On 7/13/2026 12:31 PM, [email protected] wrote:
From: MITSUNARI Shigeo <[email protected]> Thanks for the report. The check-function-bodies pattern in this test expects the instruction order produced with generic tuning. With -march=cascadelake the scheduler reorders the movl and movabsq instructions, so the (order-sensitive) body match fails; the generated instructions themselves are unchanged. The same applies to gcc.target/i386/mulq-highpart.c from r17-2334, so the patch below fixes both tests by pinning -mtune=generic, as many other i386 tests do. Tested on x86_64-linux with both the default target board and unix{-m64 -march=cascadelake}: both tests pass with the patch and the reported FAIL is reproducible without it. I don't have write access, so could someone push this for me? -- >8 -- Subject: i386: Add -mtune=generic to highpart multiplication tests The check-function-bodies patterns in bmi2-mulx-highpart-1.c and mulq-highpart.c expect the instruction order produced with generic tuning. With -march=cascadelake the scheduler swaps the movl and movabsq instructions, so pin -mtune=generic. gcc/testsuite/ChangeLog: * gcc.target/i386/bmi2-mulx-highpart-1.c: Add -mtune=generic to dg-options. * gcc.target/i386/mulq-highpart.c: Likewise. Signed-off-by: MITSUNARI Shigeo <[email protected]>
THanks. I've pushed this to the trunk. jeff
