RE: [PATCH PR95199] vect: Remove extra variable created for memory reference

Hi, I modified the issue you mentioned. Bootstrap and tested on aarch64 Linux platform again. No new regression witnessed. For "*vec_offset", it is indeed a point optimizable. However, it is not able to eliminate it by the same way as "*dataref_bump". The cse_and_gimplify_to_preheader wil

[committed][GCC10/11] testsuite: Re-add gdc.test prefix in test names (PR testsuite/95575)

Hi, When moving the gdc-test.exp test script into multiple test scripts (c76df72f1a9), this subdir handling got removed. This patch re-adds that, adjusted to handle the new $subdir path. Regression tested on x86_64-linux-gnu, committed to mainline, and backported to releases/gcc-10. Regards Iai

[committed] d: Fix line lengths after mechanical field renames.

Hi, This patch adjusts a few overflowed lines after the recent refactorings done in the upstream front-end that caused renames of many fields - in this case, dim -> length. Bootstrapped and regression tested on x86_64-linux-gnu, and committed to mainline. Regards Iain. gcc/d/ChangeLog:

[patch, fortran] PR 27318, warn if interfaces do not match

Hello world, this patch solves an PR which just had its 14th birthday, continuing the mission of alerting the user to mismatches where possible. Regression-tested (which led to a few of the extra checks for errors). OK for trunk? Regards Thomas Test global identifiers against what is

[patch, fortran, committed] Fix PR 95037, NULL location in error message

Hi, I just committed as obvious and simple the attached patch, as r11-1297-g4644e8f15f835a9934a8d289ee08ba4cb46cbfac . Regards Thomas Always use locations from get and put arguments for error messages. A simple and obvios patch - the error location was taken from a variable that was n

Ping [PATCH] PR fortran/95088 - Buffer overflows with PDTs, submodules and long symbols

> Here's a fix for another buffer overflow which is triggered by one of > Gerhard's (in)famous testsuite. Looking at the way symbols are generated, > I decided to partly use a simple solution recommended by Jakub Jelinek. > > Regtested with no new failures on x86_64-pc-linux-gnu. > > OK for master

Re: Ping [PATCH] PR fortran/95088 - Buffer overflows with PDTs, submodules and long symbols

Hi Harald, Here's a fix for another buffer overflow which is triggered by one of Gerhard's (in)famous testsuite. Looking at the way symbols are generated, I decided to partly use a simple solution recommended by Jakub Jelinek. Regtested with no new failures on x86_64-pc-linux-gnu. OK for mast

[patch, Fortran, committed] Fix a crash with a developer's option

Committed as obvious, simple and risk-free after checking that it still compiles. It does not fix the PR in question, it it just something that was noted in one of the comments. Regards Thomas Avoid crash when global symbol table is empty with -fdump-fortran-global. This just avoids a

Re: [PATCH] Treat { 0 } specially for structs with the designated_init attribute.

Hello, Asher Gordon writes: > I also added a note after the warning showing where the field was > defined in the structure, like this: > > inform (DECL_SOURCE_LOCATION (info.field), > "in definition of %qT", constructor_type); > > However, I think it would be preferable to point to

Re: [PATCH] avoid false positives due to compute_objsize (PR 95353)

On Sat, 2020-06-13 at 17:49 -0600, Martin Sebor wrote: > On 6/13/20 3:50 PM, Sandra Loosemore wrote: > > On 6/2/20 6:12 PM, Martin Sebor via Gcc-patches wrote: > > > The compute_objsize() function started out as a thin wrapper around > > > compute_builtin_object_size(), but over time developed its

[PATCH] gfortran: Improve translation of POPPAR intrinsic

The following patch to gfortran's trans-instrinsic.c tweaks the generic that is produced for popcnt on integer(kind=16). Currently, the double word popcnt is implemented as parityll(hipart(x))^parityll(lopart(x)), but with this patch this is now translated as parityll(hipart(x)^lopart(x)).

[PATCH] libstdc++: Fix char_traits move with overlap

Hi, Upon constexpr evaluation, char_traits move uses copy_backward, but its last argument should be to the range end rather than its beginning. I include the fix and a test. This is my first patch, so if it looks OK, perhaps someone could commit for me. Regards, Paul Please consider the en

Ping: [PATCH V2] Add pattern for pointer-diff on addresses with same base/offset (PR 94234)

Thanks, Feng From: Feng Xue OS Sent: Friday, June 5, 2020 5:20 PM To: Richard Biener; gcc-patches@gcc.gnu.org; Marc Glisse Subject: Re: [PATCH] Add pattern for pointer-diff on addresses with same base/offset (PR 94234) As Marc suggested, removed the new

[committed] RISC-V: Suppress warning for signed and unsigned integer comparison.

gcc/ChangeLog: * config/riscv/riscv.c (riscv_gen_gpr_save_insn): Change type to unsigned for i. (riscv_gpr_save_operation_p): Change type to unsigned for i and len. --- gcc/config/riscv/riscv.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git

Re: [PATCH 1/2] RISC-V: Describe correct USEs for gpr_save pattern [PR95252]

Hi Andreas: Thanks for your reminder, fixed on trunk: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=82a3008e56c620008b4575a97e459e2769df54db On Sat, Jun 13, 2020 at 4:58 PM Andreas Schwab wrote: > > On Jun 10 2020, Kito Cheng wrote: > > > - Verified on rv32emc/rv32gc/rv64gc bare-metal target

RE: [PATCH PR95199] vect: Remove extra variable created for memory reference

On Sun, 14 Jun 2020, zhoukaipeng (A) wrote: > Hi, > > I modified the issue you mentioned. Bootstrap and tested on aarch64 Linux > platform again. No new regression witnessed. > > For "*vec_offset", it is indeed a point optimizable. However, it is not > able to eliminate it by the same way