Re: [CAUTION - External Sender] Re: gcc relies on RISC-V vcompress instruction undefined behaviour

2024-10-31 Thread Anton Blanchard via Gcc
Hi Richard, > Please file a bugreport so this issue doesn't get lost. I was struggling to reproduce it on a recent toolchain, but I managed to take one of the gcc test cases and recreate it I think. I submitted https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117383 Thanks, Anton

Re: gcc relies on RISC-V vcompress instruction undefined behaviour

2024-10-31 Thread Richard Biener via Gcc
On Thu, Oct 31, 2024 at 3:55 AM Anton Blanchard via Gcc wrote: > > Hi, > > I think gcc is relying on undefined behaviour with the vcompress instruction. > Unfortunately my test case isn't reproducing on mainline, but gcc looks to > use the fields between the last mask selected field and vl while s

Re: gcc relies on RISC-V vcompress instruction undefined behaviour

2024-10-31 Thread Robin Dapp via Gcc
> Hi, > > I think gcc is relying on undefined behaviour with the vcompress instruction. > Unfortunately my test case isn't reproducing on mainline, but gcc looks to > use the fields between the last mask selected field and vl while setting > tail agnostic. > > This thread explains how vcompress is