629057 - fix unaligned accesses in qcms on sparc64

Status: RESOLVED FIXED

(Core :: Graphics: Color Management, defect)

Description

[Landry Breuil (:gaston)]

User-Agent:       Midori/0.2 (X11; OpenBSD; U; fr-fr) WebKit/531.2+
Build Identifier: 

Followup to https://bugzilla.mozilla.org/show_bug.cgi?id=504766 (and its duplicates), this bug is still present in branches 1.9.1 and 1.9.2. Browsing a particular page on OpenBSD/sparc64 using ffx 3.6.13 crashes the browser.

Cherrypicking http://hg.mozilla.org/mozilla-central/rev/e8207773d54d fixes the issue.

Reproducible: Always

Comment 1

[Landry Breuil (:gaston)]

Crash has been confirmed on http://www.heise.de/netze/ with ffx 3.5.16.

Comment 2

[Landry Breuil (:gaston)]

Attachment: Fix unaligned access for 1.9.1

backport of http://hg.mozilla.org/mozilla-central/rev/e8207773d54d

Comment 3

[Landry Breuil (:gaston)]

Attachment: Fix unaligned access for 1.9.2

Backport of http://hg.mozilla.org/mozilla-central/rev/e8207773d54d

Comment 5

[j.j.]

You should request approval

Comment 6

[Daniel Veditz [:dveditz]]

Comment on attachment 521016 [details] [diff] [review]
Fix unaligned access for 1.9.1

1.9.1.19 is code-frozen

Comment 7

[Daniel Veditz [:dveditz]]

Comment on attachment 521016 [details] [diff] [review]
Fix unaligned access for 1.9.1

Approved for 1.9.1.20, a=dveditz for release-drivers

Comment 8

[Daniel Veditz [:dveditz]]

Comment on attachment 521017 [details] [diff] [review]
Fix unaligned access for 1.9.2

Approved for 1.9.2.18, a=dveditz for release-drivers

Comment 9

[Daniel Veditz [:dveditz]]

Comment on attachment 521017 [details] [diff] [review]
Fix unaligned access for 1.9.2

Didn't make 3.6.18

Comment 10

[Landry Breuil (:gaston)]

And what's needed to get that commited ? approval-xxx is not enough now ?

Comment 11

[Boris Zbarsky [:bzbarsky]]

> And what's needed to get that commited ?

Someone to do the actual push...  So either poking someone directly or adding checkin-needed.

dveditz, does this need to be reapproved for 3.6.19, or can I just push it?

Comment 12

[Daniel Veditz [:dveditz]]

Comment on attachment 521017 [details] [diff] [review]
Fix unaligned access for 1.9.2

Approved for 1.9.2.19, a=dveditz

Comment 13

[Boris Zbarsky [:bzbarsky]]

And the other thing that would help would be posting a patch that can actually be imported into hg...

Comment 14

[Boris Zbarsky [:bzbarsky]]

http://hg.mozilla.org/releases/mozilla-1.9.2/rev/7976c26c70e8

Comment 15

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Per bz's request, pushed:
http://hg.mozilla.org/releases/mozilla-1.9.1/rev/920bcf17a9e1

Comment 16

[Wolfgang Rosenauer [:wolfiR]]

JFYI (I can fix it locally on my side but I'd like you to know it ;-))

iccread.c: In function 'read_u32':
iccread.c:91: warning: implicit declaration of function 'memcpy'
iccread.c:91: warning: incompatible implicit declaration of built-in function 'memcpy'
iccread.c: In function 'read_u16':
iccread.c:103: warning: incompatible implicit declaration of built-in function 'memcpy'

This breaks the build in our buildsystem:
I: Program is using implicit definitions of special functions.
   these functions need to use their correct prototypes to allow
   the lightweight buffer overflow checking to work.
     - Implicit memory/string functions need #include <string.h>.
     - Implicit *printf functions need #include <stdio.h>.
     - Implicit *printf functions need #include <stdio.h>.
     - Implicit *read* functions need #include <unistd.h>.
     - Implicit *recv* functions need #include <sys/socket.h>.
E: MozillaThunderbird implicit-fortify-decl iccread.c:91, 103

Comment 17

[Landry Breuil (:gaston)]

Right.. so it lacks a #include <string.h>. I suppose a separate bug should be opened against trunk for that ?

Comment 18

[Benoit Girard (:BenWa)]

A new bug would be great. We will fix this soon.

Comment 19

[Landry Breuil (:gaston)]

It's already fixed in trunk by bug 674230 (http://hg.mozilla.org/mozilla-central/rev/77305d830151)... so i suppose a new bug should be filed against 1.9.2 branch ?

Comment 20

[Al Billings [:abillings - ex-MoCo]]

Landry, can you verify that the crash is fixed in 1.9.2 and trunk on sparc64? I don't have access to such machines.

Comment 21

[Landry Breuil (:gaston)]

As i said in Bug 643137, i have limited time for that in the coming days, but i'll try 3.6.20 when possible.

Oh, and noone will never know if the crash is fixed in trunk, since it is definitely broken on sparc64, and will probably stay that way.

Comment 22

[Landry Breuil (:gaston)]

Firefox 3.6.20 Candidate build 1 works fine on sparc64 with the patch integrated, i can browse the offending sites http://www.heise.de/netze/ and http://www.onlinemovies.top-site-list.com/ without crashes. Thanks!

Comment 23

[Al Billings [:abillings - ex-MoCo]]

Thanks. Marking as verified for 1.9.2.