https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118100
Jeffrey A. Law <law at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Known to fail| |14.1.0
CC| |rdapp at gcc dot gnu.org
Blocks|120763 |
Resolution|--- |FIXED
Known to work| |15.1.0
Status|WAITING |RESOLVED
--- Comment #5 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Not sure how this escaped from both Robin and myself, but it did.
The good news is this was reported against "15.0.0 20241117" so not an actual
release. I checked gcc-15.1.0 and it's fine along with the trunk.
Bisection points to this as the fix:
commit 9b4cc2710f97138805a1db6a89bc792e22f02db2 (HEAD)
Author: Pan Li <[email protected]>
Date: Fri Dec 13 10:45:38 2024 +0800
RISC-V: Make vector strided load alias all other memories
The vector strided load doesn't include the (mem:BLK (scratch)) to
alias all other memories. It will make the alias analysis only
consider the base address of strided load and promopt the store
before the strided load. For example as below
Which fits. Given we're not backporting to gcc-14, I'll add known-to-work/fail
and close.
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120763
[Bug 120763] [meta-bug] Tracker for bugs to visit during weekly RISC-V meeting
