On Tue, Sep 09, 2025 at 06:49:22PM +, Qing Zhao wrote:
>
> > On Sep 4, 2025, at 20:24, Kees Cook wrote:
> > +For indirect call sites:
> > +
> > +- Keeping indirect calls from being merged (see above) by adding a
> > + wrapping type so that equality was tested based on type-id.
>
> I still t
C++17 has a 'Requires:' precondition that the two random access iterator
types have the same value type. In C++20 that is a 'Mandates:'
requirement which we must diagnose.
Although we could diagnose it in C++17, that might be a breaking change
for any users relying on it today. Also I am lazy and
Hi Kwok,
sorry for the belated review.
[Some minor but trivial changes are required due to bitrotting -
both in gimplify.cc and in libgomp/target.c]
Kwok Cheung Yeung wrote:
Subject: [PATCH 04/11] openmp, fortran: Add support for map iterators in
OpenMP target construct (Fortran)
This add
Don't upgrade TLS model when cplus_decl_attributes is called on a thread
local variable whose TLS model isn't set yet.
gcc/cp/
PR c++/121889
* decl2.cc (cplus_decl_attributes): Don't upgrade TLS model if
TLS model isn't set yet.
gcc/testsuite/
PR c++/121889
On Thu, Sep 11, 2025 at 12:55:52AM +0200, Alejandro Colomar wrote:
> Hi,
>
> On Tue, Sep 09, 2025 at 09:58:47PM +0200, Alejandro Colomar wrote:
> > Here's v2. It's still an RFC because I was unable to build GCC, due to
> > an unrelated build error in the master branch.
>
> I've finished regressi
Hi,
On Tue, Sep 09, 2025 at 09:58:47PM +0200, Alejandro Colomar wrote:
> Here's v2. It's still an RFC because I was unable to build GCC, due to
> an unrelated build error in the master branch.
I've finished regression testing. All's good. Nothing has broken, and
the new tests all pass.
On Wed, 10 Sept 2025 at 22:55, Bruce Korb wrote:
>
> Looks reasonable to me. I don't see a patch to the testing results tho.
> The "SVN-MERGE" warns against using subversion merging the file instead
> of rebuilding and replacing it. The text for "SVN-MERGE" lives in the
> template used to create t
libstdc++-v3/ChangeLog:
* include/bits/unique_ptr.h: Remove blank line.
---
Tested powerpc64le-linux. Pushed to trunk.
libstdc++-v3/include/bits/unique_ptr.h | 1 -
1 file changed, 1 deletion(-)
diff --git a/libstdc++-v3/include/bits/unique_ptr.h
b/libstdc++-v3/include/bits/unique_ptr
Looks reasonable to me. I don't see a patch to the testing results tho.
The "SVN-MERGE" warns against using subversion merging the file instead
of rebuilding and replacing it. The text for "SVN-MERGE" lives in the
template used to create the "fixincl.x" file.
On 8/28/25 06:56, Jonathan Wakely
libstdc++-v3/ChangeLog:
* testsuite/std/time/year_month_day/io.cc: Additional tests.
---
Adding test coverate, before I touch the implementation.
Testing on x86_64-linux locally.
.../testsuite/std/time/year_month_day/io.cc | 48 +++
1 file changed, 28 insertions(+), 20 d
>
>
> > Am 09.09.2025 um 12:54 schrieb Juergen Christ :
> >
> > The length returned by vect_get_loop_len is REALLEN + BIAS, but was
> > assumed to be REALLEN - BIAS. If BIAS is -1, this leads to wrong
> > code.
> >
> > Bootstrapped and regtested on s390. Ok for trunk?
>
> Ok. Can you also
Hi Robin,
> > diff --git a/gcc/config/riscv/tt-ascalon-d8.md
> > b/gcc/config/riscv/tt-ascalon-d8.md
> > index a57c0b31a81..25b99b6129e 100644
> > --- a/gcc/config/riscv/tt-ascalon-d8.md
> > +++ b/gcc/config/riscv/tt-ascalon-d8.md
> > @@ -285,38 +285,38 @@ (define_insn_reservation
> > "tt_ascalo
On 9/9/25 12:00 PM, Tomasz Kaminski wrote:
On Mon, Sep 8, 2025 at 10:14 PM Luc Grosheintz
wrote:
I decided to deviate once (described in detail in the commit message).
Without this deviation writing tests is very clumsy and sometimes not
possible, e.g. when mixing uint8_t and dynamic paddin
C2Y removes various instances of undefined behavior, typically making
them either constraint violations or implementation-defined.
In many but not all such cases, GCC's existing behavior is compatible
with the C2Y changes. For an initial batch of such cases, add
explicit tests for how GCC behaves
On Tue, Sep 9, 2025 at 7:36 AM Takayuki 'January June' Suwa
wrote:
>
> That was introduced in commit 42db504c2fb2b460e24ca0e73b8559fc33b3cf33,
> over 15 years ago!
>
> gcc/ChangeLog:
>
> * config/xtensa/xtensa.h (TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P):
> Change "Xtrnase" in the
On Mon, Sep 8, 2025 at 7:47 AM H.J. Lu wrote:
>
> -mno-direct-extern-access is used to disable direct access to external
> symbol from executable with and without PIE for x86. Require PIE and
> pass -fPIE to disable direct access to external symbol for other targets.
>
> PR fortran/107421
I think this part is no contirvatial, so I will commit this patch
after CI passes.
On Wed, Sep 10, 2025 at 6:13 PM Kito Cheng wrote:
>
> can_find_related_mode_p incorrectly handled VLS (Vector Length Specific)
> types by using TARGET_MIN_VLEN directly, which is in bits, instead of
> converting it
When producing output, the libstdc++ format implementation only uses _Sink_iter
specializations. Since users cannot construct basic_format_context, this is the
only iterator type actually used. The __format_padded helper relies on this
property to efficiently pad sequences from tuples and ranges.
From: Matthew Malcomson
Working on PR119588. From that PR:
We've seen on some internal workloads (NVPL BLAS running GEMM routine
on a small matrix) that the overhead of a `#pragma omp parallel`
statement when running with a high number of cores (72 or 144) is much
higher with the libgomp impleme
On Wed, 10 Sept 2025 at 09:44, Jonathan Wakely wrote:
>
> On Wed, 10 Sept 2025 at 07:49, Luc Grosheintz
> wrote:
> >
> > In libstdc++ the prefix _S is used for static members only. In
> > there's several type aliases that also used the prefix _S. They now use
> > double underscore instead.
>
>
On Wed, 10 Sept 2025 at 10:01, Tomasz Kaminski wrote:
>
>
>
> On Wed, Sep 10, 2025 at 10:54 AM Jonathan Wakely wrote:
>>
>> On Wed, 10 Sept 2025 at 09:52, Jonathan Wakely wrote:
>> >
>> > On Thu, 4 Sept 2025 at 14:39, Tomasz Kaminski wrote:
>> > >> + if constexpr (is_same_v<_Out, _Iter_for
On Wed, 10 Sep 2025, H.J. Lu wrote:
> Since C, C++, and Fortran front-ends now set the TLS access model after
> a variable has been fully processed, not in the middle of processing it,
> add an option, -fipa-tls-access, for TLS access optimization as the part
> of IPA whole-program visibility op
On Wed, 2025-09-10 at 18:23 +0100, Richard Earnshaw (lists) wrote:
> On 10/09/2025 14:06, Jeff Law wrote:
> >
> >
> > On 9/10/25 4:23 AM, Iain Sandoe wrote:
> >
> > > Now we have this facility - and it is firing on my testboxes
> > > (since i use this
> > > script to post-process [per .sum file
On Mon, Sep 8, 2025 at 3:10 PM H.J. Lu wrote:
>
> Both C and C++ frontends should set a tentative TLS model in grokvardecl
> and update TLS mode with the default TLS access model after a TLS variable
> has been fully processed if the default TLS access model is stronger.
>
> PR c/107419
>
On Wed, Sep 10, 2025 at 2:44 AM Alexander Monakov wrote:
>
>
>
> On Wed, 10 Sep 2025, Alexander Monakov wrote:
>
> > Sorry, I think this comment is more confusing than helpful. As mentioned
> > before,
> > decl_default_tls_model does not scan the attribute list, and may return a
> > weaker
> > m
On Tue, Sep 09, 2025 at 04:52:08PM +0200, Jason Merrill wrote:
> > Ok. I wonder whether to keep adding CLOBBER (bob) as before to the start
> > of some of the ctors for -std=c++23 and earlier (unless
> > -ftrivial-auto-var-init= is specified). The current patch does that. If we
> > stop doing th
On Wed, 10 Sep 2025, Patrick Palka wrote:
> Bootstrapped and regtested on x86_64-pc-linux-gnu, does this
> look OK for trunk?
>
> -- >8 --
>
> Some accessor macros unintentionally accept more kinds of nodes than
> they should, e.g. PACK_EXPANSION_PATTERN accepts any EXPR_P node.
Oops, disregard
Bootstrapped and regtested on x86_64-pc-linux-gnu, does this
look OK for trunk?
-- >8 --
Some accessor macros unintentionally accept more kinds of nodes than
they should, e.g. PACK_EXPANSION_PATTERN accepts any EXPR_P node.
gcc/cp/ChangeLog:
* cp-tree.h (PACK_EXPANSION_PATTERN): Assert
Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look
OK for trunk/15?
-- >8 --
We weren't explicitly treating a pack index specifier as a non-deduced
context.
PR c++/121795
gcc/cp/ChangeLog:
* pt.cc (unify) : New non-deduced context
case.
gcc/testsuite/Cha
On Wed, Sep 10, 2025 at 4:57 PM Tomasz Kamiński wrote:
> libstdc++-v3/ChangeLog:
>
> * testsuite/std/time/year_month_day/io.cc: Additional tests.
> ---
> Adding test coverate, before I touch the implementation.
>
Test test shows that %U/%W computest week count since start of specified
yea
When the vectorizer removes a forwarder created earlier by split_edge
it uses redirect_edge_pred for convenience and efficiency. That breaks
down when the edge split is originating from an asm goto as that is
a jump that needs adjustments from redirect_edge_and_branch. The
following factores a si
On today's RISC-V GCC patch review call, someone mentioned there might be an
alternative that uses a hook instead. Jeff mentioned this type of check
may be needed in other pipeline descriptions either now or in the future,
so I thought I'd post what I have so we can discuss which form is preferre
On Wed, 10 Sept 2025 at 15:41, Iain Sandoe wrote:
>
>
>
> > On 10 Sep 2025, at 13:42, Christophe Lyon
> > wrote:
> >
> > On Wed, 10 Sept 2025 at 12:23, Iain Sandoe wrote:
> >>
> >>
> >>
> >>> On 10 Sep 2025, at 11:12, Richard Earnshaw (lists)
> >>> wrote:
> >>>
> >>> On 10/09/2025 10:54, Chri
> On 10 Sep 2025, at 14:06, Jeff Law wrote:
> ]
> On 9/10/25 4:23 AM, Iain Sandoe wrote:
>
>> Now we have this facility - and it is firing on my testboxes (since i use
>> this
>> script to post-process [per .sum file for stability]) - I looked through a
>> few of
>> the reported cases and th
> On 10 Sep 2025, at 13:42, Christophe Lyon wrote:
>
> On Wed, 10 Sept 2025 at 12:23, Iain Sandoe wrote:
>>
>>
>>
>>> On 10 Sep 2025, at 11:12, Richard Earnshaw (lists)
>>> wrote:
>>>
>>> On 10/09/2025 10:54, Christophe Lyon wrote:
If the results include several configurations (sch
> Am 10.09.2025 um 09:27 schrieb Jakub Jelinek :
>
> Hi!
>
> I thought this wouldn't be necessary because RAW_DATA_CST can only appear
> inside of (array) CONSTRUCTORs within DECL_INITIAL of TREE_STATIC vars,
> so there shouldn't be a need to expand it. Except that we have an
> optimization
On Thu, 4 Sept 2025 at 22:27, Christophe Lyon
wrote:
>
> Hi Torbjorn,
> Sorry for the delay
>
>
> On Mon, 1 Sept 2025 at 17:14, Torbjorn SVENSSON
> wrote:
> >
> >
> >
> > On 2025-09-01 16:59, Christophe Lyon wrote:
> > > On Wed, 27 Aug 2025 at 13:25, Torbjorn SVENSSON
> > > wrote:
> > >>
> >
On Wed, 10 Sept 2025 at 12:23, Iain Sandoe wrote:
>
>
>
> > On 10 Sep 2025, at 11:12, Richard Earnshaw (lists)
> > wrote:
> >
> > On 10/09/2025 10:54, Christophe Lyon wrote:
> >> If the results include several configurations (schedule of
> >> variations), do not report summary lines as duplicate
On 10/09/2025 10:54, Christophe Lyon wrote:
> If the results include several configurations (schedule of
> variations), do not report summary lines as duplicates. Indeed with
> several configurations, it's likely that the results contain the same
>
># of expected passesX
>
>
On 9/10/25 2:13 PM, Nathaniel Shead wrote:
Bootstrapped and regtested on x86_64-pc-linux-gnu, OK for trunk?
OK.
-- >8 --
A static member template doesn't always have a DECL_INITIAL, as in the
below testcase, and so checking its TREE_CODE performs a null-pointer
dereference. I don't think we
On Wed, Sep 10, 2025 at 10:20 AM Jonathan Wakely wrote:
> On Sun, 7 Sept 2025 at 16:34, Jonathan Wakely wrote:
> >
> > On Sat, 6 Sept 2025 at 14:12, Luc Grosheintz wrote:
> > >
> > > The concept __integral_constant_like doesn't consider traits with a
> > > boolean member `value` as an integer co
On Wed, Sep 10, 2025 at 12:33 PM Jonathan Wakely wrote:
> On Wed, 10 Sept 2025 at 11:12, Luc Grosheintz
> wrote:
> >
> > In libstdc++ the prefix _S is used for static members only. In
> > there's several type aliases that also used the prefix _S. They now use
> > a single leading underscore fol
Jakub, you did the spaceship cmov adjustments last year.
Is a change like the following where we introduce a cmov ok? It looks like two
of the four functions in the test file were already branchless.
I ran a full bootstrap and regtest again, this time the test is unchanged. I
think I tested
Hi Robin,
> diff --git a/gcc/config/riscv/tt-ascalon-d8.md
b/gcc/config/riscv/tt-ascalon-d8.md
> index a57c0b31a81..25b99b6129e 100644
> --- a/gcc/config/riscv/tt-ascalon-d8.md
> +++ b/gcc/config/riscv/tt-ascalon-d8.md
> @@ -285,38 +285,38 @@ (define_insn_reservation
"tt_ascalon_d8_vec_ordered
LGTM :)
On Wed, Sep 10, 2025 at 6:24 PM Paul-Antoine Arras wrote:
>
> This pattern enables the combine pass (or late-combine, depending on the case)
> to merge a float_extend'ed vec_duplicate into a minus RTL instruction. The
> other
> minus operand is already wide.
>
> Before this patch, we hav
On Wed, 10 Sept 2025 at 11:08, Luc Grosheintz wrote:
>
>
>
> On 9/10/25 11:24 AM, Jonathan Wakely wrote:
> > On Wed, 10 Sept 2025 at 09:44, Jonathan Wakely wrote:
> >>
> >> On Wed, 10 Sept 2025 at 07:49, Luc Grosheintz
> >> wrote:
> >>>
> >>> In libstdc++ the prefix _S is used for static member
can_find_related_mode_p incorrectly handled VLS (Vector Length Specific)
types by using TARGET_MIN_VLEN directly, which is in bits, instead of
converting it to bytes as required.
This patch fixes the issue by dividing TARGET_MIN_VLEN by 8 to convert
from bits to bytes when calculating the number o
> On 10 Sep 2025, at 11:12, Richard Earnshaw (lists)
> wrote:
>
> On 10/09/2025 10:54, Christophe Lyon wrote:
>> If the results include several configurations (schedule of
>> variations), do not report summary lines as duplicates. Indeed with
>> several configurations, it's likely that the r
This pattern enables the combine pass (or late-combine, depending on the case)
to merge a float_extend'ed vec_duplicate into a minus RTL instruction. The other
minus operand is already wide.
Before this patch, we have four instructions, e.g.:
fcvt.d.sfa0,fa0
vsetvli a5,zero,e64
In libstdc++ the prefix _S is used for static members only. In
there's several type aliases that also used the prefix _S. They now use
a single leading underscore follow by a capital letter instead.
libstdc++-v3/ChangeLog:
* include/std/mdspan (_ExtentsStorage::_Base): New name for
On Wed, Sep 10, 2025 at 02:20:58AM -0700, Andrew Pinski wrote:
> On Tue, Aug 26, 2025 at 10:42 PM Stefan Schulze Frielinghaus
> wrote:
> >
> > Bootstrapped and regtested on x86_64 and s390. Ok for mainline?
> >
> > -- >8 --
> >
> > In case an asm operand is an error node, constraints etc. are sti
> Am 10.09.2025 um 09:50 schrieb Jakub Jelinek :
>
> Hi!
>
> THe lowering of .{ADD,SUB,MUL}_OVERFLOW ifns is optimized, so that we don't
> in the common cases uselessly don't create a large _Complex _BitInt
> temporary with the first (real) part being the result and second (imag) part
> just
On Wed, 10 Sep 2025, Alexander Monakov wrote:
> Sorry, I think this comment is more confusing than helpful. As mentioned
> before,
> decl_default_tls_model does not scan the attribute list, and may return a
> weaker
> model than the attribute. If there was no attribute, we can expect that
> r
> Am 10.09.2025 um 10:01 schrieb Jakub Jelinek :
>
> Hi!
>
> This is something that has bothered me for a few years but I've only found
> time for it now.
> The glob used for finding *_1.* etc. counterparts to the *_0.* tests is too
> broad, so if one has say next to *_1.c file also *_1.c~ or
Hi Mike,
On 24/07/25 12:47 am, Michael Meissner wrote:
> diff --git a/gcc/config/rs6000/rs6000-cpus.def
> b/gcc/config/rs6000/rs6000-cpus.def
> index 4a1037616d7..df833e3d54b 100644
> --- a/gcc/config/rs6000/rs6000-cpus.def
> +++ b/gcc/config/rs6000/rs6000-cpus.def
> @@ -83,6 +83,8 @@
> #defin
Robin, given the time since submission, I would suggest a fresh bootstrap and
regression test on x86. Given the V4 passed on x86 and aarch64 in the past
and earlier versions tested well on s390, I think just the sanity check on
x86 is needed.
There are a few regressions on riscv which are all
Hi Robin:
The document is written down in psABI:
https://github.com/riscv-non-isa/riscv-elf-psabi-doc/pull/418/files
Related wording for that is:
A fixed-length vector argument is passed in one vector argument register if the
size of the vector is less than or equal to ABI_VLEN bit.
A fixed-len
ctfstd.org has certificate issue, ignoring which redirects to a
GitHub project and after two links one gets to the ctfstd standard.
There does not appear to be a good, permanent fit for our news list.
Pushed.
Gerald
---
htdocs/gcc-12/changes.html | 2 +-
1 file changed, 1 insertion(+), 1 deletio
On 9/10/25 10:18 AM, Jakub Jelinek wrote:
On Wed, Sep 10, 2025 at 10:06:07AM +0200, Jason Merrill wrote:
It looks like these tests won't run on m68k, can you make at least the -2
test run (and pass) there?
Sure, tested there just with a cross (and only on those 2 new tests).
OK, thanks.
20
On Wed, 10 Sept 2025 at 09:52, Jonathan Wakely wrote:
>
> On Thu, 4 Sept 2025 at 14:39, Tomasz Kaminski wrote:
> >> + if constexpr (is_same_v<_Out, _Iter_for_t<_CharT>>)
> >
> > This would be better if we use _Drop_iter here. This is result of me doing
> > some other tests.
>
> Yes, I agree
On Wed, Sep 10, 2025 at 10:06:35AM +0200, Juergen Christ wrote:
> To properly implement __builtin_ffs for SI mode, implement clz and
> (for >= z17) ctz for SI mode. Otherwise, gcc falls back to a libcall
> which causes problems for Linux kernel code.
>
> Also adjust the C?Z_DEFINED_VALUE_AT_ZERO
On Tue, Sep 09, 2025 at 04:21:14PM +0200, Maximilian Immanuel Brandtner wrote:
> Because of a wrong define_insn for vec_extract_plus a vector access wasn't
> combined with a preceeding plus operation which set the offset at which
> to perform the vector access even though the instruction offers tha
On Sun, 7 Sept 2025 at 16:34, Jonathan Wakely wrote:
>
> On Sat, 6 Sept 2025 at 14:12, Luc Grosheintz wrote:
> >
> > The concept __integral_constant_like doesn't consider traits with a
> > boolean member `value` as an integer constant. This is done to reject
> > various completely unrelated traits
On Wed, Sep 10, 2025 at 10:06:07AM +0200, Jason Merrill wrote:
> It looks like these tests won't run on m68k, can you make at least the -2
> test run (and pass) there?
Sure, tested there just with a cross (and only on those 2 new tests).
2025-09-10 Jakub Jelinek
* mangle.cc (write_rea
Hi Kito,
we discussed this in yesterday's patchwork sync. Would you mind sharing what
the current LLVM implementation does and if this is written down/documented
somewhere?
In particular the chunk size we split large vectors. Like for a 1024b vector
with the "128b ABI", does LLVM use LMUL8
To properly implement __builtin_ffs for SI mode, implement clz and
(for >= z17) ctz for SI mode. Otherwise, gcc falls back to a libcall
which causes problems for Linux kernel code.
Also adjust the C?Z_DEFINED_VALUE_AT_ZERO macros to return 2. Since
the optabs now return exactly the value set by
Hi!
This is something that has bothered me for a few years but I've only found
time for it now.
The glob used for finding *_1.* etc. counterparts to the *_0.* tests is too
broad, so if one has say next to *_1.c file also *_1.c~ or *_1.c.~1~
or *_1.c.orig or *_1.c.bak etc. files, lto.exp will repor
Hi!
THe lowering of .{ADD,SUB,MUL}_OVERFLOW ifns is optimized, so that we don't
in the common cases uselessly don't create a large _Complex _BitInt
temporary with the first (real) part being the result and second (imag) part
just being a huge 0 or 1, although we still do that if it can't be done.
I have double checked that implementation-defined behavior in the [compliance]
(whether the implementation is freestanding) and [stringbuf.const]
(initialization
of sequence pointers) are indeed null, and there are no corresponding entires in
earlier standards.
libstdc++-v3/ChangeLog:
*
Hi!
I thought this wouldn't be necessary because RAW_DATA_CST can only appear
inside of (array) CONSTRUCTORs within DECL_INITIAL of TREE_STATIC vars,
so there shouldn't be a need to expand it. Except that we have an
optimization when reading ARRAY_REF from such a CONSTRUCTOR which will
try to exp
This pattern enables the combine pass (or late-combine, depending on the case)
to merge a float_extend'ed vec_duplicate into a plus RTL instruction.
Before this patch, we have four instructions, e.g.:
fcvt.d.sfa0,fa0
vsetvli a5,zero,e64,m1,ta,ma
vfmv.v.fv3,fa0
vfwad
71 matches
Mail list logo
