On Tue, Feb 11, 2025 at 4:38 PM Hongtao Liu <crazy...@gmail.com> wrote: > > On Tue, Feb 11, 2025 at 4:27 PM H.J. Lu <hjl.to...@gmail.com> wrote: > > > > On Tue, Feb 11, 2025 at 4:13 PM Hongtao Liu <crazy...@gmail.com> wrote: > > > > > > > PR117081 is about regression in povray. The reducted testcase: > > > Just for clarification. PR117081 is not about regression in povray. > > > it's related to FAIL: gcc.target/i386/pr91384.c scan-assembler-not > > > testl > > > The pr91384.c is added by r12-7417 which is peephole optimization > > > expecting some specific instruction sequence, the regression can be > > > fixed by adding new peephole pattern. > > > > > > H.J patch actually regressed povray by introducing extra push/pops > > > (since it adds preference for callee save registers, in the benchmark > > > using caller saved registers is much better). > > > Sorry, I may not have been clear in > > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117081#c9 > > > > > > > My patch doesn't change the codegen for that code as shown by > Real benchmark scenarios are a little more complex, the testcase in > the PR is just one of the scenes, but not all. > We are currently investigating this case and hope to find a better solution.
We need testcases to make sure that there are no regressions. > > > > commit 846837c2406ae7a52d9123b29c13e4b8b9d14224 > > Author: H.J. Lu <hjl.to...@gmail.com> > > Date: Fri Feb 7 13:49:30 2025 +0800 > > > > x86: Verify that PUSH/POP can be skipped > > > > > > -- > > H.J. > > > > -- > BR, > Hongtao -- H.J.
