[Webkit-unassigned] [Bug 245321] Ran clang-tidy on JSC, WTF and bmalloc

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=245321

Nikolas Zimmermann <zimmermann at kde.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |zimmermann at kde.org

--- Comment #3 from Nikolas Zimmermann <zimmermann at kde.org> ---
This caused a regression on ARM 32bit -- SIGILL during WebProcess startup.
I've split this patch into multiple pieces, WebKit main + only the bmalloc/ portion of this patch is enough to have WebProcess constantly abort with SIGILL.

It's hard to produce actual stacktraces, as I have to selectively build certain files with debug symbols, a full WebKit debug or RelWithDebInfo builds leads to large libWPEWebKit.so, exceeding > 2GB, not possible to use as-is on embedded.

Anyhow, I'm 100% certain by bisect that this commit is guilty and we're likely facing a compiler problem, since by code inspection I cannot find anything suspicious in the clang-tidy-patch itself.

However the patch already triggered a debug compilation problem with clang10, where KeyValuePairs "= default" constructor leads to a compilation error, whereas using " { }" instead works fine :/

Looking forward to nail down the culprit within the Source/bmalloc/ changes.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20220930/b3ca965d/attachment.htm>