The following flips the vectorizer to forced SLP, there is almost
no expected fallout at this point, the remains should be target
specific cost modeling issues.
Bootstrap and regtest on x86_64-unknown-linux-gnu in progress.
* params.opt (vect-force-slp): Default to 1.
---
gcc/params.opt
Kyrylo Tkachov writes:
>> On 15 Nov 2024, at 10:20, Andrew Pinski wrote:
>>
>> On Thu, Nov 14, 2024 at 7:50 PM Soumya AR wrote:
>>>
>>> The SVE SUBR instruction performs a reversed subtract from an immediate.
>>>
>>> This patches enables the emission of SUBR for Neon modes and avoids the
>>>
Hi Jonathan,
On Thu, Nov 14, 2024 at 08:39:07PM +, Jonathan Wakely wrote:
> On 14/11/24 12:36 +, Yury Khrustalev wrote:
> > From: Szabolcs Nagy
> >
> > ...
> >
> > @@ -28093,7 +28093,7 @@ if test "${enable_standard_branch_protection+set}"
> > = set; then :
> > enableval=$enable_standa
Jakub Jelinek 于2024年11月15日周五 16:20写道：
>
> On Fri, Nov 15, 2024 at 04:04:55PM +0800, Hongyu Wang wrote:
> > Following the discussion in pr116738, the insn for UNSPEC_IEEE_MAXMIN
> > actually matches the behavior of if_then_else, so remove the UNSPEC and
> > rewrite related pattern with if_then_else
Hi Kyrill,
> This would make USE_NEW_VECTOR_COSTS effectively the default.
> Jennifer has been trying to do that as well and then to remove it (as it
> would be always true) but there are some codegen regressions that still >
> need to be addressed.
Yes, that's the goal - we should use good tun
Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
-- >8 --
The error here should also check that we aren't nested in another
lambda; in it, at_function_scope_p() will be false.
PR c++/117602
gcc/cp/ChangeLog:
* parser.cc (cp_parser_lambda_introducer): Check if we're i
Hi,
while working on -fmalloc-dce I noticed that tree-ssa-dce.cc still has an
outdated list of builtions that are known to not read memory that can be
replaced by query to fnspec and modref.
If I get things right, dce does some dead store removal, but only on those
memory object that are non-alias
On 15/11/24 08:53 -0500, Jason Merrill wrote:
On 11/15/24 6:04 AM, Jonathan Wakely wrote:
On 14/11/24 23:09 -0500, Jason Merrill wrote:
On 10/18/24 9:38 AM, Jason Merrill wrote:
Currently this installs the sources under $(gxx_include_dir)/bits/,
i.e. /usr/include/c++/15/bits. So with my -fs
On Fri, Nov 15, 2024 at 1:45 PM Jan Hubicka wrote:
>
> >
> > The patch only ever skips just one conversion, never a chain of them and
> > deliberately so, for the reasons shown in the example.
> >
> > However, I have just realized that combining pass_through jump functions
> > during inlining may
Hello Richard,
On Fri, 15 Nov 2024 15:00:44 +0100 (CET), Richard Biener wrote:
> fnspec was originally introduced for PTA, but 'I' and 'U' given the
> pointers escape, do not help PTA, as soon as pointers escape all
> (most) bets are lost.
>
> So where and how exactly do you plan to use those?
On Fri, 15 Nov 2024, Jₑₙₛ Gustedt wrote:
> Hello Richard,
>
> On Fri, 15 Nov 2024 15:00:44 +0100 (CET), Richard Biener wrote:
>
> > fnspec was originally introduced for PTA, but 'I' and 'U' given the
> > pointers escape, do not help PTA, as soon as pointers escape all
> > (most) bets are lost.
>
When trying to change a IV from IV0 < IV1 to IV0' != IV1' we apply
fancy adjustments to the may_be_zero condition we compute rather
than using the obvious IV0->base >= IV1->base expression (to be
able to use > instead of >=?). This doesn't seem to go well.
Bootstrapped and tested on x86_64-unknow
>
> The patch only ever skips just one conversion, never a chain of them and
> deliberately so, for the reasons shown in the example.
>
> However, I have just realized that combining pass_through jump functions
> during inlining may unfortunately have exactly the same effect, so we
> indeed need
This series adds support for:
- FEAT_SVE_B16B16 (contains both SVE and SME instructions)
- FEAT_SME_B16B16
- FEAT_SME_F16F16
- FEAT_SME2p1
I've bundled them together because they depend on each other and because
they need some common preparatory patches. They also depend on
https://gcc.gnu.org/p
Many of the SME ZA intrinsics have two type suffixes: one for ZA
and one for the vectors. The ZA suffix only conveys an element
size, while the vector suffix conveys both an element type and
an element size. Internally, the ZA suffix maps to an integer mode;
e.g. za32 maps to VNx4SI.
For SME2, i
After commit 8833389e90d676baabb35c3e7a021a4f5444a5ba.
gcc/c-family/
* c.opt.urls: Regenerate.
---
gcc/c-family/c.opt.urls | 3 +++
1 file changed, 3 insertions(+)
diff --git a/gcc/c-family/c.opt.urls b/gcc/c-family/c.opt.urls
index 3bfe2b86b24..04f8e2ee008 100644
--- a/gcc/c-family/c.
Now that we no longer exercise vcond{,u,eq} patterns remove unused
infrastructure.
Bootstrap and regtest running on x86_64-unknown-linux-gnu.
* optabs-query.h (get_vcond_icode): Remove.
(get_vcond_eq_icode): Likewise.
* optabs-tree.h (expand_vec_cond_expr_p): Remove code
> > As I wrote in another mail, the only difference I'm aware of is
> > the DECL_NO_INLINE_WARNING_P flag, which is set on bar and not on bar2
> > in this case. That flag is used just for -Winline warning.
> >
> > int baz (void *);
> >
> > struct S {
> > int foo (int n) { return baz (__builtin_a
Hi,
When working to get rid of mask bit OPTION_MASK_P8_VECTOR,
I noticed that the check on ISA_3_0_MASKS_IEEE is actually
to check TARGET_P9_VECTOR, since we check all three mask
bits together and p9 vector guarantees p8 vector and vsx
should be enabled. So this patch is to adjust this first
as p
Hi,
When I was making a patch to rework TARGET_P8_VECTOR, I
noticed that there are some redundant checks and dead code
related to TARGET_DIRECT_MOVE, so I made this patch as one
separated preparatory patch, it consists of:
- Check either TARGET_DIRECT_MOVE or TARGET_P8_VECTOR only
according
On Thu, Nov 14, 2024 at 06:25:49PM +0100, Jakub Jelinek wrote:
> On Thu, Nov 14, 2024 at 10:05:05AM -0500, Andrew MacLeod wrote:
> > The inferred range mechanism is also initialized using cfun, so again
> > introducing a use of cfun shouldnt be an issue.
> >
> > Something like this ought to work I
We fail to demote this to VMAT_ELEMENTWISE and thus run into the three
vector permutation limit (and would not consider to use strided loads
or gathers).
This resolves another bunch of SVE regressions with --param
vect-force-slp=1
Bootstrapped and tested on x86_64-unknown-linux-gnu.
PR t
On Fri, 15 Nov 2024, Filip Kastl wrote:
> Hi,
>
> Andi's greedy bit test finding algorithm was reverted. I found a fix for the
> problem that caused the revert. I made this patch to reintroduce the greedy
> alg into GCC. However I think we should keep the old slow but more powerful
> algorithm
On Fri, 15 Nov 2024, Jennifer Schmitz wrote:
>
>
> > On 7 Nov 2024, at 13:47, Richard Biener wrote:
> >
> > External email: Use caution opening links or attachments
> >
> >
> > On Tue, 5 Nov 2024, Jennifer Schmitz wrote:
> >
> >> We are working on a patch to improve the codegen for the foll
The following ensures that peeling a single iteration for gaps is
sufficient by enforcing niter masking (partial vector use) given
we cannot (always) statically decide when the vector size isn't known.
The condition guarding this and thus statically giving a pass in
some cases for VL vectors is que
Hi Wilco,
> On 14 Nov 2024, at 18:44, Wilco Dijkstra wrote:
>
>
> Add AARCH64_EXTRA_TUNE_USE_NEW_VECTOR_COSTS and
> AARCH64_EXTRA_TUNE_MATCHED_VECTOR_THROUGHPUT
> to the baseline tuning since all modern cores use it. Fix the neoverse512tvb
> tuning to be
> like Neoverse V1/V2.
>
This would
On Fri, 15 Nov 2024, Richard Sandiford wrote:
> Richard Biener writes:
> > The following ensures that peeling a single iteration for gaps is
> > sufficient by enforcing niter masking (partial vector use) given
> > we cannot (always) statically decide when the vector size isn't known.
> > The cond
Gentle ping on the middle-end review for this patch.
There are two parts of this patch:
1. Diagnostic part (Part 2), which has been reviewed by David;
2. Middle end part (Part 1 and 3), mainly on the copy_history information
collection during transformation.
Thanks,
Qing
> On Nov 5, 2024,
On 11/15/24 3:25 AM, Robin Dapp wrote:
So this is really the biggest question in my mind. When we kicked this
around in the patchwork meeting several weeks ago I got the impression
Robin had a correctness concern with this code. Robin, do you remember
what had you worried?
Unfortunately I h
This is both a performance optimization and a partial fix for
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98723.
This commit fixes the issue for bracket expressions that are
not locale-dependant.
---
libstdc++-v3/include/bits/regex_compiler.tcc | 11 +++
1 file changed, 7 insertions(+),
With this change, we can finally build ardour-8.10.0+ds.
Tested on hppa-unknown-linux-gnu and hppa64-hp-hpux11.11. Committed
to gcc-13, gcc-14 and trunk.
Dave
---
hppa: Fix typos in 32-bit SFmode peephole2 patterns
2024-11-15 John David Anglin
gcc/ChangeLog:
PR target/117564
On 29/10/2024 13:41, Richard Biener wrote:
> On Mon, 28 Oct 2024, Alex Coplan wrote:
>
> > From: Tamar Christina
> >
> > The alignment peeling changes exposed a latent missing dominator update
> > with early break vectorization, specifically when inserting the vector
> > skip edge, since the new
* Andrew Pinski:
> On Fri, Nov 15, 2024 at 9:13 AM Florian Weimer wrote:
>>
>> This is another recent GCC extension whose use is apparently
>> difficult to spot in code reviews.
>>
>> The name of the option is due to Jonathan Wakely. Part of it
>> could apply to C++ as well (for labels at the en
On Fri, Nov 15, 2024 at 08:13:28PM +0100, Florian Weimer wrote:
> * Andrew Pinski:
>
> > On Fri, Nov 15, 2024 at 9:13 AM Florian Weimer wrote:
> >>
> >> This is another recent GCC extension whose use is apparently
> >> difficult to spot in code reviews.
> >>
> >> The name of the option is due to
On Thu, Nov 14, 2024 at 06:26:11PM -0600, Peter Bergner wrote:
> On 11/8/24 1:49 PM, Michael Meissner wrote:
> > As part of the architecture flags patches, this patch changes the use of
> > TARGET_POPCNTB to TARGET_POWER5. The POPCNTB instruction was added in ISA
> > 2.02
> > (power5).
>
> I lik
Please, scratch that. I have now sent an updated patch using git
send-email:
https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668988.html
Sorry for the noise! :-)
Best regards,
Luca
Il giorno ven 15 nov 2024 alle ore 16:55 Luca Bacci
ha scritto:
> See https://gcc.gnu.org/bugzilla/show_b
On Fri, Nov 15, 2024 at 11:13:55AM -0500, Andrew MacLeod wrote:
>
> On 11/15/24 04:36, Jakub Jelinek wrote:
> > On Thu, Nov 14, 2024 at 06:25:49PM +0100, Jakub Jelinek wrote:
> > > On Thu, Nov 14, 2024 at 10:05:05AM -0500, Andrew MacLeod wrote:
> > > > The inferred range mechanism is also initiali
On 15/11/24 11:43 -0500, Jason Merrill wrote:
On 11/15/24 10:21 AM, Jonathan Wakely wrote:
On Fri, 15 Nov 2024 at 15:20, Jonathan Wakely wrote:
On Fri, 15 Nov 2024 at 15:12, Jason Merrill wrote:
On 11/15/24 9:20 AM, Jonathan Wakely wrote:
On 15/11/24 08:53 -0500, Jason Merrill wrote:
On
On Fri, Nov 15, 2024 at 6:30 AM Tamar Christina wrote:
>
> Hi All,
>
> This patch makes it so that when you use any of the Cortex-A53 errata
> workarounds but have specified an -march or -mcpu we know is not affected by
> it
> that we suppress the errata workaround.
>
> This is a driver only patc
A kind remind/ping on the patch.
Kind regards,
Aleksandar Rakić
From: Aleksandar Rakic
Sent: Thursday, October 31, 2024 12:46 PM
To: richard.guent...@gmail.com
Cc: gcc-patches@gcc.gnu.org; Djordje Todorovic; Jovan Dmitrovic
Subject: Re: [Bug tree-optimiza
>
> On 11/15/24 10:19, Jan Hubicka wrote:
> > > I have seen this happen when the result of the intersection is an empty
> > > set (one of the few comments in value-range.h actually describes an
> > > undefined range as an "empty range"). I have only seen this happen when
> > > the edge has been r
See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98723
From bf651c78c6ad4fedd0fe6cec3d7953cbf0a4f03a Mon Sep 17 00:00:00 2001
From: Luca Bacci
Date: Fri, 15 Nov 2024 16:32:14 +0100
Subject: [PATCH] Avoid unneeded calls to regex_traits::transform_primary()
To: gcc-patches@gcc.gnu.org
This is both
Hello,
This patch series adds SEH, stack unwinding and C++ exceptions
implementation to the aarch64-w64-mingw32 target. The patch series is
based on
"[PATCH v3 0/8] SMALL code model fixes, optimization fixes, LTO, and minimal
C++ enablement"
patch series, which should be merged next week.
https:/
Unfortunately that broke bootstrap.
https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668554.html
https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668699.html
https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668700.html
https://gcc.gnu.org/pipermail/gcc-patches/2024-November/66
Bootstrapped and regression tested on x86_64.
c: Fix ICE when forming composite type for two structures / unions
[PR117548]
When forming the composite type from two tagged type, we need to find the
original type for a typedecl to get the correct tag.
PR c/117
Hi,
On Fri, Nov 15 2024, Jan Hubicka wrote:
>>
>> On 11/15/24 10:19, Jan Hubicka wrote:
>> > > I have seen this happen when the result of the intersection is an empty
>> > > set (one of the few comments in value-range.h actually describes an
>> > > undefined range as an "empty range"). I have on
The patch bypasses missing prototypes warnings in MinGW until it is
fixed.
libgcc/ChangeLog:
* config/aarch64/t-mingw: Update.
---
libgcc/config/aarch64/t-mingw | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libgcc/config/aarch64/t-mingw b/libgcc/config/aarch64/t-min
From: Alfie Richards
Hi Joseph and all,
I worked through Josephs feedback, and as I fixed certain issues I came to the
conclusion he was correct that a rethink was required.
I reworked this to only have the one FMV binding for each function set which
gets replaced with the dispatched symbol dec
This patch adds support for `target_version` function multiversioning to
the C frontend, specifically intended for enabling this for Aarch64
targets.
The functionality and behavior matches the CPP frontend. Which is to
say it is not ACLE compliant, as I have another patch achieving this.
The maj
Richard, ok for trunk then?
Am Freitag, dem 15.11.2024 um 02:11 -0300 schrieb Thiago Jung Bauermann:
> Hello,
>
> Martin Uecker writes:
>
> > I added a max element as suggested by Richard to force
> > the type to an int.
> >
> > Regression tested on x86_64 but needs testing on arm-eabi.
> >
On Fri, 15 Nov 2024, Martin Uecker wrote:
> Bootstrapped and regression tested on x86_64.
>
>
> c: Fix ICE when forming composite type for two structures / unions
> [PR117548]
>
> When forming the composite type from two tagged type, we need to find the
> original type for a ty
gcc/fortran/ChangeLog
PR middle-end/112779
PR middle-end/113904
* decl.cc (gfc_match_end): Handle COMP_OMP_BEGIN_METADIRECTIVE and
COMP_OMP_METADRIECTIVE.
* dump-parse-tree.cc (show_omp_node): Handle EXEC_OMP_METADIRECTIVE.
(show_code_node): Likewise.
gcc/testsuite/ChangeLog
* c-c++-common/gomp/attrs-metadirective-1.c: New.
* c-c++-common/gomp/attrs-metadirective-2.c: New.
* c-c++-common/gomp/attrs-metadirective-3.c: New.
* c-c++-common/gomp/attrs-metadirective-4.c: New.
* c-c++-common/gomp/attrs-metadirec
This patch reimplements the middle-end support for "declare variant"
and extends the resolution mechanism to also handle metadirectives
(PR112779). It also adds partial support for dynamic selectors
(PR113904) and fixes a selector scoring bug reported as PR114596. I hope
this rewrite also improve
Successfully bootstrapped & regrtested on x86_64-pc-linux-gnu.
OK for trunk?
gcc/cp/ChangeLog:
* name-lookup.cc (suggest_alternative_in_explicit_scope):
Gracefully handle non-namespaces, such as scoped enums.
* parser.cc (cp_parser_name_lookup_error): Provide
a name
libgomp/ChangeLog
* libgomp.texi (OpenMP 5.0): Mark metadirective and declare variant
as implemented.
(OpenMP 5.1): Mark target_device as supported.
Add changed interaction between declare target and OpenMP context
and dynamic selector support.
(OpenM
After reimplementing late resolution of "declare variant", the
declare_variant_alt and calls_declare_variant_alt flags on struct
cgraph_node are no longer used by anything. For the purposes of
marking functions that need late resolution, the
has_omp_variant_constructs flag has replaced
calls_decla
Additional shared C/C++ testcases are included in a subsequent patch in this
series.
gcc/c-family/ChangeLog
PR middle-end/112779
PR middle-end/113904
* c-common.h (enum c_omp_directive_kind): Add C_OMP_DIR_META.
(c_omp_expand_variant_construct): Declare.
* c
r15-5327-g55e3bd376b2214 changes the default to -std=gnu23 but this
test relies on unprototyped functions. Follow Joseph's advice
in that commit and tweak the test accordingly.
gcc/testsuite/ChangeLog:
* gcc.target/i386/pr66891.c: Pass -std=gnu17.
---
OK for trunk?
v2 fixes the commit me
This is the first patch in improving this code.
Since there are a few places which get the index and they
check the same thing, let's factor that out into one function.
Bootstrapped and tested on x86_64-linux-gnu.
gcc/ChangeLog:
* cfgexpand.cc (INVALID_STACK_INDEX): New defined.
On Sat, 2024-11-16 at 03:44 +, Prathamesh Kulkarni wrote:
> diff --git a/libatomic/Makefile.in b/libatomic/Makefile.in
> index 9798e7c09e9..62cd5e0a76b 100644
> --- a/libatomic/Makefile.in
> +++ b/libatomic/Makefile.in
> @@ -1,7 +1,7 @@
> -# Makefile.in generated by automake 1.15.1 from Makefil
This patch adds basic support for three new tree node types that will
be used in subsequent patches to support OpenMP metadirectives and
dynamic selectors.
OMP_METADIRECTIVE is the internal representation of parsed OpenMP
metadirective constructs. It's produced by the front ends and is expanded
d
This is yet another revised version of the patch set for metadirective
and dynamic selector support, along with "declare variant"
fixes/improvements, incorporating comments received on the last
posted version in July:
https://gcc.gnu.org/pipermail/gcc-patches/2024-July/657841.html
The current ver
Presently, the code to handle OpenMP attribute-syntax pragmas in the C
front end assumes nothing else is messing with redirecting
parser->tokens, and makes no provision for restoring it from anything
other than parser->tokens_buf when the buffer allocated for the pragma
is exhausted. Adding suppor
In r12-6650-g5c69acb32329d4 we updated our sources from .c to .cc
since for some time GCC has been implemented in C++, not C.
GCC plugins are also implemented in C++, not C, but the plugins
in our testsuite still have .c extensions.
Rename the plugin implementation files in the testsuite from .c
On Sat, 16 Nov 2024 at 01:09, Jonathan Wakely wrote:
>
> While working on fancy pointer support for the linked lists I noticed
> they didn't have any debug assertions. This adds the obvious non-empty
> assertions to front(), back(), pop_front() and pop_back().
>
> For the pop members, adding an as
Peter Bergner asked me to reorganize my V3 patch that separates the
architecture bits (set via -mcpu=) compared to the ISA bits that are based
on options.
Here is the beginning of the V3 patch for reference:
https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668643.html
As Peter has suggest
The code and test case previously implemented the OpenMP 5.0 spec,
which said in section 2.3.1:
"For functions within a declare target block, the target trait is added
to the beginning of the set..."
In OpenMP 5.1, this was changed to
"For device routines, the target trait is added to the beginni
Additional shared C/C++ testcases are included in a subsequent patch in this
series.
gcc/cp/ChangeLog
PR middle-end/112779
PR middle-end/113904
* cp-tree.h (struct saved_scope): Add new field
x_processing_omp_trait_property_expr.
(processing_omp_trait_proper
After fixing loop-im to do the correct overflow rewriting
for pointer types too. We end up with code like:
```
_9 = (unsigned long) &g;
_84 = _9 + 18446744073709551615;
_11 = _42 + _84;
_44 = (signed char *) _11;
...
*_44 = 10;
g ={v} {CLOBBER(eos)};
...
n[0] = &f;
*_44 = 8;
g ={v} {CLOBBER(eos)};
This is an expansion of the last patch to also track pointers via vector types
and the
constructor that are used with vector types.
In this case we had:
```
_15 = (long unsigned int) &bias;
_10 = (long unsigned int) &cov_jn;
_12 = {_10, _15};
...
MEM[(struct vec *)&cov_jn] ={v} {CLOBBER(bob)};
bi
101 - 171 of 171 matches
Mail list logo
