Q about crash

8 views
Skip to first unread message

Avi Drissman

unread,
May 18, 2022, 11:48:16 AM5/18/22
to Clang maintainers
A new high-frequency crash just appeared deep inside gemmlowp code, which is deeply templated and not super understandable. A bisect seems to implicate my roll of the minimum required macOS version, which is odd, and we're not sure if that's a bad bisect or if it's real and something much more subtle is going on.

I email you folks as you're the compiler people. Do you have any thoughts as to what's going on here? Assuming this is indeed the macOS repin, and seeing that this is x64-only, and seeing that this code has nothing to do with macOS, something inside me says that this might be UB that is exposed to be crashing based on something tied to the minimum macOS version. That's honestly on the borderline of absurd and exactly what a compiler would do, but I don't have anything better.

Ideas? Thanks.

Avi

Bruce Dawson

unread,
May 18, 2022, 11:57:01 AM5/18/22
to Avi Drissman, Clang maintainers
Bug link?

--
To unsubscribe from this group and stop receiving emails from it, send an email to clang+un...@chromium.org.


--
Bruce Dawson, he/him

Avi Drissman

unread,
May 18, 2022, 12:10:15 PM5/18/22
to Bruce Dawson, Clang maintainers
crbug.com/1326455

A slight update, though. It turns out that gemmlowp is unsupported obsolete code and its author asked us to move to a different library.
Reply all
Reply to author
Forward
0 new messages