[TIMOB-2658] Crash on app resume
| GitHub Issue | n/a |
|---|---|
| Type | Bug |
| Priority | Low |
| Status | Closed |
| Resolution | Fixed |
| Resolution Date | 2011-04-17T01:59:44.000+0000 |
| Affected Version/s | n/a |
| Fix Version/s | Release 1.6.0 M02 |
| Components | iOS |
| Labels | crash, ios, release-1.6.0 |
| Reporter | Stephen Tramer |
| Assignee | Stephen Tramer |
| Created | 2011-04-15T03:26:09.000+0000 |
| Updated | 2011-04-17T01:59:44.000+0000 |
Description
Under certain (very rare) circumstances, apps will crash upon resume. This appears to be because in iOS 4.0, an app may still be in the "terminating" state when it is re-resumed if the system determined that it should be shut down. If the app is locked and waiting on the main thread for a response, this may cause the OS to terminate the app for not responding in time when resuming.
The issue appears to come from -[AnalyticsModule shutdown:], where an HTTP request is scheduled on the main thread. Marking medium because it's such a weird corner case and hard to reproduce.
Bug may not be reproducible enough to reliably test a fix. This one is 100% crash log.
Attachments
| File | Date | Size |
|---|---|---|
| log1 | 2011-04-15T03:26:09.000+0000 | 25526 |
| log2 | 2011-04-15T03:26:10.000+0000 | 22558 |
| orientation_2010-12-20-145419_the-last-unicorn.crash | 2011-04-15T03:26:10.000+0000 | 24163 |
I've seen this issue pop up a number of times in my users' crash logs. My app has analytics disabled, so that doesn't seem to be the sole cause. At the time it looked like the app was freezing in redraw code, so it's possible my users won't see this once I put out a new version based on Ti 1.5.
If you would be able to provide crash logs that would be extremely helpful. If you can symbolize them first at all, that would be great. The instructions here indicate how to do that:
http://furbo.org/2008/08/08/symbolicatifination/">http://furbo.org/2008/08/08/symbolicatifination/
Here's one I have on hand.
And not sure if this is related or a separate issue, but from the same user
These seem to have a different underlying cause, but if these issues persist after an upgrade to 1.5.0, please file a ticket for them. There were a large number of changes to the view drawing system that may have fixed these issues.
yep, thought as much. I'll let you know if they crop up again after I ship with 1.5.
(from [3165d74ad25b899f6d820b2ca76352fe13d9001a]) [#2658 state:fixed-in-qa] Speed up shutdown analytics events by reducing timeout. https://github.com/appcelerator/titanium_mobile/commit/3165d74ad25b899f6d820b2ca76352fe13d9001a"> https://github.com/appcelerator/titanium_mobile/commit/3165d74ad25b...
not much I can do about this one, closing
Stephen: I haven't seen log1 pop up again, but I'm still getting log2 on 1.5.1. I created a helpdesk ticket which includes new logs with full symbols, and it appears to be a deadlock in NSCondition: http://developer.appcelerator.com/helpdesk/view/66761">http://developer.appcelerator.com/helpdesk/view/66761
Damien, we're aware that there are some deadlock problems with certain UI operations right now, and we're trying to track all of them down. If you can create a new bug to track this issue and include the HD link/crash logs with it I would appreciate it.
Getting identical issues described in this support ticket:
http://developer.appcelerator.com/helpdesk/view/76350">http://developer.appcelerator.com/helpdesk/view/76350