Bug 27626 - the "fix" for bug 25622 actually broke things
Summary: the "fix" for bug 25622 actually broke things
Status: UNCONFIRMED
Alias: None
Product: binutils
Classification: Unclassified
Component: gas (show other bugs)
Version: 2.36
: P2 normal
Target Milestone: ---
Assignee: Not yet assigned to anyone
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2021-03-23 07:08 UTC by Jan Beulich
Modified: 2022-06-22 06:31 UTC (History)
0 users

See Also:
Host:
Target: x86_64-*-*
Build:
Last reconfirmed:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Jan Beulich 2021-03-23 07:08:04 UTC
Working around a compiler issue by suppressing legitimate assembler warnings is not a way to go. In fact the suppressed warnings hide a code generation bug in gcc (up to 7 afaict) - in such a case, it seems more than just desirable to me to point out to the programmer that the compiler produced bad code. The issue there was a lack of suffix in "cvtusi2<ssescalarmodesuffix>32<round_name>" and "cvtusi2<ssescalarmodesuffix>64<round_name>", which is leading to a 32-bit source operand getting used silently in the 64-bit case.

The correct way to approach such problems is to backport the respective fixes to the affected older compiler branches. The gas change (commit 10d97a0f9907) should be reverted.