1187664 - [gtk3] 2 valgrind errors

Status: RESOLVED FIXED

(Core :: Widget: Gtk, defect)

Description

[Mike Hommey [:glandium]]

https://treeherder.mozilla.org/logviewer.html#?job_id=9775743&repo=try

Essentially there are 2 errors:
- write() with uninitialized bytes in fontconfig.
- invalid free with a weird backtrace.

Comment 1

[Karl Tomlinson (:karlt)]

The write in FcInit() seems unlikely to be related to application usage of the library.  Seems more likely an issue in the library, and so can be suppressed, assuming that is possible.

Comment 2

[Mike Hommey [:glandium]]

So, with debugging symbols, this is what the FcInit() failures look like:

==9529== Syscall param write(buf) points to uninitialised byte(s)
==9529== at 0x4E4148D: ??? (in /lib64/libpthread-2.12.so)
==9529== by 0xAB1BD43: FcDirCacheWrite (fccache.c:921)
==9529== by 0xAB22CB3: FcDirCacheScan (fcdir.c:289)
==9529== by 0xAB1EB48: FcConfigAddDirList (fccfg.c:355)
==9529== by 0xAB1EDD5: FcConfigBuildFonts (fccfg.c:388)
==9529== by 0xAB27EFA: FcInitLoadConfigAndFonts (fcinit.c:106)
==9529== by 0xAB27FC4: FcInit (fcinit.c:124)
(snip)

They are reproducible by running fc-cache, so they clearly are entirely within fontconfig.

Now, running firefox shouldn't make us write the fontconfig cache ourselves, at least not in most cases, because the cache will already exist somehow. So I'm going to test pre-generating the fontconfig cache, along pango.modules and gdb-pixbuf's loaders.cache.

Comment 3

[Mike Hommey [:glandium]]

Attachment: Create a fontconfig cache so that Firefox doesn't have to do it itself when run on build automation

Comment 4

[Mike Hommey [:glandium]]

Attachment: Don't run gst-plugin-scanner through valgrind, it only leads to irrelevant errors

Comment 5

[Nicholas Nethercote [inactive]]

Comment on attachment 8640353 [details] [diff] [review]
Don't run gst-plugin-scanner through valgrind, it only leads to irrelevant errors

Review of attachment 8640353 [details] [diff] [review]:
-----------------------------------------------------------------

Nice comment.

Comment 6

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/be94d55627f4
https://hg.mozilla.org/integration/mozilla-inbound/rev/dfc1cb1c26d9

Comment 7

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/be94d55627f4
https://hg.mozilla.org/mozilla-central/rev/dfc1cb1c26d9

Comment 8

[Ryan VanderMeulen [:RyanVM]]

Backed out because it depends on bug 1186748, which had to be backed out.
https://hg.mozilla.org/mozilla-central/rev/a2d31912a7a9

Comment 9

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/207c45dd3c2e
https://hg.mozilla.org/integration/mozilla-inbound/rev/695b89efecfa

Comment 10

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/207c45dd3c2e
https://hg.mozilla.org/mozilla-central/rev/695b89efecfa

Comment 11

[Pulsebot]

Backout:
https://hg.mozilla.org/mozilla-central/rev/91ccd37efd3e
https://hg.mozilla.org/mozilla-central/rev/f4ff9f935df5

Comment 12

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/1e2f1521b7e8
https://hg.mozilla.org/integration/mozilla-inbound/rev/b28d496da7bf

Comment 13

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/1e2f1521b7e8
https://hg.mozilla.org/mozilla-central/rev/b28d496da7bf