520058 - Flash crash on Mac OS X [@ Flash Player@0x92160 | nsNPAPIPluginInstance::HandleEvent] [@ Flash Player@0x92160] [@ Flash Player@0x91d90]

Status: RESOLVED WORKSFORME

(External Software Affecting Firefox Graveyard :: Flash (Adobe), defect)

Comment 1

[Josh Aas]

This crash happens in response to a number of different events from the browser - idle (null) events from the idle timer, mouse up, stream-has-data. Almost certainly a crash in Flash that will require a fix from Adobe.

Comment 2

[Marcia Knous [:marcia]]

There aren't many comments in crash-stats regarding this bug except that they crashed on videos.google.com when using the back button and while watching a youtube video.

Comment 3

[Carsten Book [:Tomcat]]

Crashstats link:

http://crash-stats.mozilla.com/report/list?product=Firefox&query_search=signature&query_type=startswith&query=Flash%20Player%400x92160&date=&range_value=1&range_unit=weeks&do_query=1&signature=Flash%20Player%400x92160

Comment 4

[Bob Clary [:bc] (inactive)]

this showed up in the crash logs on 8/1. It might be Flash 10.0.32.18 considering it was released about then.

Comment 5

[Damon Sicore (:damons)]

Can we get URLs for this?

Comment 6

[Damon Sicore (:damons)]

Johnny, can you add this to the list of Flash issues that you are tracking?

Comment 7

[Johnny Stenback (:jst)]

Sure thing. I pulled together the URLs for the crashes we have for this. Out of 6687 crashes, 1354 were from www.youtube.com/watch, another couple of hundred were other various sites with primarily flash videos on them, and the rest are fairly unique, so not much of value there. Flash video does appear to stand out here, but no one particular site or video. In fact, there's no one youtube video that appears in our data more than twice with this crash signature.

Comment 8

[Josh Aas]

Not having any URLs that stand out is in line with the fact that stack traces show the crash happening in response to many different kinds of events. There are probably quite a few ways to reproduce which are individually hard to hit.

Comment 9

[Samuel Sidler (old account; do not CC)]

Looking through dbaron's analysis, here's the Flash information:

100% (302/302) vs.  90% (16597/18510) Flash Player
  1% (2/302)   vs.   1% (136/18510) checksum: 27CC04C9588E482A948FB5A87E22687B0
  1% (4/302)   vs.   2% (280/18510) checksum: 38AEB67F6A0B43C6A341D7936603E84A0
 31% (95/302)  vs.  23% (4198/18510) checksum: 860692A215F054B7B9474B410ABEB5300
 66% (198/302) vs.  43% (7924/18510) checksum: B19EE2363941C9582E040B99BB5E237A0
  0% (1/302)   vs.   1% (144/18510) checksum: CF7A0D2CC2664865921FD463FC8121020
  1% (2/302)   vs.   1% (105/18510) checksum: F43004FFC4944F26AF228334F2CDA80B0

I went through and determined version numbers for our checksums. 

  https://wiki.mozilla.org/CrashKill/Mac_Flash_Identifiers

The 31% one is 10.0.22.87 on x86. The 66% one is 10.0.32.18 on x86.

Comment 10

[Josh Aas]

Sam and I talked about comment #9 and we think the data there is incorrect for some reason. Looking at the data again we see:

        100% (593/593) vs.  43% (7924/18510) checksum: B19EE2363941C9582E040B99BB5E237A0

which indicates that this is a new crash in Flash 10.0.32.18 on x86.

Comment 11

[Damon Sicore (:damons)]

Updates here?

Comment 12

[Bob Clary [:bc] (inactive)]

Not really. We can't reproduce but we don't trust our urls either.

Comment 13

[Josh Aas]

This is now the #1 crash for Firefox 3.5.3.

Comment 14

[Josh Aas]

Sorry, it is the #1 crash for Firefox 3.5.4.

Comment 15

[Samuel Sidler (old account; do not CC)]

(On Mac)

Comment 16

[Chris Ilias [:cilias]]

Looks like we'll need to update <http://support.mozilla.com/en-US/kb/Pages+that+use+the+Flash+plugin+crash+on+Mac>.
The solution on that article is to update Flash, but AIUI this crash occurs with the latest version of Flash. For this crash, the cause is still being investigated, correct?

Comment 17

[Tim Riley [:timr]]

Marcia-  Is there anything else we can do to work with Josh or ping our community to try to repro this.  It was the #1 crash on 3.5.4 and it's qawanted.

Comment 19

[Marcia Knous [:marcia]]

Best thing we can do is maybe put a post on QMO. This is one of many bugs that involve flash crashes that are often difficult to reproduce. 

(In reply to comment #17)
> Marcia-  Is there anything else we can do to work with Josh or ping our
> community to try to repro this.  It was the #1 crash on 3.5.4 and it's
> qawanted.

Comment 20

[Damon Sicore (:damons)]

Never nominated, but marking blocking1.9.2- to explicitly mark [CrashKill] bugs as either blocking or not.

Comment 22

[Charles]

Flash player 10.1 has Out Of Memory handling.  Its possible that change could have addressed this crash?  

Is it possible for you guys to find out based on the stacks of there are any reports from that version of the player?

Comment 23

[Chris Lawson (gone)]

(In reply to comment #22)
> Flash player 10.1 has Out Of Memory handling.  Its possible that change could
> have addressed this crash?

If it has, it's been totally masked by the tremendous uptick in crashes that were introduced in the 10.1 betas.

Comment 24

[Chris Lawson (gone)]

For the record, though, I just took a quick look at the 10 most recent reports at each of these URLs:

http://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&signature=Flash%20Player%400x92160&version=Camino%3A2.0.1

http://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&signature=Flash%20Player%400x92160&version=Firefox%3A3.6

http://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&signature=Flash%20Player%400x92160  (In the time it took me to type this comment, six more reports came in at this URL.)

B19EE2363941C9582E040B99BB5E237A0 is the module ID for Flash Player in all 36 (three times 10 plus the six that came in later) of them, which, according to

https://wiki.mozilla.org/CrashKill/Mac_Flash_Identifiers

is 10.0.32.18 on x86.

Assuming those 36 reports are a representative sample, and assuming that people are equally likely to upgrade Firefox as they are their Flash plugin, you might be able to draw the conclusion that a newer version of the plugin has fixed this specific crash. Two huge caveats, though:

1) That's far more assumptions than I'm comfortable making.
2) What does out-of-memory protection have to do with user events? (See comment 0.)

What we really need is a way to search crash-stats by module ID, so that we could search specifically for the module IDs for later Flash 10.0 versions to see if this crash appears with those. Without that ability, I don't think there's a way to say for certain that this crash is solved with the 10.0.42.34 release.

The 10.1 betas have a different library name and show up with different stack signatures. Again, without steps to reproduce or Flash symbols, it's pretty much impossible to know for sure whether this crash doesn't occur in 10.1+ versions of the Flash plugin.

Comment 25

[Chris Lawson (gone)]

(In reply to comment #16)
> Looks like we'll need to update
> <http://support.mozilla.com/en-US/kb/Pages+that+use+the+Flash+plugin+crash+on+Mac>.
> The solution on that article is to update Flash, but AIUI this crash occurs
> with the latest version of Flash.

At the time of this comment, I believe 10.0.32.18 was still the "latest". With all the caveats in comment 24, there does exist a *possibility* that 10.0.42.34 has fixed this specific crash, so maybe the solution is to tell people to update to that version (*NOT* the 10.1 betas).

cl

Comment 26

[Chris Lawson (gone)]

Heh. Looks like [@Flash Player@0x91d90] *could* be the same thing in 10.0.42.34:

http://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&signature=Flash%20Player%400x91d90

All the Flash module IDs from the 11 most recent there (another one came in while I was typing this) correspond to 10.0.42.34.

Comment 27

[Charles]

We're tracking this as looking into the association with "Flash Player@0x91d90" which appears to be reports with Flash Player 10.0.42.34

Comment 28

[Stuart Morgan]

Adding the 10.0r42 variant to the summary so crash-stats picks it up.
FYI, the Chromium equivalent of this is http://crbug.com/32082

Comment 29

[Marcia Knous [:marcia]]

I spied one of these crashes today on the trunk: https://crash-stats.mozilla.com/report/index/3f1082fd-4ed5-4d5e-984e-da3f72100511 - mentioning it since many of the bug comments here are discussing it happening on 3.5.x.

Comment 30

[Jet Villegas (inactive)]

Picking this up.

Comment 31

[Jet Villegas (inactive)]

Bear with me as I parse through the comments. It looks like this crash breaks down as follows:

1% on 9.0.151.0
1% on 9.0.159.0
1% on 10.0.12.36
31% on 10.0.22.87 
66% on 10.0.32.18
unknown% on 10.0.42.34

We've fixed a bunch of video/memory/crash bugs in 10.1.53.64.
I'd like to see if i can mine the data for correlations to this ID:

4FE0EBB4DD56E42E17926466E720365F0

I'm marking this WORKSFORME to get the regressions and FP 10.1 crash dumps (if any) that correlate to this crash.

JV

Comment 32

[BMO Automation]

Version and milestone values are being reset to defaults as part of product refactoring.