pogo59 wrote:
I'm seeing two bot failures:
https://lab.llvm.org/buildbot/#/builders/60/builds/17462
https://lab.llvm.org/buildbot/#/builders/119/builds/18845
both for sysroot.c.
These are Arm/AArch64 Windows bots, based on the names, which is not an
environment I have available. I cannot reprodu
https://github.com/pogo59 created
https://github.com/llvm/llvm-project/pull/94253
After #94055 this test failed on ARM/AArch64-hosted Windows, but it's not clear
why.
>From b31aa18d5abb553e4c44eff338a8c9e8393986b7 Mon Sep 17 00:00:00 2001
From: Paul Robinson
Date: Mon, 3 Jun 2024 09:38:45 -07
https://github.com/pogo59 closed https://github.com/llvm/llvm-project/pull/94253
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 created
https://github.com/llvm/llvm-project/pull/94258
See #94000 for a report of a downstream failure, this fixes it.
>From 21fea52ae628f1105ea562b77f7987f87908d948 Mon Sep 17 00:00:00 2001
From: Paul Robinson
Date: Mon, 3 Jun 2024 10:08:42 -0700
Subject: [PATCH] [D
https://github.com/pogo59 closed https://github.com/llvm/llvm-project/pull/94258
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
pogo59 wrote:
Let's hope so! Done.
https://github.com/llvm/llvm-project/pull/94000
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
pogo59 wrote:
Correctly written Driver tests do not depend on any particular target. All
Driver code relevant to all targets is present in all cases.
Exactly one test (out of 133) was failing on exactly two bots. I'm not saying
what I did was the best thing; it was expedient, to get the bots g
pogo59 wrote:
FTR, when I say I could not reproduce on X86 Windows, exactly what I did is
build on an X86 Windows host but include only the AArch64 target. The test
passed.
https://github.com/llvm/llvm-project/pull/94055
___
cfe-commits mailing list
https://github.com/pogo59 created
https://github.com/llvm/llvm-project/pull/94276
A DEFAULT_SYSROOT interfered with the test, apparently. See #95055.
>From c89183017ef9719c92077c1be6fd282e22a98e4e Mon Sep 17 00:00:00 2001
From: Paul Robinson
Date: Mon, 3 Jun 2024 12:43:59 -0700
Subject: [PATCH
pogo59 wrote:
Pls try in your environments to see if this solves the problem.
https://github.com/llvm/llvm-project/pull/94276
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 edited https://github.com/llvm/llvm-project/pull/94276
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 updated
https://github.com/llvm/llvm-project/pull/94276
>From b4b78efe9c8119e6b4b1f539847e5678808bf715 Mon Sep 17 00:00:00 2001
From: Paul Robinson
Date: Mon, 3 Jun 2024 12:43:59 -0700
Subject: [PATCH] [Driver] Fix the sysroot.c test properly
A DEFAULT_SYSROOT interfe
https://github.com/pogo59 closed https://github.com/llvm/llvm-project/pull/94276
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
pogo59 wrote:
> @pogo59 Hmm - I thought you held a strong preference that debugger tuning
> never be the only way to access a certain feature, and that we should always
> have direct control of these features via flags (but with default (or
> explicit) debugger tuning setting some of these fla
@@ -4584,6 +4584,32 @@ renderDebugOptions(const ToolChain &TC, const Driver &D,
const llvm::Triple &T,
}
}
+ // Emit DW_TAG_template_alias for template aliases? True by default for SCE.
+ const auto *DebugTemplateAlias = Args.getLastArg(
+ options::OPT_gtemplate_
https://github.com/pogo59 deleted
https://github.com/llvm/llvm-project/pull/87623
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -4584,6 +4584,32 @@ renderDebugOptions(const ToolChain &TC, const Driver &D,
const llvm::Triple &T,
}
}
+ // Emit DW_TAG_template_alias for template aliases? True by default for SCE.
+ const auto *DebugTemplateAlias = Args.getLastArg(
+ options::OPT_gtemplate_
@@ -5361,7 +5383,56 @@ static bool IsReconstitutableType(QualType QT) {
return T.Reconstitutable;
}
-std::string CGDebugInfo::GetName(const Decl *D, bool Qualified) const {
+bool CGDebugInfo::HasReconstitutableArgs(
+ArrayRef Args) const {
+ return llvm::all_of(Args, [&
@@ -465,3 +465,15 @@
// MANGLED_TEMP_NAMES: error: unknown argument
'-gsimple-template-names=mangled'; did you mean '-Xclang
-gsimple-template-names=mangled'
// RUN: %clang -### -target x86_64 -c -g %s 2>&1 | FileCheck
--check-prefix=FULL_TEMP_NAMES
--implicit-check-not=debu
https://github.com/pogo59 commented:
The flag situation is good. Tuning influences the default for the new option,
which is how we want that to work.
I'm happy, a couple of nits left. I'll leave it to others to do the final
approval.
https://github.com/llvm/llvm-project/pull/87623
___
@@ -1332,6 +1332,54 @@ llvm::DIType *CGDebugInfo::CreateType(const
TemplateSpecializationType *Ty,
auto PP = getPrintingPolicy();
Ty->getTemplateName().print(OS, PP, TemplateName::Qualified::None);
+ SourceLocation Loc = AliasDecl->getLocation();
+
+ if (CGM.getCodeGenO
https://github.com/pogo59 edited https://github.com/llvm/llvm-project/pull/87623
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -465,3 +465,15 @@
// MANGLED_TEMP_NAMES: error: unknown argument
'-gsimple-template-names=mangled'; did you mean '-Xclang
-gsimple-template-names=mangled'
// RUN: %clang -### -target x86_64 -c -g %s 2>&1 | FileCheck
--check-prefix=FULL_TEMP_NAMES
--implicit-check-not=debu
@@ -310,6 +310,24 @@ namespace llvm {
DINode::DIFlags Flags = DINode::FlagZero,
DINodeArray Annotations = nullptr);
+/// Create debugging information entry for a template alias.
+/// \param Ty O
@@ -310,6 +310,24 @@ namespace llvm {
DINode::DIFlags Flags = DINode::FlagZero,
DINodeArray Annotations = nullptr);
+/// Create debugging information entry for a template alias.
+/// \param Ty O
pogo59 wrote:
FTR, got an internal report about this. Luckily it was my turn to catch new
bugs and I recognized the issue.
@MaskRay Is it too late to add a Release Note for LLVM 18?
https://github.com/llvm/llvm-project/pull/74809
___
cfe-commits mail
pogo59 wrote:
Yes, driver part LGTM.
https://github.com/llvm/llvm-project/pull/87623
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 edited
https://github.com/llvm/llvm-project/pull/100160
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 approved this pull request.
One nit on the test, otherwise LGTM.
https://github.com/llvm/llvm-project/pull/100160
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-comm
@@ -0,0 +1,33 @@
+; This checks that .debug_aranges is always generated for the SCE debugger
+; tuning.
+
+; RUN: llc -mtriple=x86_64 -debugger-tune=sce -filetype=obj %s -o %t
+; RUN: llvm-dwarfdump -debug-aranges %t | FileCheck %s
+
+; CHECK: .debug_aranges contents:
+; CHEC
https://github.com/pogo59 approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/100423
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/101034
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/101202
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 created
https://github.com/llvm/llvm-project/pull/101257
-falign-functions with no argument is supposed to be equivalent to
-falign-functions=16, according to the documentation. Make it so.
>From a95ab9bfb7be8da6c4ebe0d207beff5a0a72bb9d Mon Sep 17 00:00:00 2001
From:
@@ -1982,7 +1982,7 @@ unsigned tools::ParseFunctionAlignment(const ToolChain
&TC,
return 0;
if (A->getOption().matches(options::OPT_falign_functions))
-return 0;
+return 4; // log2(16)
pogo59 wrote:
Ah, I should have said "the commentary" on Pa
https://github.com/pogo59 updated
https://github.com/llvm/llvm-project/pull/101257
>From a95ab9bfb7be8da6c4ebe0d207beff5a0a72bb9d Mon Sep 17 00:00:00 2001
From: Paul Robinson
Date: Tue, 30 Jul 2024 15:17:14 -0700
Subject: [PATCH 1/2] [Driver] Pass correct alignment for -falign-functions
with n
pogo59 wrote:
The driver can't really be poking down into LLVM's TargetLowering; at best it
could be querying Clang's TargetInfo, which knows a lot about data alignment
but does not currently have any knowledge of code alignment. So, I've reverted
the change and instead fixed the incorrect com
https://github.com/pogo59 edited
https://github.com/llvm/llvm-project/pull/101257
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 edited
https://github.com/llvm/llvm-project/pull/101257
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 closed
https://github.com/llvm/llvm-project/pull/101257
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -354,6 +354,9 @@ DwarfDebug::DwarfDebug(AsmPrinter *A)
UseLocSection = !TT.isNVPTX();
+ // Always emits .debug_aranges for SCE tuning.
pogo59 wrote:
```suggestion
// Always emit .debug_aranges for SCE tuning.
```
https://github.com/llvm/llvm-project
https://github.com/pogo59 approved this pull request.
LGTM with a comment nit.
https://github.com/llvm/llvm-project/pull/99629
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 approved this pull request.
LGTM.
https://github.com/llvm/llvm-project/pull/99862
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 created
https://github.com/llvm/llvm-project/pull/89775
Commit b876596 corrected default compiler-rt library names for many targets,
this patch restores the arch suffix for PlayStation and Windows which both
distribute these libraries with the arch suffix.
>From c8b0
https://github.com/pogo59 edited https://github.com/llvm/llvm-project/pull/89775
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
pogo59 wrote:
Microsoft obviously builds and distributes with the arch suffix, supporting
both 32-bit and 64-bit in the same distro. So I think they will continue to
need it. (I can't actually speak for MS, but it's what I see in the Visual
Studio installation.)
I don't know enough about how
pogo59 wrote:
> if you build with runtimes on windows (which is the suggested way) it will
> end up with the new style without arch suffix
Was someone from Microsoft party to that decision? If not, perhaps it needs to
be revisited. We should not be making decisions about how they want to do
pogo59 wrote:
> what gets emitted as embedded directive, when the compiler doesn't see either
> of them at compile time (with e.g. distributed build systems), while they
> might be available at link time.
This is exactly the scenario we had. I will double-check that it is solved by
my patch,
pogo59 wrote:
> It's been like that for maybe 2-3 years now and no one has complained about it
As a downstream vendor I can tell you we work around things all the time
without bothering to try to fix things upstream (I try to encourage more
engagement, but I can't control what people actually
pogo59 wrote:
> My recollection of the past discussions is that we want users to switch to
> the new hierarchy.
Is Microsoft a "user" who agreed to this change? Was clang-vendors tagged on
this discussion about a new file hierarchy? I don't recall it. The library name
change is news to me, a
pogo59 wrote:
> I think the distributed build system users need to prepare a file hierarchy
> to get the intended --dependent-lib= (like how to prepare --sysroot in
> clang/test/Driver tests), or turn off it with the recent -frtlib-defaultlib.
Is there a Release Note for this new requirement o
pogo59 wrote:
> There have been a few cases before and we did not write specific release
> notes for the driver behavior:
Possibly users of those targets do not depend on distributed builds the way our
users do, and didn't run into the problem? They don't look like Windows
targets, anyway. Di
pogo59 wrote:
> LLVM_ENABLE_PER_TARGET_RUNTIME_DIR=on
I'm unable to find what code this affects. I don't see it mentioned anywhere in
clang/lib or clang/include.
It does seem like it should control the behavior of `ToolChain::getCompilerRT`;
where I had added the Windows/PS check, seems like
pogo59 wrote:
Poking around more, it looks like the canonical way to convert a CMake variable
to a C++ define is to add an entry to
llvm/include/llvm/Config/llvm-config.h.cmake. I'll have a go at doing it that
way.
https://github.com/llvm/llvm-project/pull/89775
__
pogo59 wrote:
> `optnone` is a very strange attribute; it's a directive to the compiler with
> no clear semantics.
In an ideal world, an `optnone` function would be compiled as-if the entire
compilation unit had `-O0`. At `-O0`, Clang attaches `optnone` to every
function, implying that this e
@@ -0,0 +1,105 @@
+// RUN: %clang_cc1 -triple i386-unknown-unknown -target-feature +mmx \
+// RUN: -target-feature +sse2 -O0 -emit-llvm %s -o - | FileCheck %s
+
+// Test that mmx/sse2 shift intrinsics map to the expected builtins.
+
+// Don't include mm_malloc.h, it's system spec
@@ -0,0 +1,387 @@
+// RUN: %clang_cc1 -triple x86_64-unknown-unknown -fsyntax-only \
pogo59 wrote:
Okay.
https://github.com/llvm/llvm-project/pull/105852
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://
https://github.com/pogo59 updated
https://github.com/llvm/llvm-project/pull/105852
>From fbdf1c4a2bc88e1872bb6b91265b142319b7ccc8 Mon Sep 17 00:00:00 2001
From: Paul Robinson
Date: Fri, 23 Aug 2024 09:40:08 -0700
Subject: [PATCH 1/2] [Headers][X86] Add a couple of tests for MMX/SSE
intrinsics
https://github.com/pogo59 edited
https://github.com/llvm/llvm-project/pull/105852
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 edited
https://github.com/llvm/llvm-project/pull/105852
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 edited
https://github.com/llvm/llvm-project/pull/105852
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 closed
https://github.com/llvm/llvm-project/pull/105852
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 created
https://github.com/llvm/llvm-project/pull/106463
This was accidentally matching a metadata record that happend to have three
elements, but wasn't the record of interest. Add CHECKs to make sure we've
found the correct record.
>From 845fb7105d0e23df57ced47c1f1
https://github.com/pogo59 commented:
I didn't really look at the tests.
Rather than saying the new options are "incompatible" with -O0, I'd say they
are "ignored" at -O0, which is in fact how they are implemented. "Incompatible"
suggests to me that there'd be a diagnostic, and there's no need
@@ -3558,15 +3562,26 @@ static llvm::StoreInst
*findDominatingStoreToReturnValue(CodeGenFunction &CGF) {
llvm::BasicBlock *IP = CGF.Builder.GetInsertBlock();
if (IP->empty()) return nullptr;
-// Look at directly preceding instruction, skipping bitcasts and lifetim
@@ -2523,6 +2575,15 @@ llvm::Function *CodeGenModule::getLLVMLifetimeEndFn() {
return LifetimeEndFn;
}
+/// Lazily declare the @llvm.fake.use intrinsic.
+llvm::Function *CodeGenModule::getLLVMFakeUseFn() {
+ if (FakeUseFn)
+return FakeUseFn;
+ FakeUseFn =
+ llvm::
https://github.com/pogo59 edited
https://github.com/llvm/llvm-project/pull/106724
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -1412,6 +1420,39 @@ void
CodeGenFunction::EmitAndRegisterVariableArrayDimensions(
}
}
+/// Return the maximum size of an aggregate for which we generate a fake use
+/// intrinsic when -fextend-lifetimes is in effect.
+static uint64_t maxFakeUseAggregateSize(const ASTCont
@@ -2543,6 +2543,10 @@ void CodeGenModule::ConstructAttributeList(StringRef
Name,
if (shouldDisableTailCalls())
FuncAttrs.addAttribute("disable-tail-calls", "true");
+// Suppress the machine instruction scheduler when -fextend-lifetimes is
on.
+if (CodeGenO
@@ -36,9 +35,6 @@
// RUN: env SCE_ORBIS_SDK_DIR=.. %clang -Winvalid-or-nonexistent-directory
-### -emit-ast -isysroot foo -target x86_64-scei-ps4 %s 2>&1 | FileCheck
-check-prefix=WARN-ISYSROOT -check-prefix=NO-WARN %s
// RUN: env SCE_ORBIS_SDK_DIR=.. %clang -Winvalid-or-nonex
https://github.com/pogo59 approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/98884
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
pogo59 wrote:
Looks like you don't have write permission so I'll merge it for you.
https://github.com/llvm/llvm-project/pull/98884
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 closed https://github.com/llvm/llvm-project/pull/98884
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 updated
https://github.com/llvm/llvm-project/pull/89775
>From c8b04739159fcd4775656aabb4b964272add8c36 Mon Sep 17 00:00:00 2001
From: Paul Robinson
Date: Tue, 23 Apr 2024 08:10:47 -0700
Subject: [PATCH 1/2] [Driver] Restore arch suffix for PS and Windows
Commit b87659
pogo59 wrote:
I've redone the decision to be based on the setting of
LLVM_ENABLE_PER_TARGET_RUNTIME_DIR and updated the tests to work with either
setting as much as possible. Tests that explicitly look for the per-target
_directory_ are enabled only if the CMake variable is ON.
https://github
pogo59 wrote:
Worth noting that this only affects the presumed layout if the driver can't
find either layout. In most installations, it will detect which one you have
installed, and use it correctly. It's when the driver can't find the libs, and
has to guess, that we run into difficulty.
I st
https://github.com/pogo59 edited https://github.com/llvm/llvm-project/pull/89775
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
pogo59 wrote:
@mstorsjo On reflection I agree that a build-time choice for handling this is
not appropriate. It really wants to be target-based. Latest version (a) falls
back to old style for Windows/PS as well as AIX, (b) adds a driver option to
influence that default.
@MaskRay @tru @nico PT
pogo59 wrote:
My understanding is that this change never had a proper Discourse RFC, which
means it was quite possible for vendors not to have noticed it. Certainly I (on
behalf of Sony) was not aware, and I make some effort to keep an eye on things.
There is no good reason to suppose Visual S
pogo59 wrote:
To answer @MaskRay 's question directly, with respect to PlayStation (not
Windows, which is the major problem):
We have exactly one target, so neither the arch suffix nor a target-specific
directory are useful. (Arguably we have two targets, but as we deliver entirely
separate S
pogo59 wrote:
Driver tests are supposed to verify Driver behavior. That is, use `-###` and
make sure the correct option is being passed down to the compiler. They should
not be compiling anything.
https://github.com/llvm/llvm-project/pull/89727
___
c
https://github.com/pogo59 commented:
The LLVM changes don't have a test. I expect there are existing
expression-to-final-DWARF tests that you can modify to throw in a few bit-piece
operators and show they come out correctly.
Offhand it seems quite reasonable to use DW_OP_bit_piece for bitfield
pogo59 wrote:
> I want to hear about opinions why the old compiler-rt file hierarchy is
> preferred by some users.
I think if you really want a range of opinions, you can get them from an RFC.
Not many people will be watching this PR.
https://github.com/llvm/llvm-project/pull/89775
__
https://github.com/pogo59 created
https://github.com/llvm/llvm-project/pull/91401
Debug-info metadata does not have a strictly defined order. Check that elements
are linked to each other correctly, not that metadata appears in a particular
order.
>From 8e23d2bea291ad003417f7f1af6e2143e1352bb2
pogo59 wrote:
This test was failing in our downstream repo because the metadata didn't come
out in exactly the same order.
https://github.com/llvm/llvm-project/pull/91401
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/c
https://github.com/pogo59 closed https://github.com/llvm/llvm-project/pull/91401
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
pogo59 wrote:
Thanks for the link. I've put a comment there pointing here, explaining that
the current default state (at least on Windows) is broken and that this patch
has come to no agreement about how to fix it.
https://github.com/llvm/llvm-project/pull/89775
___
pogo59 wrote:
> @pogo59 - you might find this interesting in terms of bitrotten tests, etc.
Fixing these typos is great. Identifying recurring patterns in the typos would
help motivate changes to FileCheck itself to detect them sooner. I think
@jdenny-ornl might have had some private patches,
@@ -0,0 +1,33 @@
+/// Check PS5 specific interactions between visibility options.
+/// Detailed testing of -fvisibility-from-dllstorageclass is covered elsewhere.
+
+/// Check defaults.
+// RUN: %clang -### -target x86_64-sie-ps5 -x cl -c -emit-llvm %s 2>&1 | \
+// RUN: FileChec
https://github.com/pogo59 edited https://github.com/llvm/llvm-project/pull/92091
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/pogo59 approved this pull request.
LGTM. I noted a couple of redundant checks, but if you think it's better for
them to be explicit, I'm okay with keeping them.
https://github.com/llvm/llvm-project/pull/92091
___
cfe-commits mailing
@@ -0,0 +1,32 @@
+/// Check PS4 specific interactions between visibility options.
+/// Detailed testing of -fvisibility-from-dllstorageclass is covered elsewhere.
+
+/// Check defaults.
+// RUN: %clang -### -target x86_64-scei-ps4 -x cl -c -emit-llvm %s 2>&1 | \
+// RUN: FileChe
https://github.com/pogo59 approved this pull request.
LGTM
https://github.com/llvm/llvm-project/pull/92549
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -0,0 +1,115 @@
+/// For a class that has a vtable and typeinfo symbol for RTTI, if a user marks
+/// either:
+///
+/// (a) The entire class as dllexport (dllimport)
+/// (b) Any non-inline method of the class as dllexport (dllimport)
+///
+/// then Clang must export the vta
https://github.com/pogo59 commented:
I am moderately sure this does what we (Sony) want, but I would like a second
set of eyes on it so I'm not marking it approved.
https://github.com/llvm/llvm-project/pull/92579
___
cfe-commits mailing list
cfe-commi
https://github.com/pogo59 edited https://github.com/llvm/llvm-project/pull/92579
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -1793,6 +1793,36 @@ void ItaniumCXXABI::EmitDestructorCall(CodeGenFunction
&CGF,
ThisTy, VTT, VTTTy, nullptr);
}
+// Check if any non-inline method has the specified attribute.
+template
+static bool CXXRecordNonInlineHasAttr(const CXXRecordDec
https://github.com/pogo59 approved this pull request.
I've been persuaded this is the right direction. LGTM.
https://github.com/llvm/llvm-project/pull/92579
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/
https://github.com/pogo59 commented:
Does the test exercise both modified paths? I'm guessing it only exercises
GetOrCreateLLVMFunction, but I'm not a codegen expert.
https://github.com/llvm/llvm-project/pull/93302
___
cfe-commits mailing list
cfe-com
https://github.com/pogo59 edited https://github.com/llvm/llvm-project/pull/93302
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
201 - 300 of 336 matches
Mail list logo
