Re: [PATCH v6 0/3] fmv: Add -ftarget-clones-table option support

@Jeff: I think my outlook accidentally tried to add an emoji to one of the previous emails in this series as I was reading through them, apologies for that. I'd like to also note my interest in this feature, I think this is the kind of thing that would make it easier to make use of FMV. In fa

[gcc-wwwdocs PATCH] gcc-12/13/14/15/16: Mention changes for Intel x86_64 and add changes between minor versions

Hi all, It is yet another year end and time to add the changes for Intel x86_64 backend for GCC 16. Nothing special for that part. In GCC 16 timeframe, there are quite a few changes for ISA enabled on compiler switches backported to previous release branch, making the switch behavior has minor di

Re: [PATCH v2] x86: Add AMD znver6 processor support

On Mon, Dec 15, 2025 at 12:21 PM Umesh Kalvakuntla wrote: > > Hi, > > Thank you Hongtao Liu, Haochen Jiang for the reviews. > Addressed all the comments received so far. > > > You can directly use _mm512_undefined_epi32 and _mm512_setzero_epi32, > > and the definition of _mm512_setzero_epi8/_mm512

RE: [PATCH v2] x86: Add AMD znver6 processor support

> From: Umesh Kalvakuntla > Sent: Monday, December 15, 2025 12:21 PM > To: [email protected] > Subject: [PATCH v2] x86: Add AMD znver6 processor support > > Hi, > > Thank you Hongtao Liu, Haochen Jiang for the reviews. > Addressed all the comments received so far. > > > You can directly u

Re: [PATCH 2/5] a68: add Poke description of module exports for 32-bit systems

Thanks for the patch. I have installed it on your behalf after fixing the changelog in the log. > This commit adds new types for 32-bit systems and rename current > types to have a `_64' suffix. > > Signed-off-by: Mohammad-Reza Nabipoor > > gcc/ChangeLog > > * algol68/ga68-exports.pk (ga

Re: [PATCH 1/5] a68: fix comment and magic number in ga68-exports.pk

Hi Mohammad. Thanks for the patch. I have applied it on your behalf with a nit below. > Signed-off-by: Mohammad-Reza Nabipoor > > gcc/ChangeLog > > * algol68/ga68-exports.pk: Fix comment and value for magic > number. The right changelog to use in this case is gcc/algol68/ChangeLo

[COMMITTED] Add myself to DCO section

For contributions made under Signed-off-by: Jose E. Marchesi ChangeLog * MAINTAINERS: Add myself to DCO section. --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index 90fe5e7cd10..8d3e7ec8c9b 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -

Re: [PATCH] cprop_hardreg: Prevent copy propagation from a non-frame, related insn to a frame related insn

Hi Eric, On 14/12/25 10:43 pm, Eric Botcazou wrote: >> Thanks for reviewing the patch! > > You're welcome. > >> I have upstreamed the patch after incorporating your comments. > > OK, but please post the final version of the patch here next time. My apologies. I misunderstood your review commen

[PATCH v2] x86: Add AMD znver6 processor support

Hi, Thank you Hongtao Liu, Haochen Jiang for the reviews. Addressed all the comments received so far. > You can directly use _mm512_undefined_epi32 and _mm512_setzero_epi32, > and the definition of _mm512_setzero_epi8/_mm512_undefined_epi8 is not > needed. Removed them and used the existing ones

Re: [PATCH] cleanupcfg: Reject forwarder blocks where dest is eh landing pad or a non-local dest [PR123110]

On 12/13/25 10:08 PM, Andrew Pinski wrote: r16-5250-g88db06d7da393f901 changed which block was doing the check on the label being a non-local dest or an eh landing pad. Which is fine except then when we move labels we moved them to before the label for the non-local dest/eh landing pad. This c

Re: [PATCH v5 2/2] asf: Enable pass at O2 or higher on AArch64

On 12/9/25 10:01 AM, Konstantinos Eleftheriou wrote: From: kelefth This patch enables the avoid-store-forwarding patch by default at O2 or higher on AArch64. The assembly patterns in `bitfield-bitint-abi-align16.c` and `bitfield-bitint-abi-align8.c` have been updated to account for the asf

Re: [PATCH v5 1/2] expmed: Copy back the destination register in store_bit_field_1, when needed

On 12/9/25 10:01 AM, Konstantinos Eleftheriou wrote: The call to `gen_lowpart` in `store_bit_field_1` might copy the destination register into a new one, which may lead to wrong code generation, as the bit insertions update the new register instead of updating `str_rtx`. This patch copies bac

Re: [PATCH] RISC-V: Rename UPPERCAE_NAME to UPPERCASE_NAME

On 12/14/25 12:52 AM, Jerry Zhang Jian wrote: This is a follow-up to d99af4e12bf8. gcc/ChangeLog: * config/riscv/gen-riscv-ext-opt.cc: Rename UPPERCAE_NAME to UPPERCASE_NAME. * config/riscv/gen-riscv-ext-texi.cc: Likewise. * config/riscv/riscv-ext-corev.def: L

Re: [PATCH] Remove score files in libgcc

On 12/14/25 11:21 AM, Peter Damianov wrote: This target was deleted in r5-3909-g3daa7bbf791203, but a few files in libgcc were missed. Delete them. gcc/ChangeLog: * score/crti.S: Delete. * score/crtn.S: Delete. * score/sfp-machine.h: Delete. Signed-off-by: Peter Dami

[PATCH 4/5] a68: minor readability improvement

Signed-off-by: Mohammad-Reza Nabipoor gcc/ChangeLog * algol68/a68-exports.cc (a68_asm_output_extract): Re-order cases in switch statement to be sorted based on the GA68_EXTRACT_* values. --- gcc/algol68/a68-exports.cc | 10 +- 1 file changed, 5 insertions(+), 5 d

[PATCH 5/5] a68: add import from .m68 file

Signed-off-by: Mohammad-Reza Nabipoor gcc/ChangeLog * algol68/a68-imports.cc (a68_find_export_data): Implement reading from module's .m68 file if available. --- gcc/algol68/a68-imports.cc | 50 ++ 1 file changed, 45 insertions(+), 5 deletions(

[PATCH 1/5] a68: fix comment and magic number in ga68-exports.pk

Signed-off-by: Mohammad-Reza Nabipoor gcc/ChangeLog * algol68/ga68-exports.pk: Fix comment and value for magic number. --- gcc/algol68/ga68-exports.pk | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/gcc/algol68/ga68-exports.pk b/gcc/algol68/ga68-exports.p

[PATCH 3/5] a68: fix dump of encoded string mode

Signed-off-by: Mohammad-Reza Nabipoor gcc/ChangeLog * algol68/a68-import.cc (dump_encoded_mode): Replace "basic" with "string". --- gcc/algol68/a68-imports.cc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/gcc/algol68/a68-imports.cc b/gcc/algol68/a68-imports

[PATCH 2/5] a68: add Poke description of module exports for 32-bit systems

This commit adds new types for 32-bit systems and rename current types to have a `_64' suffix. Signed-off-by: Mohammad-Reza Nabipoor gcc/ChangeLog * algol68/ga68-exports.pk (ga68_text_reloc_64): Renamed and adapted from ga68_text_reloc. (ga68_data_reloc_64): Renamed and

RE: [r16-6032 Regression] FAIL: gfortran.dg/vect/pr86421.f90 -O (test for excess errors) on Linux/x86_64

Well it seems to me something crashed in my script not rendering anything. But it seems fixed by the following commit. Thx, Haochen From: Jiang, Haochen Sent: Friday, December 12, 2025 6:27 PM To: [email protected]; [email protected]; [email protected]; Jiang, Haochen

Re: [PATCH] aarch64: Cache the PCS value for a function

On Sun, Dec 14, 2025 at 11:41 AM Alfie Richards wrote: > > On 14/12/2025 15:36, Iain Sandoe wrote: > > Hi Alfie, > > > > (just a head’s up) > > > > This change breaks bootstrap on my darwin development branch. > > > > The issue is in aarch64_output_mi_thunk () > > and it is that > > - the thunk

[PATCH] PR122936 Read for user defined data types in Fortran fails ...

Hi all, See attached patch and change log below. Took we a while to sort this one out. Ultimately, my fix for PR105361 was wrong and we were calling hit_eof in the wrong place for the wrong reasons. This is also a regression because the new testcase passes on 14 and fails on 15. Regression

[PATCH] [Bug gcov-profile/123019] Fix Virtual SSA ICE

Hi, This patch shows up with AutoFDO but IMO this is not limited to AutoFDO. The bug is a stale Virtual SSA on calls to functions that have been marked const or pure. pure_const pass analyzes function rocksdb::y::y() and determines it has no side effects and marks it as const. At this point, ex

Re: [PATCH] final_cleanup: Don't create forwarder blocks for degenerate_phis that are ifconvertable [PR123111]

On Sun, Dec 14, 2025 at 10:20 AM Richard Biener wrote: > > > > > Am 14.12.2025 um 17:30 schrieb Jeff Law : > > > >  > > > >> On 12/14/25 2:04 AM, Andrew Pinski wrote: > >> So it turns out creating a forwarder block in some cases causes a > >> regression. > >> The rtl ifcvt does like the case wer

Re: [PATCH] middle-end: Fix spurious -Walloc-size-larger-than warning during LTO [PR106409]

On Sat, Dec 13, 2025 at 11:27:34AM +0100, Richard Biener wrote: > IMO the fix is to simply drop the lang_GNU_CXX () check. > DECL_IS_OPERATOR_NEW_P is good enough. > > OK with that change. > > Richard. Thanks! Pushed the below patch. -Lewis The implementation of -Walloc-size-larger-than has log

Re: [PATCH] aarch64: Cache the PCS value for a function

On Sun, Dec 14, 2025 at 11:42 AM Alfie Richards wrote: > > On 14/12/2025 15:36, Iain Sandoe wrote: > > Hi Alfie, > > > > (just a head’s up) > > > > This change breaks bootstrap on my darwin development branch. > > > > The issue is in aarch64_output_mi_thunk () > > and it is that > > - the thunk

Re: [PATCH] aarch64: Cache the PCS value for a function

On 14/12/2025 15:36, Iain Sandoe wrote: Hi Alfie, (just a head’s up) This change breaks bootstrap on my darwin development branch. The issue is in aarch64_output_mi_thunk () and it is that - the thunk is referring to a virtual dtor - the compilation of the virtual dtor is “complete” and

Re: [PATCH v2] c, plugin: Add c_parse_component_ref callback

Sorry for the late reply, I was out for a while... That series looks really interesting and would probably work for a parser::build_component_ref channel. What is its current state? I would gladly rebase my patch onto the pub-sub system, should it get merged. Also, this patch is still pending rev

[PATCH] Remove score files in libgcc

This target was deleted in r5-3909-g3daa7bbf791203, but a few files in libgcc were missed. Delete them. gcc/ChangeLog: * score/crti.S: Delete. * score/crtn.S: Delete. * score/sfp-machine.h: Delete. Signed-off-by: Peter Damianov --- libgcc/config/score/crti.S| 13

Re: [PATCH] final_cleanup: Don't create forwarder blocks for degenerate_phis that are ifconvertable [PR123111]

> Am 14.12.2025 um 17:30 schrieb Jeff Law : > >  > >> On 12/14/25 2:04 AM, Andrew Pinski wrote: >> So it turns out creating a forwarder block in some cases causes a regression. >> The rtl ifcvt does like the case were the middle bb does not have a single >> predecessor. >> So in some cases i

Re: [PATCH] cprop_hardreg: Prevent copy propagation from a non-frame, related insn to a frame related insn

> I'm sure we can find some old release where things did not yet fail > (simply because ancient stuff missed many features completely), so this > is a regression looked at that way ;-) > > Bugfixes are often backported. But features not usually no. No disagreement, but the policy is nevertheless

Re: [PATCH] cprop_hardreg: Prevent copy propagation from a non-frame, related insn to a frame related insn

Hi! On Sun, Dec 14, 2025 at 06:13:55PM +0100, Eric Botcazou wrote: > > Thanks for reviewing the patch! > > You're welcome. > > > I have upstreamed the patch after incorporating your comments. > > OK, but please post the final version of the patch here next time. > > > After a week on the trunk

Re: [PATCH] cprop_hardreg: Prevent copy propagation from a non-frame, related insn to a frame related insn

> Thanks for reviewing the patch! You're welcome. > I have upstreamed the patch after incorporating your comments. OK, but please post the final version of the patch here next time. > After a week on the trunk, I would like to backport the patch to gcc 13 > as this is a customer reported issue.

Re: [PATCH] final_cleanup: Don't create forwarder blocks for degenerate_phis that are ifconvertable [PR123111]

On 12/14/25 2:04 AM, Andrew Pinski wrote: So it turns out creating a forwarder block in some cases causes a regression. The rtl ifcvt does like the case were the middle bb does not have a single predecessor. So in some cases ifcvt does not happen any more. So the way to fix this is not to cre

Re: [PATCH] aarch64: Cache the PCS value for a function

Hi Alfie, (just a head’s up) This change breaks bootstrap on my darwin development branch. The issue is in aarch64_output_mi_thunk () and it is that - the thunk is referring to a virtual dtor - the compilation of the virtual dtor is “complete” and function.cc :free_afte

Re: [PATCH] cprop_hardreg: Prevent copy propagation from a non-frame, related insn to a frame related insn

Hi Eric, Thanks for reviewing the patch! I have upstreamed the patch after incorporating your comments. After a week on the trunk, I would like to backport the patch to gcc 13 as this is a customer reported issue. Hope this is fine. Regards, Surya On 18/11/25 8:26 pm, Eric Botcazou wrote: >> In

Re: [PATCH] forwardprop: Reject vector comparisons that don't exist [PR123107]

> Am 14.12.2025 um 13:39 schrieb Andrew Pinski : > > Right now forwprop calls into fold to do some folding of comparisons. > But fold is designed for generic which does not have a limitation on > vector comparisons in many cases. So fold turns `a >> signbit != 0` > into `a < 0` but `a < 0` doe

[PATCH 2/2] xtensa: Improve usage of SALT/SALTU instructions

In the expansion of cstoresi4 insn patterns, LT[U] comparisons where the second operand is an integer constant are canonicalized to LE[U] ones with one less than the original. /* example */ int test0(int a) { return a < 100; } unsigned int test1(unsigned int a) { retur

[PATCH 1/2] xtensa: Remove variables only used to pass the return of end_sequence()

As a result of the automatic replacement by commit 4dd13988c93c24ba3605f4b9cafc97515c34f2ac, there are several code fragments that receive the return value of end_sequence() and immediately use it as the return value of the function itself. rtx_insn *insn; ... insn = end_sequence (); ret

[PATCH] forwardprop: Reject vector comparisons that don't exist [PR123107]

Right now forwprop calls into fold to do some folding of comparisons. But fold is designed for generic which does not have a limitation on vector comparisons in many cases. So fold turns `a >> signbit != 0` into `a < 0` but `a < 0` does not exist for the vector type (with a resulting bool). Really

[PATCH] final_cleanup: Don't create forwarder blocks for degenerate_phis that are ifconvertable [PR123111]

So it turns out creating a forwarder block in some cases causes a regression. The rtl ifcvt does like the case were the middle bb does not have a single predecessor. So in some cases ifcvt does not happen any more. So the way to fix this is not to create a forwarder block for those edges which cau