Closed Bug 1307226 Opened 8 years ago Closed 8 years ago

clang 3.8 crashes when compiling changes from bug 1151204

Categories

(Core :: Layout, defect)

defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla52
Tracking Status
firefox52 --- fixed

People

(Reporter: keeler, Assigned: keeler)

References

Details

Attachments

(1 file)

On Fedora with clang-3.8, clang crashes when compiling with the changes from bb3080bc1aad from bug 1151204. The issue seems to be similar to one that's already worked around in that code (see https://hg.mozilla.org/mozilla-central/rev/bb3080bc1aad#l1.28 and https://hg.mozilla.org/mozilla-central/rev/bb3080bc1aad#l1.183 ). If I extend the CLANG_CRASH_BUG macro to cover 3.8 and wrap the assertion at https://hg.mozilla.org/mozilla-central/rev/bb3080bc1aad#l1.146 with the same #ifndef CLANG_CRASH_BUG, compilation succeeds.
That's odd, I'm using 3.8 myself and it compiles fine for me: clang version 3.8.0-2ubuntu4 (tags/RELEASE_380/final) Target: x86_64-pc-linux-gnu What's the output from your "clang --version" ?
Flags: needinfo?(dkeeler)
Interesting - I was wondering why it seemed others hadn't encountered this and filed a bug before me... In any case: clang version 3.8.0 (tags/RELEASE_380/final) Target: x86_64-unknown-linux-gnu Thread model: posix InstalledDir: /usr/bin
Flags: needinfo?(dkeeler)
Weird, maybe the clang bug is still present then, but depends on some other unknown factor which are no longer present on my system. It sounds like you wrote a patch already, can you attach it and ask for review please? BTW, I see that we use g++ on mozilla-inbound so the assertions should still be active there.
Flags: needinfo?(dkeeler)
Ok - there's the patch. I'll keep investigating, though - I'm still confused as to why it seems only I have encountered this issue.
Assignee: nobody → dkeeler
Flags: needinfo?(dkeeler)
Comment on attachment 8797283 [details] bug 1307226 - extend clang crash workaround to work around similar crash in SetGridItemCount https://reviewboard.mozilla.org/r/82884/#review81524 Thanks! r=mats with this nit: Could we make the last factor there "__clang_minor__ <= 8" instead? I see no reason why 3.7 isn't affected in the light of 3.8 crashing (for some). That will of course disable the assertions on all 3.x up to and including 3.8, but I think that's fine.
Attachment #8797283 - Flags: review?(mats) → review+
Pushed by dkeeler@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/1a4288cc3cea extend clang crash workaround to work around similar crash in SetGridItemCount r=mats
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla52
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: