[Bug rtl-optimization/125176] [16/17 regression] A possible regression in GCC 16 breaking qt5-qtwebkit rendering of simple HTML

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125176

Sam James <sjames at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|A possible regression in    |[16/17 regression] A
                   |GCC 16 breaking             |possible regression in GCC
                   |qt5-qtwebkit rendering of   |16 breaking qt5-qtwebkit
                   |simple HTML                 |rendering of simple HTML
           Keywords|                            |needs-source, wrong-code
           See Also|                            |https://bugs.gentoo.org/sho
                   |                            |w_bug.cgi?id=969496,
                   |                            |https://bugs.gentoo.org/sho
                   |                            |w_bug.cgi?id=968755
   Target Milestone|---                         |16.2

--- Comment #1 from Sam James <sjames at gcc dot gnu.org> ---
We had a few reports during the GCC 16 cycle downstream:
* qtwebengine (with Qt 6): https://bugs.gentoo.org/968755 where it is
apparently miscompiled with -O3
* webkit-gtk: https://bugs.gentoo.org/969496 where it started working again a
few weeks ago

I couldn't ever repro the webkit-gtk one, and qtwebengine is so massive that
it's not easy to work on. The build time to get started is so huge that it's
impractical for me.