8 +57,10 @@ static void rewriteFuncWithReturnType(Function &OldF, Value
*NewRetValue) {
BasicBlock::iterator NewValIt =
NewRetI ? NewRetI->getIterator() : EntryBB.end();
+ Type *OldRetTy = OldFuncTy->getReturnType();
+
// Hack up any return values in other
https://github.com/fhahn approved this pull request.
LGTM ,thanks!
https://github.com/llvm/llvm-project/pull/134685
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-comm
https://github.com/fhahn approved this pull request.
LGTM, thanks
https://github.com/llvm/llvm-project/pull/134035
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commi
https://github.com/fhahn approved this pull request.
LGTM, thanks
https://github.com/llvm/llvm-project/pull/134691
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commi
https://github.com/fhahn commented:
Just checking why this was closed?
https://github.com/llvm/llvm-project/pull/135799
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-
https://github.com/fhahn approved this pull request.
LGTM, thanks
https://github.com/llvm/llvm-project/pull/137179
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commi
https://github.com/fhahn commented:
Should be good to include, assuming the test failures are just missed test
updates
https://github.com/llvm/llvm-project/pull/136863
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lis
@@ -9545,7 +9545,9 @@ AArch64TargetLowering::LowerCall(CallLoweringInfo &CLI,
// Do what the frontend tells us: if the rvmarker module flag is present,
// emit the marker. Always emit the call regardless.
// Tell the pseudo expansion using an additional boolean op.
@@ -9545,7 +9545,9 @@ AArch64TargetLowering::LowerCall(CallLoweringInfo &CLI,
// Do what the frontend tells us: if the rvmarker module flag is present,
// emit the marker. Always emit the call regardless.
// Tell the pseudo expansion using an additional boolean op.
@@ -160,10 +160,7 @@ static bool lowerObjCCall(Function &F, const char *NewFn,
auto *CB = cast(U.getUser());
if (CB->getCalledFunction() != &F) {
- objcarc::ARCInstKind Kind = objcarc::getAttachedARCFunctionKind(CB);
- (void)Kind;
- assert((Kind == objca
fhahn wrote:
I'm not sure if it is feasible to strip the fix down, as it depends quite a few
refactoring patches.
For 20.x, it might be best just not enable epilogue vectorization for
FindLastIV: https://github.com/llvm/llvm-project/pull/135666
https://github.com/llvm/llvm-project/pull/135231
https://github.com/fhahn created
https://github.com/llvm/llvm-project/pull/135666
Back-porting properly freezing of the start value during epilogue
vectorization
(https://github.com/llvm/llvm-project/commit/2bdc1a1337692a5743658ba6b680e5d914e684a4)
is non-trivial. For the 20.x release, just
di
https://github.com/fhahn updated
https://github.com/llvm/llvm-project/pull/135666
>From e21b3211d2129e3e0110bcaa75ab0b451397ef42 Mon Sep 17 00:00:00 2001
From: Florian Hahn
Date: Mon, 14 Apr 2025 17:16:08 +0100
Subject: [PATCH 1/2] [LV] Add tests with FindLastIV and epilogue
vectorization.
Te
@@ -2432,12 +2437,40 @@ static void
tryToCreateAbstractReductionRecipe(VPReductionRecipe *Red,
Red->replaceAllUsesWith(AbstractR);
}
+/// This function tries to create an abstract recipe from a partial reduction
to
+/// hide its mul and extends from cost estimation.
+stati
https://github.com/fhahn approved this pull request.
LGTM, thanks!
https://github.com/llvm/llvm-project/pull/138681
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-comm
fhahn wrote:
Without the pointer-tbaa changes (which is new on by default in Clang 20), we
would always generate `any pointer`. Without this fix, we will generate
different tags for different `void` pointer depths.
With this fix, we will generate `any pointer` again for `void` pointers. I
th
https://github.com/fhahn edited https://github.com/llvm/llvm-project/pull/129317
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
https://github.com/fhahn created
https://github.com/llvm/llvm-project/pull/129317
Backport 275baedfde9dcd344bc4f11f552b046a69a4bf3f
In some cases, it is possible for the same underlying object to be accessed via
pointers to different address spaces. This could lead to pointers from
different
https://github.com/fhahn milestoned
https://github.com/llvm/llvm-project/pull/129317
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
https://github.com/fhahn milestoned
https://github.com/llvm/llvm-project/pull/128879
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
https://github.com/fhahn approved this pull request.
This fixes a potential crash with assertions enabled, would be good to pick to
avoid assertions.
https://github.com/llvm/llvm-project/pull/128694
___
llvm-branch-commits mailing list
llvm-branch-com
https://github.com/fhahn created
https://github.com/llvm/llvm-project/pull/128879
Backport
https://github.com/llvm/llvm-project/commit/8009c1fd81ad0b6ac65724d2b134a92db48f8fbf
>From 508d5e45a57affceeb87e69a99ebb65efde88478 Mon Sep 17 00:00:00 2001
From: Elvis Wang
Date: Tue, 25 Feb 2025 11:0
fhahn wrote:
I created https://github.com/llvm/llvm-project/pull/128879 with a test update
https://github.com/llvm/llvm-project/pull/128694
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/l
https://github.com/fhahn updated
https://github.com/llvm/llvm-project/pull/128879
>From 508d5e45a57affceeb87e69a99ebb65efde88478 Mon Sep 17 00:00:00 2001
From: Elvis Wang
Date: Tue, 25 Feb 2025 11:09:09 +0800
Subject: [PATCH 1/2] [LV][VPlan] Prevent calculate cost for skiped
instructions in pr
https://github.com/fhahn approved this pull request.
LGTM, thanks
https://github.com/llvm/llvm-project/pull/133412
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commi
https://github.com/fhahn approved this pull request.
LGTM, thanks
https://github.com/llvm/llvm-project/pull/133408
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commi
https://github.com/fhahn commented:
Could you add an Alive2 proof showing this transform is valid given the
precondition for completeness?
https://github.com/llvm/llvm-project/pull/135785
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.
https://github.com/fhahn approved this pull request.
LGTM, thanks
https://github.com/llvm/llvm-project/pull/145639
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commi
https://github.com/fhahn commented:
does the latest version of this depend on
https://github.com/llvm/llvm-project/pull/147026?
https://github.com/llvm/llvm-project/pull/147513
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
ht
https://github.com/fhahn approved this pull request.
LGTM, thanks
https://github.com/llvm/llvm-project/pull/145638
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commi
@@ -2744,6 +2702,12 @@ class VPSingleDefBundleRecipe : public VPSingleDefRecipe
{
/// vector operands, performing a reduction.add on the result, and adding
/// the scalar result to a chain.
MulAccumulateReduction,
+/// Represent an inloop multiply-accumulate re
@@ -2532,7 +2533,10 @@ class VPPartialReductionRecipe : public
VPReductionRecipe {
Opcode(Opcode), VFScaleFactor(ScaleFactor) {
[[maybe_unused]] auto *AccumulatorRecipe =
getChainOp()->getDefiningRecipe();
-assert((isa(AccumulatorRecipe) ||
+// When
@@ -2857,15 +2857,26 @@ tryToMatchAndCreateExtendedReduction(VPReductionRecipe
*Red, VPCostContext &Ctx,
VPValue *VecOp = Red->getVecOp();
// Clamp the range if using extended-reduction is profitable.
- auto IsExtendedRedValidAndClampRange = [&](unsigned Opcode, bool isZ
@@ -2955,12 +2966,14 @@
tryToMatchAndCreateMulAccumulateReduction(VPReductionRecipe *Red,
// Match reduce.add(mul(ext, ext)).
if (RecipeA && RecipeB &&
-(RecipeA->getOpcode() == RecipeB->getOpcode() || A == B) &&
+(RecipeA->getOpcode() == RecipeB->getO
@@ -2678,6 +2684,23 @@ InstructionCost
VPExpressionRecipe::computeCost(ElementCount VF,
case ExpressionTypes::ExtNegatedMulAccReduction:
case ExpressionTypes::ExtMulAccReduction: {
bool Negated = ExpressionType ==
ExpressionTypes::ExtNegatedMulAccReduction;
+if (i
@@ -2470,7 +2470,8 @@ class VPReductionRecipe : public VPRecipeWithIRFlags {
static inline bool classof(const VPRecipeBase *R) {
return R->getVPDefID() == VPRecipeBase::VPReductionSC ||
- R->getVPDefID() == VPRecipeBase::VPReductionEVLSC;
+ R->getVPDe
https://github.com/fhahn created
https://github.com/llvm/llvm-project/pull/149815
Back-port https://github.com/llvm/llvm-project/pull/146526
(https://github.com/llvm/llvm-project/commit/02d3738be92eac38cebfb7b670673abb9538ca76)
for the 21.x release, just for Apple
CPUs. As discussed during the
https://github.com/fhahn milestoned
https://github.com/llvm/llvm-project/pull/149815
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
https://github.com/fhahn updated
https://github.com/llvm/llvm-project/pull/149736
>From 104183777e3a6bd14f209c74da53e6f592b72d9f Mon Sep 17 00:00:00 2001
From: Florian Hahn
Date: Fri, 18 Jul 2025 22:58:19 +0200
Subject: [PATCH 1/2] [LV] Vectorize maxnum/minnum w/o fast-math flags.
(#148239)
U
https://github.com/fhahn commented:
> Nit: name could be shortened from
> `‏handleMaxMinNumReductionsWithoutFastMath()` to
> `‏handleMaxMinNumReductions()`
Ah yes thanks! done on main in 3813567e08e202280546dd9fd5ae891d067cbda9 and
also adjusted in this PR.
https://github.com/llvm/llvm-proj
https://github.com/fhahn updated
https://github.com/llvm/llvm-project/pull/149736
>From 104183777e3a6bd14f209c74da53e6f592b72d9f Mon Sep 17 00:00:00 2001
From: Florian Hahn
Date: Fri, 18 Jul 2025 22:58:19 +0200
Subject: [PATCH 1/4] [LV] Vectorize maxnum/minnum w/o fast-math flags.
(#148239)
U
https://github.com/fhahn updated
https://github.com/llvm/llvm-project/pull/149736
>From 104183777e3a6bd14f209c74da53e6f592b72d9f Mon Sep 17 00:00:00 2001
From: Florian Hahn
Date: Fri, 18 Jul 2025 22:58:19 +0200
Subject: [PATCH 1/3] [LV] Vectorize maxnum/minnum w/o fast-math flags.
(#148239)
U
fhahn wrote:
Hm, happy to include the full change. We evaluated carefully evaluated this for
Apple CPUs, where this is a significant improvement, with little risk given the
evaluation on Apple platforms.
> Is this mitigating a regression? If not, I don't think we should backport
> optimizatio
fhahn wrote:
> Can this be squashed for the release branch? We ideally want to have just one
> commit per fix.
@tru is there a way to update this PR here or do I need to create a new PR?
https://github.com/llvm/llvm-project/pull/149736
___
llvm-branc
fhahn wrote:
> I'm not a huge fan of hacks like this on the release branch. It should be the
> same patch as on trunk (or not be backported).
@davemgreen I agree it's not really pretty, wdyt about including the full
change?
https://github.com/llvm/llvm-project/pull/149815
https://github.com/fhahn created
https://github.com/llvm/llvm-project/pull/150193
Backport
https://github.com/llvm/llvm-project/commit/004c67ea257039e4e98abc26dd4ac6e8f3d7a171
Requested by: @fhahn
>From 95c38757d148670018134d1dc6ad6ec5e52f8433 Mon Sep 17 00:00:00 2001
From: Florian Hahn
Date
fhahn wrote:
@tru put up a new PR with the commits squashed:
https://github.com/llvm/llvm-project/pull/150193
https://github.com/llvm/llvm-project/pull/149736
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.o
https://github.com/fhahn milestoned
https://github.com/llvm/llvm-project/pull/150193
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
fhahn wrote:
This is a version of https://github.com/llvm/llvm-project/pull/149736 with the
commits squashed
https://github.com/llvm/llvm-project/pull/150193
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.or
301 - 355 of 355 matches
Mail list logo