Closed Bug 494448 Opened 16 years ago Closed 16 years ago

[Mac] Crash in [@ CFReadStreamGetStatus ] or [@ CoreFoundation@0x6f78d ]

Categories

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

Product:
External Software Affecting Firefox Graveyard
Component:
Flash (Adobe)
Platform:
x86
macOS
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: marcia, Unassigned)

References

(
URL
)

Details

(Keywords: crash, Whiteboard: Fixed 10.0.32.18)

Crash Data

I crashed in this stack last night using the latest nightly. Breakpad shows 3946 crashes, all on Mac with this stack. There is a mix of 3.5b4 and 3.0.0.10 versions. The notes section reveals sites such as NY Times, Myspace, mlb.com, hulu.com as sites they were visiting when
Forgot to note that 274 of the crashes occurred using B4, the remainder are all 3.0.0.10 crashes. http://tinyurl.com/qc6lj4 shows all the crashes to date in this stack.
I don't have definitive STR yet but adding smichaud to get his opinion on whether this might be our issue or potentially a flash issue.
> adding smichaud to get his opinion on whether this might be our > issue or potentially a flash issue Since this crash happens deep in Flash code (and on a secondary thread probably created by the Flash plugin), it's very likely a Flash problem.
The machine I crashed on was using 10.0. r22.
This is with 3.5 beta? Can some please post reproducible steps as well as OS level info?
My crash actually happened on the trunk using 20090520033508, but when I searched the crash data and filed this bug people using the 3.5 Beta 4 have hit it as well. If I remember correctly, I was clicking on a flash ad when the crash occurred. There are some additional comments in the link Comment 2 which indicate what the person was doing when the crash occurred. Let me know what OS info you need.
Some sites noted in the comments: hulu.com nytimes.com IMDB.com
The machine I crashed on was a Macbook running 10.5.7. The crash reports in Comment 2 indicate that it is affecting 10.5.6 as well.
Severity: major → critical
Summary: [Mac] Crash in [@ CFReadStreamGetStatus ] → [Mac] Crash in [@ CFReadStreamGetStatus ] or [@ CoreFoundation@0x6f78d ]
From bug 496285, comment 0: Crash Reports query: http://is.gd/NTvf Flash crashes Firefox in all 1.9.1 versions since 3.1b1 (through 3.5b4), but only on Mac OS X 10.4. 66 reports in the last two weeks of crash signature [ CoreFoundation@0x6f78d ].
So far I cannot reproduce this crash on Mac 10.4 running FF 3.5B4. I have run various videos on Hulu, youTube and NYtimes. Can some one help and provide more details as to how many tabs where running and what was done when you crashed? Can anyone reproduce this same crash in Safari or on Windows?
http://tinyurl.com/m4fexj shows 591 people crashed in this stack on Mac using the Beta Preview release. Sites mentioned in the comments are universalsports.com and http://www.deredactie.be/cm/vrtnieuws
More info noted in the crash data - myspace crashe firefox everytime that song player loads it crashes, reader.google.com. I have not been able to reproduce this since my initial report, but there are now close to 10K reports using 3.0.11 and 3.5 according to a crash-stats query.
Firefox 3.5 has amassed 1030 crashes to date in this stack, all Mac. Other sites noted in the comments are http://www.wral.com/traffic/, www.msnbc.com and bbc.com.
The Flash Player team has internally made a potential fix a for the CFReadStreamGetStatus crash. If you are aware of a way to reproduce that crash (other than using Valgrind), please let me know.
The urls in bug 500939 reproduced the CFReadStreamGetStatus stacks.
The Flash Player team was able to reproduce the CFReadStreamGetStatus crash with http://www.classique-shop.ch/ and I've put this URL in the URL field for this bug. Thank you for pointing us to those URLs! The CoreFoundation@0x6f78d is likely to be a separate bug. We would like to reproduce that as well, if you have URLs to recommend.
URL: http://www.classique-shop.ch/
(In reply to comment #19) > The CoreFoundation@0x6f78d is likely to be a separate bug. We would like to > reproduce that as well, if you have URLs to recommend. Michelle, just out of curiosity, why do you think it's a separate bug? The crashed threads look nearly identical (with the main exception of the one set not having been resymbolized), and the 0x6f78d address is in the CFReadStreamGetStatus function in 8S2167 OS builds for which we have symbols: PUBLIC 6f77c 0 CFReadStreamGetStatus [6f78d] PUBLIC 6f7f6 0 CFReadStreamSignalEvent For reasons unknown (and quite frustrating) to me, there seem to be OS builds with the same build numbers, e.g. 10.4.11 8S2167, but where the "Debug Identifier"/module ID for some of the libraries differ. I can't say for certain that CoreFoundation in the "mystery 8S2167" in these crashes is the same as the CoreFoundation in the 8S2167 we pulled symbols from, but in the cases where we have been able to compare two libraries with different IDs but from identical OS build numbers, that has been the case.
(In reply to comment #21) Smokey, you have excellent data on this. I take back my thought about this being a separate issue. We have repro'd the CFReadStreamGetStatus crash in house and have verified that it is fixed for a future Flash Player release. Thanks for everyone's help!!
Michelle, is the just-released Flash Player 10.0.32.18 supposed to resolve this crash, or will the fix be in a future release yet? I've heard anecdotal evidence that Flash/video sites are crashing less with the new version (and this particular crash was quite common), but I've seen/heard nothing concrete as of yet.
I can tell you that we targeted this bug with the latest release 10.0.32.18. We expect it to be fixed should you find otherwise we would be interested in knowning.
No crashes with Flash 10.0.32.18 for the urls in bug 500939. Do we mark this as fixed or wfm since it is in Flash?
(In reply to comment #25) > No crashes with Flash 10.0.32.18 for the urls in bug 500939. Do we mark this as > fixed or wfm since it is in Flash? Typically INVALID, since the bug was in a third-party product.
Great news on this one guys. Can we close it?
(In reply to comment #27) > Great news on this one guys. Can we close it? Sure. --> INVALID, per comment 26
Status: NEW → RESOLVED
Closed: 16 years ago
Resolution: --- → INVALID
We're still seeing ~1500 crashes/day in CFReadStreamGetStatus under Flash Player. What version corresponds to debug identifier 860692A215F054B7B9474B410ABEB5300 ? Example: http://crash-stats.mozilla.com/report/index/f35cb871-59b7-4336-b7b8-0e9092090929
What are the version numbers associated with these crashes. I am guess not 10.0.32.18. Since its Mac it should be available in the raw dump. I do not know what to do with the debug identifier Mike. Can you explain?
The raw dump line from the crash I linked above (click "raw dump" from that page) is: Module|Flash Player||Flash Player|860692A215F054B7B9474B410ABEB5300|0x11a5d000|0x12080fff|0 The debug identifier is the 5th column there, I believe it's unique per version, perhaps a hash of something.
(In reply to comment #31) > The raw dump line from the crash I linked above (click "raw dump" from that > page) is: > > Module|Flash Player||Flash > Player|860692A215F054B7B9474B410ABEB5300|0x11a5d000|0x12080fff|0 > > The debug identifier is the 5th column there, I believe it's unique per > version, perhaps a hash of something. That debug identifier is Flash 10.0.22.87 on Intel. I'm compiling a list now.
Component: Plug-ins → Flash (Adobe)
Product: Core → Plugins
QA Contact: plugins → adobe-flash
Resolution: INVALID → FIXED
Whiteboard: Fixed 10.0.32.18
Target Milestone: --- → Jul 2009
Version: 1.9.1 Branch → 10.x
Crash Signature: [@ CFReadStreamGetStatus ] [@ CoreFoundation@0x6f78d ]
Version and milestone values are being reset to defaults as part of product refactoring.
Target Milestone: Jul 2009 → ---
Version: 10.x → unspecified
Product: External Software Affecting Firefox → External Software Affecting Firefox Graveyard
You need to log in before you can comment on or make changes to this bug.