[Bug target/120651] [15/16 Regression] RISC-V: Miscompile at -O3 with -flto since r15-3228-g771256bcb9d

2025-07-02 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120651 Alexey Merzlyakov changed: What|Removed |Added CC||alexey.merzlyakov at samsung dot c

[Bug target/120356] [15/16 Regression] RISC-V: Miscompile at -O[23] since r15-6881-g7b815107f40

2025-06-30 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120356 --- Comment #5 from Alexey Merzlyakov --- Analysis update: vector calculation chain in the loop from the above comment remains from the initial expand pass, while other duplicated code was moved to the first function BB by loop2_invariant. Afte

[Bug target/120714] RISC-V: incorrect frame pointer CFA address for stack-clash protection loops

2025-06-29 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120714 --- Comment #3 from Alexey Merzlyakov --- I've checked the patch from the above suggested link. Unfortunately, it does not work for the reported in this ticket test-case: GDB is missing to get into goo() breakpoint. I also tested it on the test

[Bug target/120356] [15/16 Regression] RISC-V: Miscompile at -O[23] since r15-6881-g7b815107f40

2025-06-25 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120356 Alexey Merzlyakov changed: What|Removed |Added CC||alexey.merzlyakov at samsung dot c

[Bug middle-end/120714] RISC-V: incorrect frame pointer CFA address for stack-clash protection loops

2025-06-19 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120714 --- Comment #1 from Alexey Merzlyakov --- Initial bugfix has been prepared locally and checked on simple test-cases. Once it will pass internal tests for RISC-V, the patch to be submitted to the mailing list.

[Bug middle-end/120714] New: RISC-V: incorrect frame pointer CFA address for stack-clash protection loops

2025-06-19 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
Severity: normal Priority: P3 Component: middle-end Assignee: unassigned at gcc dot gnu.org Reporter: alexey.merzlyakov at samsung dot com Target Milestone: --- Created attachment 61666 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=61666&action=edi

[Bug middle-end/108016] RISC-V:Bad codegen in scalar code comparing to LLVM

2025-05-07 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108016 --- Comment #13 from Alexey Merzlyakov --- > There is https://gcc.gnu.org/pipermail/gcc-patches/2024-August/660968.html > which (though I am not 100% sure) will help the situtation. Thank you for the reference, this definitely makes sense. If

[Bug middle-end/108016] RISC-V:Bad codegen in scalar code comparing to LLVM

2025-04-30 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108016 --- Comment #11 from Alexey Merzlyakov --- > So I'd first look at why we created an explicit store through memory rather > than simple x.y assignment like we see for D.19566.first. For the examples from "Item1" and "Item2", GCC does not gener

[Bug middle-end/108016] RISC-V:Bad codegen in scalar code comparing to LLVM

2025-04-02 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108016 --- Comment #7 from Alexey Merzlyakov --- > For DSE to kick in, I'm pretty sure we'd need to eliminate the memory load > first. Eliminating the memory load will likely be nontrivial. For this, I think we could start from loads that was partiall

[Bug target/119519] [14 Regression] RISC-V, autovectorization, internal compiler error when "V" RISC-V extension used.

2025-03-30 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119519 Alexey Merzlyakov changed: What|Removed |Added CC||alexey.merzlyakov at samsung dot c

[Bug rtl-optimization/119099] [15 regression] Compile-time hang in ext-dce

2025-03-06 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119099 --- Comment #8 from Alexey Merzlyakov --- Oops... It looks like we've submitted the patch in maillist in parallel two times.

[Bug rtl-optimization/119099] [15 regression] Compile-time hang in ext-dce

2025-03-06 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119099 --- Comment #7 from Alexey Merzlyakov --- > That should result in the livein sets never contracting. That API will return > a boolean indicating if that livein set actually changed. Now I just have to > see if there's any fallout. Thanks for n

[Bug rtl-optimization/119099] New: Compile-time hang in DCE

2025-03-03 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
Assignee: unassigned at gcc dot gnu.org Reporter: alexey.merzlyakov at samsung dot com Target Milestone: --- Created attachment 60644 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=60644&action=edit Reduced test-case The following code c-reduced from csmith ge

[Bug middle-end/108016] RISC-V:Bad codegen in scalar code comparing to LLVM

2025-02-07 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108016 --- Comment #5 from Alexey Merzlyakov --- For "Item3" (extra sign extension): it seem that a slightly more elegant solution was found - generate sign_extend(plus) + subreg chain during expand-rtx. Currently expanding code snippet for umulv4:

[Bug middle-end/108016] RISC-V:Bad codegen in scalar code comparing to LLVM

2025-02-05 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108016 Alexey Merzlyakov changed: What|Removed |Added CC||alexey.merzlyakov at samsung dot c

[Bug target/117688] [15 Regression] RISC-V: Wrong code for .SAT_SUB

2025-01-17 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117688 Alexey Merzlyakov changed: What|Removed |Added CC||alexey.merzlyakov at samsung dot c

[Bug rtl-optimization/117476] [15 regression] bad generated code at -O1 since r15-4991-g69bd93c167fefb

2024-11-11 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117476 --- Comment #25 from Alexey Merzlyakov --- Updates on GCC regression testing: Locally checked the GCC make check with enable-languages=all before the patching vs. r15-4991-g69bd93c167fefb + fix we are discussing in this ticket. Testing was perf

[Bug rtl-optimization/117476] [15 regression] bad generated code at -O1 since r15-4991-g69bd93c167fefb

2024-11-08 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117476 --- Comment #16 from Alexey Merzlyakov --- Current situation is: Patch was tested locally, and it seem to eliminate the problem in incorrect subreg mode check. The following regressions were fixed: * Reported here and in pr117480 testcases on

[Bug rtl-optimization/117476] [15 regression] bad generated code at -O1 since r15-4991-g69bd93c167fefb

2024-11-07 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117476 --- Comment #11 from Alexey Merzlyakov --- Analysis shown that the problem seem is related to incorrect check in the optimization, introduced by the aforementioned patch: (zero_extend:M (subreg:N (not:M (X:M -> (xor:M (X:M, mask)) In

[Bug rtl-optimization/117476] [15 regression] bad generated code from -finline-small-functions since r15-4991-g69bd93c167fefb

2024-11-07 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117476 --- Comment #7 from Alexey Merzlyakov --- Thank you for the issue simplification the repro-case. Confirming: the problem exists on trunk since r15-4991-g69bd93c167fefb. Working on it.

[Bug rtl-optimization/112398] Suboptimal code generation for xor pattern on subreg

2024-10-03 Thread alexey.merzlyakov at samsung dot com via Gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112398 Alexey Merzlyakov changed: What|Removed |Added CC||alexey.merzlyakov at samsung dot c

[Bug libstdc++/60758] Infinite backtrace in __cxa_end_cleanup

2014-05-19 Thread alexey.merzlyakov at samsung dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60758 --- Comment #9 from Alexey Merzlyakov --- The following PR has been opened for Thumb1 problem: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61223

[Bug libstdc++/60758] Infinite backtrace in __cxa_end_cleanup

2014-05-19 Thread alexey.merzlyakov at samsung dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60758 --- Comment #8 from Alexey Merzlyakov --- Created attachment 32820 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=32820&action=edit Fix for thumb fail Proposed bugfix (build OK, but not regtested).

[Bug libstdc++/60758] Infinite backtrace in __cxa_end_cleanup

2014-05-19 Thread alexey.merzlyakov at samsung dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60758 --- Comment #7 from Alexey Merzlyakov --- The problem does not appear for thumb2 targets. On older architectures (armv6 and below) in thumb-mode the LR indeed can not be used as argument of POP instruction. To support __cxa_end_cleanup backtraci

[Bug libstdc++/60758] Infinite backtrace in __cxa_end_cleanup

2014-05-19 Thread alexey.merzlyakov at samsung dot com
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60758 --- Comment #6 from Alexey Merzlyakov --- The issue was reproduced at my side. Let me make necessary investigations to fix the problem as soon as possible.

[Bug libstdc++/60758] Infinite backtrace in __cxa_end_cleanup

2014-04-06 Thread alexey.merzlyakov at samsung dot com
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60758 --- Comment #3 from Alexey Merzlyakov --- Thank you very much! Reg. test - no changes. http://gcc.gnu.org/ml/gcc-patches/2014-04/msg00195.html

[Bug libstdc++/60758] Infinite backtrace in __cxa_end_cleanup

2014-04-04 Thread alexey.merzlyakov at samsung dot com
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60758 --- Comment #1 from Alexey Merzlyakov --- Created attachment 32543 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32543&action=edit Proposed patch Proposed patch is attached.

[Bug libstdc++/60758] New: Infinite backtrace in __cxa_end_cleanup

2014-04-04 Thread alexey.merzlyakov at samsung dot com
++ Assignee: unassigned at gcc dot gnu.org Reporter: alexey.merzlyakov at samsung dot com CC: v.garbuzov at samsung dot com, y.gribov at samsung dot com Target: arm Created attachment 32542 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32