803470 - Freeze when visiting Google Data Center via Google StreetView

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Attachment: activity monitor sample

Mozilla/5.0 (Macintosh; Intel Mac OS X 10.7; rv:18.0) Gecko/18.0 Firefox/18.0 ID:20121018042010

Today Firefox froze completely when I was doing a walk through the Google datacenter via Google StreetView.

Steps:
1. Load https://maps.google.com/maps?hl=en&ll=35.898645,-81.548671&spn=0.005945,0.012306&sll=35.900197,-81.547024&layer=c&cid=7373938251588581469&panoid=QJRzrgIKmN4uPQg7lmOGww&cbp=13,119.83,,0,3.72&gl=US&t=m&z=17&cbll=35.898621,-81.548447

2. Enter the building and take the stairs into the second floor

3. Turn right and quickly go forward so you pass the green wall

At this point Firefox freeze and I  had to force kill the application. I will attach a sample made with activity monitor which will hopefully be of help. So far I haven't been able to reproduce it.

Comment 1

[Boris Zbarsky [:bzbarsky]]

Hmm.  This looks like a dup of bug 799282 (via bug 800393), but that claims to be fixed on Aurora...

Comment 2

[Marty Rosenberg [:mjrosenb]]

*groan* not again :(

Comment 3

[David Anderson [:dvander] - inactive, e-mail if emergency]

Marty, how hard would it be to introduce a cutoff to range analysis? We've seen a few sporadic cases of it hanging browsers, even after bug 799282 landed. If this is simple, it would be a free stability win until we get better STR.

Comment 4

[Marty Rosenberg [:mjrosenb]]

oh, I have a patch for that lying around.  I'll post it as soon as I can find it.

Comment 5

[Marty Rosenberg [:mjrosenb]]

Attachment: /home/mrosenberg/patches/bailOnILoop-r1.patch

fixed up the non-posted patch to be more compatible with any future changes.

Comment 6

[David Anderson [:dvander] - inactive, e-mail if emergency]

Comment on attachment 681791 [details] [diff] [review]
/home/mrosenberg/patches/bailOnILoop-r1.patch

Review of attachment 681791 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks!

Comment 7

[Marty Rosenberg [:mjrosenb]]

landed: https://hg.mozilla.org/integration/mozilla-inbound/rev/4f37c3ad5cda
We may want to not close this bug, since this is just a workaround / a measure to detect when this happens in the future, not an actual fix.

Comment 8

[Ryan VanderMeulen [:RyanVM]]

;)

Comment 9

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/4f37c3ad5cda

Comment 10

[David Anderson [:dvander] - inactive, e-mail if emergency]

Marking FIXED since range analysis has changed significantly, the workaround will only exist on branch.

Comment 11

[David Anderson [:dvander] - inactive, e-mail if emergency]

Comment on attachment 681791 [details] [diff] [review]
/home/mrosenberg/patches/bailOnILoop-r1.patch

[Approval Request Comment]
Bug caused by (feature/regressing bug #): IonMonkey
User impact if declined: Possible hang on Google street maps
Testing completed (on m-c, etc.): Yes
Risk to taking this patch (and alternatives if risky): Very little, it detects when a small performance analysis pass in IonMonkey might hang and aborts the pass.
String or UUID changes made by this patch:

Comment 12

[Scoobidiver (away)]

It has been two weeks since the Beta approval without any patch landing.

Comment 13

[Ryan VanderMeulen [:RyanVM]]

This landed on beta but was just never marked as so...
https://hg.mozilla.org/releases/mozilla-beta/rev/ba252f7d3be6

Comment 14

[Virgil Dicu [:virgil] [QA]]

Could not reproduce in 2012-10-18 Aurora
Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:18.0) Gecko/18.0 Firefox/18.0
Build ID:  20121018042010

But there is no freeze in Latest Aurora
Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:19.0) Gecko/20121211 Firefox/19.0
Build ID: 20121211042015

and in Firefox 18.0 Beta 3
Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:18.0) Gecko/20100101 Firefox/18.0
Build ID:  20121205060959

Comment 15

[Henrik Skupin [:whimboo][⌚️UTC+2]]

(In reply to Virgil Dicu [:virgil] [QA] from comment #14)
> Could not reproduce in 2012-10-18 Aurora

I just want to note that this was a one-time crash for me and I was never able to reproduce. So it might be hard to really verify that.

Comment 16

[u279076]

Marking this [qa-] since this was a one-time crash that was not reproducible, as per comment 15. We won't be able to reliably verify this is fixed.