Thanks. The patch is LGTM.
Tomasz Kaminski 於 2025年10月25日 週六 上午1:35寫道：
>
>
> On Fri, Oct 24, 2025 at 7:03 PM Hewill Kang wrote:
>
>> `volatile std::reference_wrapper` will cause `_Ur __uref = __ref;`
hard error. But I'm not sure we really need to care about it.
> result_of (where
After r16-4081-g966cdec2b2 which added folding of __builtin_assume_aligned,
forwprop would propagate pointers that lower alignment replacing ones with
greater alignment. This causes us to lose alignment information that
__builtin_assume_aligned provided to expand. Normally this just loses some
opt
The current implementation of the fnmam4 instruction template requires
the third source operand to be assigned the same hard register as the
target operand, but the constraint is not documented in the instruction
manual or standard template definitions. The current constraint will
generate addition
Equalize the rtx cost of mulw.d.w[u] with mul.d.
There are two main benefits:
1. Reducing the number of instructions.
- lu32i.d $r12,0
- bstrpick.d $r4,$r4,31,0
- mul.d $r4,$r4,$r12
- srli.d $r4,$r4,33
---
+ mulh.wu $r4,$r4,
When the bit width of the data type before extension plus
the bit count of the left-shift operation equals 32,
optimizations similar to the following can from:
ext.w.b $r4,$r5
slli.d $r4,$r4,24
to:
slli.w $r4,$r5,24
gcc/ChangeLog:
* config/loongarch/loongarch.md
> On 24 Oct 2025, at 17:51, Alfie Richards wrote:
>
> On 04/10/2025 22:15, Jeff Law wrote:
>> On 10/3/25 12:27 PM, Yangyu Chen wrote:
>>> This patch adds support for target_clones table option. The
>>> target_clones table option allows users to specify multiple versions
>>> of a function and s
This patch adds missing constructors and assignment operators for tuple<>
to support construction and assignment from other empty tuple-like types
such as array, completing the C++23 tuple-like support.
The following members are added:
- tuple(_UTuple&&): Constructor from empty tuple-like types
-
This PR was fixed by one of the other patches, just adding the testcase.
Andrew
From c9a4bd0dca16ad4de299e9f0c4a2afef6b85b8f0 Mon Sep 17 00:00:00 2001
From: Andrew MacLeod
Date: Fri, 24 Oct 2025 17:32:32 -0400
Subject: [PATCH 2/2] Add test case for PR 110405
PR was fixed bu one of the other pat
We make some attempts to improve the ranges we pick up from
cond_exprs. We miss the case where its tied to a single ssa-name. ie
_2 = value_3 < 20
_4 = _2 ? 20 : value_3
This is a common idiom for min or max, but it sometimes eludes us for
no good reason.
fold_using_range::condexpr_a
On Fri, Oct 24, 2025 at 10:23 PM Georg-Johann Lay wrote:
>
> Am 24.10.25 um 15:45 schrieb H.J. Lu:
> > For x86, volatile memory can be used as source operand for loads. Add
> > -fcombine-op-with-volatile-memory-load to allow operations with volatile
> > memory load:
> >
> > 1. Add a field, try_co
Hi All,
Embarrassingly, I had run the regression tests for V1, but apparently not
looked at the results.
Sorry about that, here's a not totally broken version.
Regression tested (properly) for AArch64.
Okay for master?
Thanks,
Alfie
-- >8 --
Adds support for the AArch64 2024 fmmla extensions
This followup patch ensures that any unknown symbol spell check
error in the instrinsics uses the parameter token rather than the
procedure name token. In turn this allows the filter module to
detect and remove multiple unknowns at the same token.
The patch also adds spell checking to the instrins
After r16-3956-gf613fdc6920c83, these 2 testcases start to fail as
now SRA will fully scalarize the structs and only provide with the piece
that was used. But the testcase is testing about the padding so turning
off SRA is correct in this case.
Pushed as obvious after testing the testcases now pas
The definition of one_cmplsi2 insn pattern is not mandatory; the insn
generator recognize the absence of the pattern and instead generates an XOR
with a pseudo whose value is -1 if "XOR reg, reg, imm" is not also provided.
There is an advantage to exposing constant-valued registers early in RTL
pa
Dear All,
this patch does two things: it extends the inline generated code for
IS_CONTIGUOUS to check the dtype of a passed pointer for span equal
to elem_len. It also fixes the assignment of pointers to ensure
that the dtype is the same whether bounds-remapping is done or not.
While developing
Am 24.10.25 um 02:59 schrieb Jerry D:
On 10/23/25 1:03 PM, Harald Anlauf wrote:
Dear All,
this patch remedies the approach used for PR102891 which turned
out to be not generalizable and otherwise could lead to unwanted
temporaries. We now check directly whether the SOURCE argument
of TRANSFER
On Fri, Oct 24, 2025 at 2:09 PM Tomasz Kamiński wrote:
>
> From: Eric Pimentel Aguiar
>
> When exporting declarations in a namespace with using declarations, the
> name must be fully qualified. Recently introduced std::full_extent{,_t},
> std:strided_slice, and std::submdspan_mapping_result broke
From: Eric Pimentel Aguiar
When exporting declarations in a namespace with using declarations, the
name must be fully qualified. Recently introduced std::full_extent{,_t},
std:strided_slice, and std::submdspan_mapping_result broke module std
and module std.compat.
libstdc++-v3/ChangeLog:
Switch std::valarray memory allocation to use std::__new_allocator
to allocate and deallocate memory. This adds support for extended
alignment types without needing to duplicate all the logic from
__new_allocator::allocate and __new_allocator::dellocate.
std::__new_allocator is used instead of std
libstdc++-v3/ChangeLog:
PR libstdc++/122401
* testsuite/20_util/shared_ptr/atomic/pr118757.cc: Add -pthread
for et pthread.
* testsuite/20_util/weak_ptr/pr118757.cc: Likewise.
* testsuite/30_threads/condition_variable/members/116586.cc:
Likewise. Onl
libstdc++-v3/ChangeLog:
* testsuite/util/testsuite_iterators.h (test_container): Add
iterator and value_type typedefs.
---
I wanted to use test_container::value_type in a new test I was writing,
and though it would be worth adding independently of the new test.
Tested powerpc64le
On Fri, Oct 24, 2025 at 7:03 PM Hewill Kang wrote:
> `volatile std::reference_wrapper` will cause `_Ur __uref = __ref;` hard
>>> error. But I'm not sure we really need to care about it.
>>>
result_of (where __inv_unwrap is sourced from), does not care about
it. So I would say, we do no
According to your explanation, I have noticed that the following code
should be rejected because the constraints are not satisfied (i.e.,
is-invocable-using should be false).
Hewill Kang 於 2025年10月25日 週六 上午12:15寫道：
> Here are some examples:
>
> struct S {
> void f() const;
> };
>
> int main(
I use "!", so libstdc++ should pass, no?
Tomasz Kaminski 於 2025年10月25日 週六 上午1:09寫道：
>
>
> On Fri, Oct 24, 2025 at 7:03 PM Hewill Kang wrote:
>
>> `volatile std::reference_wrapper` will cause `_Ur __uref = __ref;`
hard error. But I'm not sure we really need to care about it.
> resu
>
> `volatile std::reference_wrapper` will cause `_Ur __uref = __ref;` hard
>> error. But I'm not sure we really need to care about it.
>>
>>> result_of (where __inv_unwrap is sourced from), does not care about it.
>>> So I would say, we do not need to.
>>
>>
Thanks for the explanation. Should I f
On Fri, Oct 24, 2025 at 7:03 PM Hewill Kang wrote:
> `volatile std::reference_wrapper` will cause `_Ur __uref = __ref;` hard
>>> error. But I'm not sure we really need to care about it.
>>>
result_of (where __inv_unwrap is sourced from), does not care about
it. So I would say, we do no
Two were fixed for GCC 15 and this fixes the other two. This also reverts an
unintentional change introduced by the initial fix.
Tested on x86-64/Linux, applied on the mainline and 15 branch.
2025-10-24 Nicolas Boulenguez
PR ada/81087
* gnatlink.adb (Is_Prefix): Move around
This patch corrects CO-RE generation for the cases where an expression
starts with a non-CO-RE access, but in the middle it requires to
generate CO-RE to correctly compute the access location.
It fixes it by splitting the expression into their CO-RE and non-CO-RE
counterparts. It performs this by
The compiler overruns the source file buffer.
Tested on x86-64/Linux, applied on the mainline and 15 branch.
2025-10-24 Eric Botcazou
PR ada/118782
* styleg.adb (Is_Box_Comment): Also stop the loop at EOF.
--
Eric Botcazoudiff --git a/gcc/ada/styleg.adb b/gcc/ada/styleg.adb
The last character of the argument is chopped as if it was a quote.
Tested on x86-64/Windows, applied on the mainline.
2025-10-24 Mivirl
PR ada/122367
* rtinit.c (__gnat_runtime_initialize) [__MINGW32__]: Fix detection
of quoted arguments.
--
Eric Botcazoudiff --git
This was caught by a static analyzer some time ago.
Tested on x86-64/Windows, applied on the mainline.
2025-10-24 Eric Botcazou
PR ada/98879
* terminals.c (__gnat_setup_child_communication) [_WIN32]: Add else
blocks in the processing of the data returned by ReadFile.
On 24/10/2025 17:27, Alice Carlotti wrote:
On Fri, Oct 24, 2025 at 04:36:46PM +0100, Karl Meakin wrote:
On 24/10/2025 15:26, Kyrylo Tkachov wrote:
On 16 Sep 2025, at 12:43, via Sourceware Forge
wrote:
From: Karl Meakin
I’ve added Karl explicitly on CC myself as I think the forge doesn’
On Fri, Oct 24, 2025 at 04:36:46PM +0100, Karl Meakin wrote:
>
> On 24/10/2025 15:26, Kyrylo Tkachov wrote:
> >
> > > On 16 Sep 2025, at 12:43, via Sourceware Forge
> > > wrote:
> > >
> > >
> > > From: Karl Meakin
> > >
> > I’ve added Karl explicitly on CC myself as I think the forge doesn’
On 10/24/25 05:29, Richard Biener wrote:
I'd say commit it and re-open the bug - it has not been "fixed", but
it was made latent anyway.
Richard.
I checked it in. I was just thinking, should I do anything to the
testcase for 111003? mark it XFAIL or something or just leave it as is?
And
Here are some examples:
struct S {
void f() const;
};
int main() {
S s;
volatile auto sr = std::ref(s);
std::function_ref fr(std::nontype<&S::f>, sr);
fr();
}
Hewill Kang 於 2025年10月25日 週六 上午12:10寫道：
> `volatile std::reference_wrapper` will cause `_Ur __uref = __ref;` hard
> error.
`volatile std::reference_wrapper` will cause `_Ur __uref = __ref;` hard
error. But I'm not sure we really need to care about it.
Tomasz Kamiński 於 2025年10月24日 週五 下午11:33寫道：
> To reduce instantiation count, function_ref(nontype<&S::x>, r) previously
> reused the invoker from function_ref(nontype<
Tomasz Kaminski [Friday, 24 October 2025, 17:40:40 CEST]:
> > #endif // __glibcxx_integer_sequence
> >
> > +#if __cpp_structured_bindings >= 202411L
> > +#if __has_builtin(__integer_pack)
> > + template
> > +inline constexpr _Tp
> > +_IotaArray[_Num] = {__integer_pack(_Tp(_Num))...};
>
On 10/24/25 05:29, Richard Biener wrote:
On Thu, Oct 23, 2025 at 8:49 PM Andrew MacLeod wrote:
The attached patch fixes PR 118254.
It does 2 things:
1) when performing a cast, we cast sub pairs and union the results.
When this gets to VARYING we immediately short circuit and return.
THe erro
The current algorithm for performing bitwise and operations predates
multi-ranges. AS such, it has difficulty with complex signed operations
when the range crosses the boundry between signed and unsigned.
In this case, we are looking at [-INF, 128] & -9, which basically
gives up and returns
On Fri, Oct 24, 2025 at 5:34 PM Matthias Kretz wrote:
> - Moved out of #if __glibcxx_integer_sequence into #if
> __cpp_structured_bindings >= 202411L.
> - Renamed to _IotaArray. Renamed test accordingly.
>
> Tested on x86_64-linux. (__integer_pack else branch tested, too)
> OK for trunk?
>
>
On 24/10/2025 15:26, Kyrylo Tkachov wrote:
On 16 Sep 2025, at 12:43, via Sourceware Forge
wrote:
From: Karl Meakin
I’ve added Karl explicitly on CC myself as I think the forge doesn’t do it
currently.
Document all the `DEF_*` macros and their arguments.
gcc/ChangeLog:
* config/aa
To reduce instantiation count, function_ref(nontype<&S::x>, r) previously
reused the invoker from function_ref(nontype<&S::x>, &r). This assumed r was
always a reference to S or a derived class. However, this constructor is also
valid for lvalues (but not rvalues) of reference_wrapper specializatio
- Moved out of #if __glibcxx_integer_sequence into #if
__cpp_structured_bindings >= 202411L.
- Renamed to _IotaArray. Renamed test accordingly.
Tested on x86_64-linux. (__integer_pack else branch tested, too)
OK for trunk?
- 8< --
This patch introduces th
On Fri, 24 Oct 2025 at 14:55, Eric Pimentel Aguiar
wrote:
>
> Hi,
>
> This is my first patch to the GCC mailing, so feedback is appreciated. I
> may have gone overboard with detail for a mere four-line change. I
> noticed trunk builds of libstdc++ broke previously working builds of an
> experiment
On 24/10/2025 15:26, Kyrylo Tkachov wrote:
On 16 Sep 2025, at 12:43, via Sourceware Forge
wrote:
From: Karl Meakin
I’ve added Karl explicitly on CC myself as I think the forge doesn’t do it
currently.
Document all the `DEF_*` macros and their arguments.
gcc/ChangeLog:
* config/aa
On Fri, Oct 24, 2025 at 4:23 PM Georg-Johann Lay wrote:
> Can it break volatile correctness or so?
I have done some smoke tests with the patched compiler with (gcc -O2):
--cut here--
volatile int m;
int foo (void)
{
int h = m;
int r = m;
asm volatile ("barrier" ::: "memory");
int p = m
l Core i7-1195G7.
I have built the latest git defconfig linux kernel with the patched
current gcc trunk. The patched compiler compiles the kernel without
problems and I was able to boot the built kernel under qemu:
Linux version 6.18.0-rc2-00120-g6fab32bb6508 (uros@fedora) (xgcc (GCC)
16.0.0 20251
Am 24.10.25 um 15:45 schrieb H.J. Lu:
For x86, volatile memory can be used as source operand for loads. Add
-fcombine-op-with-volatile-memory-load to allow operations with volatile
memory load:
1. Add a field, try_combine_insn, to recog_data_d for the combine pass to
pass the instruction in
> On 16 Sep 2025, at 12:43, via Sourceware Forge
> wrote:
>
>
> From: Karl Meakin
>
I’ve added Karl explicitly on CC myself as I think the forge doesn’t do it
currently.
> Document all the `DEF_*` macros and their arguments.
>
> gcc/ChangeLog:
>
> * config/aarch64/aarch64-neon-sve-bri
Hi Claudiu,
There’s no earlier discussion. The ">" lines in the commit message were just
used to show code examples, not replies to previous emails.
Luis
On Fri, Oct 24, 2025 at 1:59 PM Claudiu Zissulescu Ianculescu
wrote:
> Hi Luis,
>
> It seems I've got only this email, and it looks like it
On 24 October 2025 14:27:23 CEST, Jonathan Wakely wrote:
>On Fri, 24 Oct 2025 at 13:11, Matthias Kretz wrote:
>>
>> Matthias Kretz [Friday, 24 October 2025, 14:04:16 CEST]:
>> > > > > #endif // __glibcxx_integer_sequence
>> > > >
>> > > > This is defined for C++14 and later, but I think it ca
Hi Luis,
It seems I've got only this email, and it looks like it is part of a
larger discussion. Did I miss it?
Claudiu
On Wed, Oct 22, 2025 at 3:37 PM Luis Silva wrote:
>
> From: Loeka Rogge
>
> V2HI vectors, explicitly or auto-generated, could be stored in memory wrongly
> due to endianness.
Hi,
This is my first patch to the GCC mailing, so feedback is appreciated. I
may have gone overboard with detail for a mere four-line change. I
noticed trunk builds of libstdc++ broke previously working builds of an
experimental C++ codebase of mine that tests new C++26 features. The
problem can b
Updated the git commit log as suggested by Thomas. The patch itself
remains unchanged.
On 24/10/2025 10:13, Paul-Antoine Arras wrote:
Testcase c-c++-common/gomp/attrs-metadirective-2.c failed in C++ when
OFFLOAD_TARGET_NAMES=nvptx-none. That was caused by label mangling being applied
to the use
The following does a simple legitimising attempt on the SLP graph
permutations before trying to optimize them. For the case we have
a single non-zero layout we can force that to all partitions if
it is compatible. This way we end up with the most canonical
(and possibly no-op) load permutations a
On Thu, Oct 23, 2025 at 1:07 AM Takayuki 'January June' Suwa
wrote:
>
> Because 'i'-constraint clearly includes 'n'.
>
> gcc/ChangeLog:
>
> * config/xtensa/xtensa.md (call_internal, call_value_internal,
> sibcall_internal, sibcall_value_internal): Remove 'n'-constraint.
> ---
> g
On Fri, 24 Oct 2025 at 13:11, Matthias Kretz wrote:
>
> Matthias Kretz [Friday, 24 October 2025, 14:04:16 CEST]:
> > > > > #endif // __glibcxx_integer_sequence
> > > >
> > > > This is defined for C++14 and later, but I think it can't actually be
> > > > used for anything useful without C++26 pack
On Fri, 24 Oct 2025 at 10:54, Tomasz Kamiński wrote:
>
> This patch fixes a missing forwarding-reference (&&) in _Bind_fn_t::operator()
> and lambda returned from not_fn.
>
> The bind_front/bind_back tests were updated to use a structure similar
> to r16-3398-g250dd5b5604fbc to cover cases involvi
On Fri, Oct 24, 2025 at 2:04 PM Matthias Kretz wrote:
> > > > #endif // __glibcxx_integer_sequence
> > >
> > > This is defined for C++14 and later, but I think it can't actually be
> > > used for anything useful without C++26 packs introduced by structured
> > > bindings, right?
> > >
> > > I wo
Matthias Kretz [Friday, 24 October 2025, 14:04:16 CEST]:
> > > > #endif // __glibcxx_integer_sequence
> > >
> > > This is defined for C++14 and later, but I think it can't actually be
> > > used for anything useful without C++26 packs introduced by structured
> > > bindings, right?
> > >
> > > I
This followup to PR modula2/122241 reduces error message clutter by
filtering unknown symbol error ensuring that only one error message
is emitted for an unknown symbol at a particular location.
The filter is implemented using two binary trees. A new generic
(based on the address type) binary dict
> > > #endif // __glibcxx_integer_sequence
> >
> > This is defined for C++14 and later, but I think it can't actually be
> > used for anything useful without C++26 packs introduced by structured
> > bindings, right?
> >
> > I wonder if we want to restrict the definition of _IndexPack to C++26?
>
On Fri, Oct 24, 2025 at 1:45 PM Jonathan Wakely wrote:
> On Fri, 24 Oct 2025 at 12:38, Matthias Kretz wrote:
> >
> > Tested on x86_64-linux. (Not tested with Clang yet.) OK for trunk?
> >
> > - 8<
> >
> > This patch introduces the internal helper type _In
On Fri, 24 Oct 2025 at 12:38, Matthias Kretz wrote:
>
> Tested on x86_64-linux. (Not tested with Clang yet.) OK for trunk?
>
> - 8<
>
> This patch introduces the internal helper type _IndexPack to simplify
> defining a pack of indices via a structured bindi
Tested on x86_64-linux. (Not tested with Clang yet.) OK for trunk?
- 8<
This patch introduces the internal helper type _IndexPack to simplify
defining a pack of indices via a structured binding declaration:
constexpr auto [...__is] = _IndexPack;
_IndexPac
The following does a simple legitimising attempt on the SLP graph
permutations before trying to optimize them. For the case we have
a single non-zero layout we can force that to all partitions if
it is compatible. This way we end up with the most canonical
(and possibly no-op) load permutations a
On Fri, Oct 24, 2025 at 4:32 PM H.J. Lu wrote:
>
> Adjust gcc.target/i386/builtin-fabs-2.c to scan both (%rdi) and (%edi).
>
> PR target/122323
> * gcc.target/i386/builtin-fabs-2.c: Also scan (%edi)for x32.
>
> --
> H.J.
Backporting it to release branches.
--
H.J.
On Fri, 24 Oct 2025, Alex Coplan wrote:
> I recently ran into an ICE that was fixed by richi's
> r16-4571-g1ceda79ca5fe1a1a296624a98de8fd04958fbe55.
>
> This adds a testcase for that fix.
>
> The same testcase doesn't ICE with GCC 15, so there's no need for a
> backport or PR.
>
> Tested on aar
I recently ran into an ICE that was fixed by richi's
r16-4571-g1ceda79ca5fe1a1a296624a98de8fd04958fbe55.
This adds a testcase for that fix.
The same testcase doesn't ICE with GCC 15, so there's no need for a
backport or PR.
Tested on aarch64-linux-gnu (ICEs prior to r16-4571, PASSes afterwards).
Hi,
I have incorporated the changes suggested in the previous version. For now the
bare bones version of multiply by constant pattern recog is implemented in
tree-vect-generic.cc, which should help fixing the failing test cases in 119702.
But to fix PR122065, will need to implement the lowering
Hi All,
Reg tested on aarch64-linux-gnu.
Okay for master?
Alfie
-- >8 --
Adds support for the AArch64 2024 fmmla extensions.
Note this includes a work around for spurious warnings from binutils with
movprfx and fmmla instructions (PR gas/33562).
gcc/ChangeLog:
* config/aarch64/aarch
This patch fixes a missing forwarding-reference (&&) in _Bind_fn_t::operator()
and lambda returned from not_fn.
The bind_front/bind_back tests were updated to use a structure similar
to r16-3398-g250dd5b5604fbc to cover cases involving zero, one, and many bound
arguments.
PR libstdc++/122
On 04/10/2025 22:15, Jeff Law wrote:
On 10/3/25 12:27 PM, Yangyu Chen wrote:
This patch adds support for target_clones table option. The
target_clones table option allows users to specify multiple versions
of a function and select the version at runtime based on the specified
table.
The targe
The code checking whether we have a single cycle and tracking the
reduction chain was not transitioned to full SLP which now shows
when having a SLP reduction chain built after associating the
reduction operation.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
* tree-vect-lo
The following fixes the mixup between reduction operation and conversion
wrapping a reduction chain. This also exposes a missed optimization
but I'm going to fix that in a followup.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
PR tree-optimization/122406
* tree-ve
On Fri, Oct 24, 2025 at 11:15 AM Tomasz Kamiński
wrote:
> This patch fixes a missing forwarding-reference (&&) in
> _Bind_fn_t::operator().
> This issue was detected when the bind_front/bind_back tests were
> updated
> to use a structure similar to r16-3398-g250dd5b5604fbc to cover cases
> involv
On Fri, Oct 24, 2025 at 11:32 AM Josef Melcr wrote:
>
> Hi,
> this patch remvoes the annotation which causes the build to fail when
> configured with --enable-gather-detailed-mem-stats. I am very familiar
> with the mem stat system yet, so I'd leave annotating these functions
> for a future patch
Hi,
this patch remvoes the annotation which causes the build to fail when
configured with --enable-gather-detailed-mem-stats. I am very familiar
with the mem stat system yet, so I'd leave annotating these functions
for a future patch.
Bootstrapped on x86_64-linux. OK for master?
Best regards,
J
On Thu, Oct 23, 2025 at 8:49 PM Andrew MacLeod wrote:
>
> The attached patch fixes PR 118254.
>
> It does 2 things:
>
> 1) when performing a cast, we cast sub pairs and union the results.
> When this gets to VARYING we immediately short circuit and return.
> THe error was if there was a bitmask, i
On Thu, 23 Oct 2025, Richard Sandiford wrote:
> Richard Biener writes:
> > On Wed, 22 Oct 2025, Richard Sandiford wrote:
> >
> >> Richard Biener writes:
> >> > The following tries to addess one shortcoming of the SLP permute
> >> > optimization phase which is that it tries to refuse layouts that
Hi Segher,
Please ignore the v7 patch, had minor errors in capitalization.
Thanks for the review of the v6 patch. I have incorporated all the mentioned
changes in this patch.
Bootstrapped and regtested on powerpc64le-linux with no regressions.
Ok for trunk?
As mentioned in the PATCH v6 thread va
Testcase c-c++-common/gomp/attrs-metadirective-2.c failed in C++ when
OFFLOAD_TARGET_NAMES=nvptx-none. That was caused by label mangling being applied
to the use but not to the declaration.
This is now fixed by mangling the declaration as well.
PR middle-end/122378
gcc/cp/ChangeLog:
This patch fixes a missing forwarding-reference (&&) in _Bind_fn_t::operator().
This issue was detected when the bind_front/bind_back tests were updated
to use a structure similar to r16-3398-g250dd5b5604fbc to cover cases involving
zero, one, and many bound arguments.
libstdc++-v3/ChangeLog:
This patch changes the implementation of bind_front and bind_back to
return a _Bind_front_t<_Bind_fn_t, ...> and _Bind_back_t<_Bind_fn_t, ...>
respectively, replacing the previous lambda-based implementation. The prior use
of a lambda caused non-conforming behavior with respect to C++23 [func.requi
Adjust gcc.target/i386/builtin-fabs-2.c to scan both (%rdi) and (%edi).
PR target/122323
* gcc.target/i386/builtin-fabs-2.c: Also scan (%edi)for x32.
--
H.J.
From 3caec017d613f7a5f3ed2bf150c00b8c1bead6b8 Mon Sep 17 00:00:00 2001
From: "H.J. Lu"
Date: Fri, 24 Oct 2025 16:23:19 +0800
Subject: [PA
On 24/10/2025 06:15, Avinash Jayakar wrote:
Thanks for the info. I also was able to build and debug it for this
target.
For the signed mod case, there is an extra requirement for
vectorization that is to have vector operation support for ABS_EXPR,
which amdgcn does not have.
5033if (!unsigne
Good morning,
This prevents warnings complaining about the redefinition
on top of the base version, at compiler build time.
2025-10-16 Olivier Hainque
* config/rs6000/vxworks.h (SET_CMODEL): Undefine before
(re)defining.
0001-Undefine-SET_CMODEL-before-define-in-rs6000-vxwor
Greetings,
check_weak_available was reporting weak symbols unsupported
for vxworks unconditionally, while they are actually supported
vxworks 7 now (assumed >= r2). This change adjusts the
predicate to reflect that.
We used to believe we should distinguish kernel and rtp modes,
and experiments s
88 matches
Mail list logo
