nikic wrote:
Looks like using InsertPosition in IRBuilder has some overhead:
https://llvm-compile-time-tracker.com/compare.php?from=317277e4f961edf13132914a58a26408db4ab0aa&to=6481dc57612671ebe77fe9c34214fba94e1b3b27&stat=instructions:u
So it might make sense to keep some explicit overloads for
https://github.com/nikic approved this pull request.
https://github.com/llvm/llvm-project/pull/96532
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
> > Looks like using InsertPosition in IRBuilder has some overhead:
>
> Hm, we're adding an extra pointer chase when we give up passing a basicblock
> - I think then it makes sense to keep the option to pass BB+It, and do so
> when the caller already knows the BB (while removing t
nikic wrote:
Yes, this causes a massive compile-time regression:
http://llvm-compile-time-tracker.com/compare.php?from=253a294b54a6096a0b66f840931dd0e345d70c4f&to=20869e8ed55c58696ce4e9dc3d18cd6af01ed49c&stat=instructions:u
https://github.com/llvm/llvm-project/pull/96620
___
https://github.com/nikic created https://github.com/llvm/llvm-project/pull/97023
Replace it with a forward declaration instead. Analysis.h is pulled in by all
passes, but not all passes need to access the module.
>From f21bee2d46446e9238992fde4e190ce8010ede7a Mon Sep 17 00:00:00 2001
From: Niki
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/97023
>From f21bee2d46446e9238992fde4e190ce8010ede7a Mon Sep 17 00:00:00 2001
From: Nikita Popov
Date: Thu, 27 Jun 2024 17:05:54 +0200
Subject: [PATCH 1/2] [IR] Don't include Module.h in Analysis.h (NFC)
Replace it with
https://github.com/nikic closed https://github.com/llvm/llvm-project/pull/97023
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic requested changes to this pull request.
This is probably caused by some downstream patches in google3?
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
does not contain any mentions of `M->getFunction` or `M->getCont
Author: Nikita Popov
Date: 2024-06-29T11:50:28+02:00
New Revision: 5cc1287bdbb2de9ad91e4ba8dc9eeb9508c734a5
URL:
https://github.com/llvm/llvm-project/commit/5cc1287bdbb2de9ad91e4ba8dc9eeb9508c734a5
DIFF:
https://github.com/llvm/llvm-project/commit/5cc1287bdbb2de9ad91e4ba8dc9eeb9508c734a5.diff
nikic wrote:
Reduced: https://llvm.godbolt.org/z/TPdd9qore
I believe the problem is this line:
https://github.com/llvm/llvm-project/blob/c05e29bff036060f0811b887a92715104abdceb5/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp#L2513
This does an in-place modification of the GEP, but on
nikic wrote:
Should be fixed by
https://github.com/llvm/llvm-project/commit/940f89255e4a3982d94dad57837e8e658092af78.
https://github.com/llvm/llvm-project/pull/107257
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-b
Author: Nikita Popov
Date: 2024-09-13T11:15:22+02:00
New Revision: 1c298c927498983dee29037ed1d3e71b72ca0082
URL:
https://github.com/llvm/llvm-project/commit/1c298c927498983dee29037ed1d3e71b72ca0082
DIFF:
https://github.com/llvm/llvm-project/commit/1c298c927498983dee29037ed1d3e71b72ca0082.diff
nikic wrote:
ping
https://github.com/llvm/llvm-project/pull/107601
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
Zen 5 support in GCC was upstreamed more than half a year ago -- why is the
LLVM support being upstreamed only now, after missing the 19.x window? What
steps are being taken to ensure this does not happen again?
The change to X86TargetParser.h looks ABI breaking to me.
https://gi
https://github.com/nikic approved this pull request.
https://github.com/llvm/llvm-project/pull/108675
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -18197,7 +18197,7 @@ Value *CodeGenFunction::EmitPPCBuiltinExpr(unsigned
BuiltinID,
CallOps.push_back(Ops[i]);
llvm::Function *F = CGM.getIntrinsic(ID);
Value *Call = Builder.CreateCall(F, CallOps);
-return Builder.CreateAlignedStore(Call, Ops[0], MaybeAli
nikic wrote:
It looks like you're trying to undo
https://github.com/llvm/llvm-project/pull/106612? cc @mgorny
https://github.com/llvm/llvm-project/pull/108549
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mail
https://github.com/nikic approved this pull request.
https://github.com/llvm/llvm-project/pull/108738
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
https://github.com/llvm/llvm-project/pull/108736
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
https://github.com/llvm/llvm-project/pull/108735
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -3050,6 +3050,19 @@ as follows:
address space 0, this property only affects the default value to be used
when creating globals without additional contextual information (e.g. in
LLVM passes).
+``T``
+Specifies the address space for a target's 'flat' address s
https://github.com/nikic closed https://github.com/llvm/llvm-project/pull/107601
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
https://github.com/llvm/llvm-project/pull/108794
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
FYI this is a measurable compile-time regression when compiling tiny files
(e.g. see the red 7zip entries at the end of
https://llvm-compile-time-tracker.com/compare.php?from=5c348f692a8dff98a3780d0b859fb0949eccbaca&to=f5ba3e1fa6b5f862789786fbb4b342dfc2c27c33&stat=instructions%3Au&
nikic wrote:
To reuse the name with different semantics, I'd put removal and reuse in
different releases, so it takes about half a year. (Deprecation itself can be
very short-lived.)
https://github.com/llvm/llvm-project/pull/111752
___
cfe-commits ma
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/112071
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/112074
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -1391,22 +1411,60 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
https://github.com/nikic edited https://github.com/llvm/llvm-project/pull/91101
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/91101
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -171,7 +171,7 @@ bool HexagonGenExtract::convert(Instruction *In) {
// this value.
if (!LogicalSR && (SR > SL))
return false;
-APInt A = APInt(BW, ~0ULL).lshr(SR).shl(SL);
+APInt A = APInt(BW, ~0ULL, true).lshr(SR).shl(SL);
nikic wrote:
nikic wrote:
> `APInt::getAllOnes` will assert if `BitWidth == 0`.
>
> [snip]
For the record, this is fixed by
https://github.com/llvm/llvm-project/pull/112227.
https://github.com/llvm/llvm-project/pull/80309
___
cfe-commits mailing list
cfe-commits
@@ -273,8 +273,6 @@ PreservedAnalyses FunctionToLoopPassAdaptor::run(Function
&F,
llvm::any_cast(&IR));
const Loop **LPtr = llvm::any_cast(&IR);
const Loop *L = LPtr ? *LPtr : nullptr;
-if (!L)
- L = &llvm::any_cast(IR)->getOutermostLoop();
https://github.com/nikic approved this pull request.
LLVM changes LGTM
https://github.com/llvm/llvm-project/pull/108051
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic edited https://github.com/llvm/llvm-project/pull/108051
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/80309
>From 96795194fc79359363bac0423516da2a06733047 Mon Sep 17 00:00:00 2001
From: Nikita Popov
Date: Thu, 19 Sep 2024 17:27:13 +0200
Subject: [PATCH 1/2] apint only
---
clang/lib/AST/ByteCode/IntegralAP.h |
@@ -1159,7 +1159,9 @@ class ARMOperand : public MCParsedAsmOperand {
if (!isImm()) return false;
const MCConstantExpr *CE = dyn_cast(getImm());
if (!CE) return false;
nikic wrote:
Not quite, but this put me on the right path. I added a check for `i
@@ -1391,22 +1412,65 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
@@ -1358,18 +1360,37 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
auto &Context = CalledFunction->getContext();
// Collect valid attributes for all params.
- SmallVector ValidParamAttrs;
+ SmallVector ValidObjParamAttrs, ValidExactParamAttrs;
bool
@@ -787,6 +787,11 @@ class AttributeList {
[[nodiscard]] AttributeList addRangeRetAttr(LLVMContext &C,
const ConstantRange &CR) const;
+ /// Add the range attribute to the attribute set at the given arg index.
+ /// Returns a n
@@ -373,6 +373,13 @@ FPClassTest CallBase::getParamNoFPClass(unsigned i) const {
return Mask;
}
+std::optional CallBase::getParamRange(unsigned ArgNo) const {
nikic wrote:
I don't think this API is used?
https://github.com/llvm/llvm-project/pull/91101
@@ -1391,22 +1412,65 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
@@ -1391,22 +1412,65 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
@@ -1391,22 +1412,65 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
@@ -1391,22 +1412,65 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
https://github.com/nikic commented:
> I realize this is not the approach you had in mind, but it would at least be
> nice to hear the reasoning as to why something more elaborate would be
> preferred?
Basically, because this is an ABI-dependent property, so we should let the
existing ABI hand
nikic wrote:
The invoke merging problem still exists, though the impact dropped:
```
simplifycfg.NumInvokesMerged 136372 -> 136185 -0.14%
```
Need to adjust this:
https://github.com/llvm/llvm-project/blob/774893dcd929c370bad714a70a7d670bb2d6f649/llvm/lib/Transforms/Utils/SimplifyCFG.cpp#L2782
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111628
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/112017
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
> > `test_demangle` should definitely **not** be updated when renaming
> > functions.
>
> Could you please explain a bit more since this is new to me? Thanks in
> advance!
The use of LLVM symbols in this file is incidental -- it's just a data set of
mangled C++ symbol names. The
nikic wrote:
`test_demangle` should definitely **not** be updated when renaming functions.
https://github.com/llvm/llvm-project/pull/111752
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-comm
nikic wrote:
TBH I am wondering if we should revert this rename. I don't think it's a good
idea to reuse getDeclaration with substantially different semantics, and if
we're not reusing it, then there's not much point to rename...
Instead of having getOrInsertDeclaration + getDeclaration with a
@@ -1713,25 +1713,15 @@ llvm::Function
*CGOpenMPRuntime::emitThreadPrivateVarDefinition(
if (!Ctor && !Dtor)
return nullptr;
-llvm::Type *CopyCtorTyArgs[] = {CGM.VoidPtrTy, CGM.VoidPtrTy};
-auto *CopyCtorTy = llvm::FunctionType::get(CGM.VoidPtrTy, CopyCtorTy
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111327
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111272
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111273
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
Based on llvm-opt-benchmark results, it looks like there are some significant
compile-time regressions, could you please take a look?
I also see:
```
early-cse.NumCSECall 18050 -> 17883 -0.93%
simplifycfg.NumInvokesMerged 121710 -> 120984 -0.60%
```
EarlyCSE uses isIdenticalTo and
nikic wrote:
It may be a good idea to skip the inference for constant arguments. I see a
decent amount of things like `range(i64 -2147483576, 34359738361) 272` in the
diffs.
Though I still don't think that we should be inferring range at all.
https://github.com/llvm/llvm-project/pull/91101
__
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111228
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/80309
>From 96795194fc79359363bac0423516da2a06733047 Mon Sep 17 00:00:00 2001
From: Nikita Popov
Date: Thu, 19 Sep 2024 17:27:13 +0200
Subject: [PATCH 1/5] apint only
---
clang/lib/AST/ByteCode/IntegralAP.h |
@@ -3600,8 +3600,11 @@ ExprResult Sema::ActOnCharacterConstant(const Token
&Tok, Scope *UDLScope) {
ExprResult Sema::ActOnIntegerConstant(SourceLocation Loc, uint64_t Val) {
unsigned IntSize = Context.getTargetInfo().getIntWidth();
- return IntegerLiteral::Create(Context,
@@ -437,7 +437,9 @@ ExprResult Parser::createEmbedExpr() {
SourceLocation StartLoc = ConsumeAnnotationToken();
if (Data->BinaryData.size() == 1) {
Res = IntegerLiteral::Create(Context,
- llvm::APInt(CHAR_BIT,
Data->BinaryData.back()),
+
@@ -268,7 +268,7 @@ struct UniqueVirtualMethod {
/// subobject in which that virtual function occurs).
class OverridingMethods {
using ValuesT = SmallVector;
- using MapType = llvm::MapVector;
+ using MapType = llvm::SmallMapVector;
nikic wrote:
It looks l
@@ -51,7 +51,7 @@ class ConstraintSystem {
/// A map of variables (IR values) to their corresponding index in the
/// constraint system.
- DenseMap Value2Index;
+ SmallDenseMap Value2Index;
nikic wrote:
I'd add a using/typedef for this one, especially a
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/80309
>From 96795194fc79359363bac0423516da2a06733047 Mon Sep 17 00:00:00 2001
From: Nikita Popov
Date: Thu, 19 Sep 2024 17:27:13 +0200
Subject: [PATCH 1/6] apint only
---
clang/lib/AST/ByteCode/IntegralAP.h |
@@ -1391,22 +1411,60 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
@@ -1391,22 +1411,60 @@ static void AddParamAndFnBasicAttributes(const CallBase
&CB,
AttributeList AL = NewInnerCB->getAttributes();
for (unsigned I = 0, E = InnerCB->arg_size(); I < E; ++I) {
-// Check if the underlying value for the parameter is an argum
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/113071
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
> @nikic I checked several other arch-32 situations with ForceEnableInt128, and
> they all use `alloca i128, align 16`. But they did not modify the target
> datalayout content with `i128:128`, should mips be consistent?
I think either way is fine here. Something to consider though
https://github.com/nikic closed https://github.com/llvm/llvm-project/pull/80309
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
https://github.com/llvm/llvm-project/pull/112749
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111785
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111784
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111783
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111467
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111469
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111468
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111225
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111227
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -92,11 +92,10 @@ parseTargetIDWithFormatCheckingOnly(llvm::StringRef
TargetID,
if (Sign != '+' && Sign != '-')
return std::nullopt;
bool IsOn = Sign == '+';
-auto Loc = FeatureMap->find(Feature);
+auto [Loc, Inserted] = FeatureMap->try_emplace(Feature,
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111088
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111089
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111087
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic commented:
This looks technically fine, but I'm having trouble finding a normative
reference for this. Is there an ABI specification for N32/N64 somewhere?
Also, what alignment does clang use for o32 with ForceEnableInt128?
https://github.com/llvm/llvm-project/blob/cb2
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/111939
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/112156
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/112155
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/80309
>From ab1d3f23ee1a8402403a61038effb9bb15c91a13 Mon Sep 17 00:00:00 2001
From: Nikita Popov
Date: Thu, 19 Sep 2024 17:27:13 +0200
Subject: [PATCH] apint only
---
clang/lib/AST/ByteCode/IntegralAP.h | 6 +
nikic wrote:
Ping. The PR no longer contains MLIR changes now.
https://github.com/llvm/llvm-project/pull/80309
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
Looking at our current cmake config, we're using:
```
-DLIBCXX_STATICALLY_LINK_ABI_IN_STATIC_LIBRARY=ON \
-DLIBCXX_ENABLE_ABI_LINKER_SCRIPT=ON \
```
Do I understand correctly that the new replacement for that would be?
```
-DLIBCXX_ABILIB_FOR_SHARED="shared-libcxxabi" \
-DLIBCXX_ABIL
nikic wrote:
> Yes, that is correct. This patch should be changing your CMake config as
> well, unless that config is not visible in the monorepo (or if I missed it).
It's out of tree :)
> > And those LIBCXX_ABILIB options are also the defaults, if I'm reading the
> > docs right.
>
> Not qui
nikic wrote:
> Is it even possible for `isSafeToSpeculativelyExecute` to return `true` now
> with the default (nullptr) CtxI for loads? Can
> `isDereferenceableAndAlignedPointer` just short-circuit to `false` with null
> CtxI?
Yes, it's possible for anything where the derefability is not cont
nikic wrote:
I do think this change still makes sense, especially from a consistency point
of view. If SROA sets one of the value-based FMF flags (nsz) then it stands to
reason that it should also set the other two (nnan and ninf). Unless there is
some reason why nsz would be more problematic
https://github.com/nikic approved this pull request.
LG
https://github.com/llvm/llvm-project/pull/114993
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
nikic wrote:
> I think these are a net positive change; @nikic are you comfortable with the
> performance numbers? I feel like the changes come awfully close to falling
> out in the noise aside from the positive improvements with building Clang,
> but you look at compile time performance numbe
nikic wrote:
> @nikic I mean not in theory but currently. I don't see those (except allocas)
> handled anywhere unless I'm missing something?
The `I &&` part handles those. If it's a global or argument (thus not an
instruction) we'll fall through to the return true.
https://github.com/llvm/ll
nikic wrote:
> I've rebased my PR so this branch could also contain phase-ordering test for
> vectorizing predicated selects introduced by commit
> [577c7dd](https://github.com/llvm/llvm-project/commit/577c7dd7cc4c5a9f62f9654cfa30ee9d55709426)
Do I understand correctly that this PR doesn't cha
https://github.com/nikic commented:
We should probably also handle `no-infs-fp-math` to save the next person the
trouble.
https://github.com/llvm/llvm-project/pull/114271
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/c
@@ -1140,6 +1148,15 @@ void PromoteMem2Reg::RenamePass(BasicBlock *BB,
BasicBlock *Pred,
if (isa(APN) && NoSignedZeros)
APN->setHasNoSignedZeros(true);
+// This allows select instruction folding relevant to floating point
+// reductions whose
https://github.com/nikic approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/114271
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
801 - 900 of 1388 matches
Mail list logo
