Closed Bug 1206673 Opened 9 years ago Closed 9 years ago

Nightly crashes shortly after start (Sept. 20th)

Categories

(Firefox :: Untriaged, defect)

43 Branch
defect
Not set
critical

Tracking

()

RESOLVED DUPLICATE of bug 1206524

People

(Reporter: bytehead, Unassigned)

Details

(Keywords: crash, crashreportid)

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:40.0) Gecko/20100101 Firefox/40.0
Build ID: 20150826023504

Steps to reproduce:

Start Firefox


Actual results:

Page may or may not display before crash occurs, with various crash reports.


Expected results:

Firefox should have been operating correctly.

I have tried 32-bit and 64-bit (hence why you will see different architectures) with a fresh profile and an old profile.  e10s is off.

Some of the crash reports:
https://crash-stats.mozilla.com/report/index/6d016391-9f27-46ab-98fd-8c1442150921
https://crash-stats.mozilla.com/report/index/a5a01d1b-3d0d-4a29-afa4-4ea8a2150921
https://crash-stats.mozilla.com/report/index/d5930420-53d4-4527-9923-a38602150921
https://crash-stats.mozilla.com/report/index/06c1a127-3a03-4574-b0d8-896bc2150921
https://crash-stats.mozilla.com/report/index/a8e1f7ec-d5cb-447c-ab94-ba9562150921
https://crash-stats.mozilla.com/report/index/6ab9611c-4c3d-4a87-8344-77e832150921

I find it interesting that most of the crashes have to do with image processing.
Hmm, those signatures are all over the place - some are in JS, some are in the image decoders.

David, any idea what's up here? This was filed amongst a thread of people on nightly-testers saying that recent Nightly builds crash on startup for them.
Keywords: crash
Something is randomly overwriting memory. JS and imagelib are the most frequent victims merely because we spend most of our time in those areas. It could be interference from external utilities or malware.
People on the mailing list say this started with Nightly builds of 2015-09-20, makes them crash at startup and on different OSes (Windows and Linux at least).

From that, I doubt it's something 3rd-party but it sounds like something that landed in Nightly on the 19th.
Ok, that wasn't clear from the description.

If this is reproducible for many people, perhaps someone could bisect this down to a single changeset?
Alice indicated bug 1201796 as causing bug 1206753, which might be related to the crashes being mentioned here (as another instance of memory corruption making the GC fall over, like the 3rd example in comment #0).
Thanks! That sounds spot-on since one of the crash reports was in the ICO decoder.
Blocks: 1201796
This also landed in the right timespan, so it sounds very much like it could be the cause. We probably should get it backed out from both nightly and aurora branches (as it was uplifted to the latter a few hours ago by the 43 merge).
Seth Fowler is already looking in to this on bug 1206626.
Somehow i got similar surge of crashes with various signatures in latest build too

Crash Reports for the Last 3 Days
---------------------------------

Report ID: bp-622da314-bd4c-4004-849d-a64fc2150921
Submitted: 41 minutes ago

Report ID: bp-92d282fe-1859-48bf-b61b-516c32150921
Submitted: 50 minutes ago

Report ID: bp-793af2c8-80f1-4617-b097-16a642150921
Submitted: 51 minutes ago

Report ID: bp-0505a7f3-151c-446e-bcf6-7282e2150921
Submitted: 58 minutes ago

Report ID: bp-98c30792-b17c-4007-8bbb-eba3d2150921
Submitted: 59 minutes ago

Report ID: bp-b5ff249f-0129-4ed6-a2c9-78ccd2150921
Submitted: 1 hour ago

Report ID: bp-e35c04cf-dc61-4f39-9b8c-67f9f2150921
Submitted: 1 hour ago

Report ID: bp-c229ebbf-2252-4042-9057-d35a32150921
Submitted: 1 hour ago

Report ID: bp-4a8d43d6-a009-4746-8f12-5848f2150921
Submitted: 1 hour ago

Report ID: bp-ebd96109-dfe6-41a3-b4d2-9231b2150921
Submitted: 2 hours ago

Report ID: bp-378619e9-0c77-49b3-aabb-46ffd2150921
Submitted: 2 hours ago

I dunno how its all related but rolling back to 19 Sept build fixes all problems
Sorry for bugspam if its unrelated.
Same here, random crashes, they happen when I open a few amount of tabs at once.
I can post links if needed.

[@ mozilla::net::nsHttpConnection::OnSocketReadable() ]
https://crash-stats.mozilla.com/report/index/a1845bbf-c56b-4f19-956d-1a0692150921

[@ IsAboutToBeFinalizedInternal<T> ]
https://crash-stats.mozilla.com/report/index/79a014ef-5bc7-4acb-8efc-d830c2150921

[@ _cairo_hash_table_lookup_exact_key ]
https://crash-stats.mozilla.com/report/index/52b8dab4-17e6-4031-9c4b-b27e82150921

[@ JSScript::getName(unsigned char*) ]
https://crash-stats.mozilla.com/report/index/d953c552-1aa9-40c2-abb7-d0bb72150921

[@ mozilla::layers::WalkTheTree<T> ]
https://crash-stats.mozilla.com/report/index/54b01f90-2801-4d18-a94b-cc9bd2150922

[@ nsTArray_Impl<T>::IndexOf<T>(mozilla::net::nsHttpConnection* const&, unsigned int, nsDefaultComparator<T> const&) | nsRefreshDriver::RemoveLayoutFlushObserver(nsIPresShell*) ]
https://crash-stats.mozilla.com/report/index/3baad0bb-5c11-4730-9add-e17222150922
Severity: normal → critical
Status: UNCONFIRMED → NEW
Ever confirmed: true
Keywords: crashreportid
I don't think we need more information here. The crashes are the result of memory corruption, so the signatures don't mean much - and the regressing bug has been identified as bug 1201796.
No longer blocks: 1201796
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.