[Webkit-unassigned] [Bug 142333] [GTK] WebCore::TransformationMatrix::multiply segfaults when loading last.fm

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Fri Apr 3 08:37:36 PDT 2015


https://bugs.webkit.org/show_bug.cgi?id=142333

Michael Catanzaro <mcatanzaro at igalia.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |benjamin at webkit.org

--- Comment #16 from Michael Catanzaro <mcatanzaro at igalia.com> ---
Well, I have no clue what's wrong, or whether it's our code or GCC that is broken, but this is by far our biggest problem right now ("I think this crash happens 100% of the time visiting youtube.com"), so at least let's commit a workaround ASAP. Any objections to conditionally compiling the x86_64 matrix multiplication fastpath only when using GCC < 5.0? It will obviously make things slower, but that beats a crash.

Hi Ben, CCing you because this is your code (from r138866), so fates permitting, maybe you have some idea what might be wrong. None of us understand it. :)

The best backtrace we have is https://bugzilla.redhat.com/attachment.cgi?id=999725 which isn't great but does show the assembler where the crash occurs.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20150403/794e4ed7/attachment.html>


More information about the webkit-unassigned mailing list