Closed Bug 603464 Opened 14 years ago Closed 13 years ago

[PPC] Crash in [@ FlashPlayer-10.4-10.5@0x410380 ] [@ FlashPlayer-10.4-10.5@0x40fa00 ]

Categories

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

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

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: marcia, Assigned: smadayag)

References

(
URL
)

Details

(Keywords: crash, reproducible)

Crash Data

Seen while reviewing crash stats. This is the top Mac crash on the 1.9.2 and is affecting users who try to print using a PPC mac.

STR:
1. Load the site in the URL
2. Watch the flash of the print dialog.
3. File | Print
4. Crash 100%

Running Flash Version: 10.1.82.76

http://crash-stats.mozilla.com/report/index/bp-3ab0bdbb-16ce-47aa-af18-277ae2101011 was one of my reports. http://crash-stats.mozilla.com/report/index/bp-0d8519c5-351b-461c-8cce-3daa12101011 is the other, which crashes in [@ _class_isInitialized ].
Ok, this looks to have happened all the way back to 3.6 proper but we'd like to know what is going on with this reproducible case.

I also confirmed it doesn't happen on Intel. Assigning to markus (naively as his name is all over http://hg.mozilla.org/releases/mozilla-1.9.2/annotate/448d0d2d310c/widget/src/cocoa/nsPrintDialogX.mm)
(In reply to comment #1)
> Ok, this looks to have happened all the way back to 3.6 proper but we'd like to
> know what is going on with this reproducible case.

FWIW, we didn't support printing plug-ins prior to bug 191046 (which landed for 1.9.2a1, looks like)—and that bug uncovered some very nasty problems with npapi headers in Gecko.
Marcia:  If possible try on different machines (PPC ones) and with different printer drivers, on different OS versions (OS X 10.4.X and 10.5.X).
Steven: We have one main PPC machine in the lab. I can try out the only other PPC machine we have tomorrow and do some further testing.

So far the only printer I printed to was the HP 1300 in the lab, which is a networked printer. I have a few other local printers I can try as well.

(In reply to comment #3)
> Marcia:  If possible try on different machines (PPC ones) and with different
> printer drivers, on different OS versions (OS X 10.4.X and 10.5.X).
I crash on an old PPC G4 (running OS X 10.4.11) with the latest version of the Flash plugin from Adobe (10.1r85), in FF 3.6.10.  No Breakpad dialog (at least this time).

But I don't crash in FF 3.6.9.  So this may be at least partially our fault.

I don't have any printer drivers installed (nor can I install any).  Like Marcia's crashes, mine happened as the print dialog opened.
> But I don't crash in FF 3.6.9.

Oops, this is wrong.

I don't crash on most Flash sites in either FF 3.6.9 or 3.6.10.  But I do crash in both versions when I visit http://www.villagevoice.com/content/printVersion/2130170/, which seems to spawn the print dialog all by itself.
Now I also have some Breakpad crashlogs.  They're quite different from Marcia's, but maybe that's because I'm on OS X 10.4.11:

bp-739f3164-e89b-4832-ace9-b04c12101011
bp-a03cd6f6-f7a8-4f5a-9a50-d4aca2101011
bp-1abbb9cd-091b-4d45-984e-89a012101011
I don't crash in FF 3.5.13 or 3.5.12 (visiting http://www.villagevoice.com/content/printVersion/2130170/).

I think this rules out bug 191046, which is on all branches back to 1.9.0.
Here's something that's going to make this bug very difficult to track
down:

I sometimes don't crash using a given 1.9.2-branch distro.  But the same
distro crashes if I try Marcia's comment #0 STR first thing after
rebooting.  So there's a timing element to this bug, and it's probably
much more likely to happen on old (and slow) machines.

Even just after restart, though, I still don't crash in FF 3.5.13.
For what it's worth, the most recent FF 3.6.X release in which I
consistently don't crash (even after restart) is FF 3.6.3.  So the
"guilty" patch must be new with FF 3.6.4.  Markus and Josh, can you
guess which patch that might be?
Marcia, if you can truly reproduce this bug's crash 100% of the time, please try to find a proper regression range -- something better than my "between FF 3.6.3 and 3.6.4".
I'd be curious if this happens with Flash disabled and/or with Flash run in process (http://support.mozilla.com/en-US/kb/The+Adobe+Flash+plugin+has+crashed#Disable_hang_protection).

If I had a PPC machine I would test myself...
FF 3.6.10 doesn't crash with the Flash plugin disabled.  Neither does
it crash with the ancient Flash plugin version (9.0r246) which is the
latest version available from Apple's Software Update for OS X 10.4.11
PPC.

So the Flash plugin definitely plays a role in these crashes.

Firefox on PPC always runs plugins in-process -- we only support
out-of-process plugins on OS X 10.6.X, which doesn't run on PPC (only
on Intel).

By the way, I strongly suspect there's nothing inherently PPC-specific
about these crashes.  In other words, someone could probably reproduce
them with a sufficiently old and slow Intel Mac.  But even my
first-generation PowerBook Pro (with 2GB RAM and a 2Ghz Intel Core Duo
processor) doesn't seem to be old and slow enough.  I once got close
to crashing FF 3.6.10 on it -- the Print dialog disappeared
spontaneously a second or two after it appeared -- but I didn't
actually see a crash.
I spoke too soon -- here's one of these crashes on my old first-generation Intel MacBook Pro:

bp-e1231773-66fb-4eae-b29e-74ee22101011
Will work on this in the lab tomorrow since I don't have a PPC machine here with me at the moment. 

(In reply to comment #11)
> Marcia, if you can truly reproduce this bug's crash 100% of the time, please
> try to find a proper regression range -- something better than my "between FF
> 3.6.3 and 3.6.4".
I'm able to reproduce Marcia's crash stack with Flash 10.1r82:

bp-561fdccb-fc26-4966-9558-d402c2101011

By the way, to installer earlier Flash versions over later ones, you first have to delete "/Library/Receipts/Adobe Flash Player.pkg".

If you don't do this, the installer will appear to run, but it won't actually do anything (and won't display or log any errors).  Extremely annoying!!
(Following up comment #14)

Here's another crash on my first generation Intel MacBook Pro, using FF 3.6.10 and Flash 10.1r82:

bp-ca15fb96-7bdd-4f6c-9320-3477b2101011

My previous crash was with FF 3.6.10 and Flash 10.1r85.

Both crashes have the same signature.  And crash-stats doesn't see many of these crashes.  So this, bug, while not PPC-specific, does seem to be a much more serious problem on PPC machines than on Intel machines -- probably because the PPC machines are so much older and slower.

Also (judging again from the crash-stats statistics), there seem to be many more people seeing this bug with Flash 10.1r82 than with 10.1r85.  This is odd, because I'm able to reproduce this bug with 10.1r85 as easily as I'm able to reproduce it with 10.1r82.  Maybe it's just that there are still so many users on 10.1r82 (and so few on 10.1r85).
There also a bunch of crashes on the 1.9.2 branch (1327 in the last week) at FlashPlayer-10.4-10.5@0x40fa00, all on PPC, which (judging by the comments) are also the same bug.
Summary: [PPC] Crash in [@ FlashPlayer-10.4-10.5@0x410380 ] → [PPC] Crash in [@ FlashPlayer-10.4-10.5@0x410380 ] [@ FlashPlayer-10.4-10.5@0x40fa00 ]
We've been trying to find a repro case for some time...  
I'm not able to reproduce with the machines we have, which are G5 towers running 10.4.11 and 10.5.8.

Can another person confirm which processor they encounter the crash with?

Tested:
FF3.6.10 PPC G5 Flash 10,1,85,3 on OS X 10.4.11/10.5.8 Default Lexmark printer.
Dual 1.8 GHz PowerPC G5 with 1.25 GB RAM
Running OX 10.5.8
Flash Version: 10.1.82.76

Using Mozilla/5.0 (Macintosh; U; PPC Mac OS X 10.5; en-US; rv:1.9.2.10) Gecko/20100914 Firefox/3.6.10. Default Printer is a network HP 1300 Printer. URL crashes 100% in this scenario.

I will try some other printers in my lab as well.

(In reply to comment #19)
> We've been trying to find a repro case for some time...  
> I'm not able to reproduce with the machines we have, which are G5 towers
> running 10.4.11 and 10.5.8.
> 
> Can another person confirm which processor they encounter the crash with?
> 
> Tested:
> FF3.6.10 PPC G5 Flash 10,1,85,3 on OS X 10.4.11/10.5.8 Default Lexmark printer.
Using the same machine noted in Comment 20, I crash all the way back to Firefox 3.6 ->Mozilla/5.0 (Macintosh; U; PPC Mac OS X 10.5; en-US; rv:1.9.2) Gecko/20100115 Firefox/3.6.
Now that I upgraded my Flash Version to Version: 10.1.85.3, it seems I am not able to repro the crash on 3.6 or 3.6.10.
Marcia, please downgrade back to Flash 10.1.82 and see if you crash in FF 3.5.X.

If so (as I expect), please find which pre-FF-3.6 nightly the crashes start with.

Note that for the downgrade to work, you'll first have to delete "/Library/Receipts/Adobe Flash Player.pkg" (as mentioned in comment #16 above).

I suspect you *will* see crashes using Flash 10.1.85, as I have -- but you won't be able to reproduce them as reliably.  Which means that only with Flash 10.1r82 (and the rest of your setup) do we have any chance to find a proper regression range.
> If so (as I expect) ...

Oops, I meant "if you don't crash in FF 3.5.X, as I expect" ...
No crash using  Mozilla/5.0 (Macintosh; U; PPC Mac OS X 10.5; en-US; rv:1.9.1.13) Gecko/20100914 Firefox/3.5.13 and Shockwave Flash 10.1 r82. 

Now on to the regression range hunt.
I crash somewhat more reliably using Flash 10.1r82, and I've now also seen crashes going back to FF 3.6.

Still no crashes in FF 3.5.13, though.
Keywords: reproducible
I tried 3.6 Beta 1 just to make sure that it didn't regressed between beta and 3.6 final. But 3.6 B1 crashes, so the regression happened in the pre-FF 3.6 days. This will take a bit more digging but I have started the process.
Hey everyone.  We were able to reproduce in house and have a fix that will land in the next Flash Player release, Salt.

These were the repro step we used:
1. Launch FF and navigate to http://www.bestsampleresume.com/designers-resumes.html.
2. File > Print. Printer = Adobe PDF 9.0.
3. Click Print button.
(In reply to comment #28)

Thanks for the new site to test with!  I crash much more easily with
it than with the original site
(http://www.villagevoice.com/content/printVersion/2130170/).

And I crash without the "Adobe PDF 9.0" printer driver, or any printer
driver, installed.  All I need to do is open the Print dialog and
click on the Preview button.

Thanks also for promising to fix this!
has anyone encountered any repeat crashers after the Flash Player Salt upgrade 10,1,102,64?  i'm going to close this fix resolved.
Status: NEW → RESOLVED
Closed: 13 years ago
Resolution: --- → FIXED
Assignee: nobody → smadayag
Component: Plug-ins → Flash (Adobe)
Product: Core → Plugins
QA Contact: plugins → adobe-flash
Target Milestone: --- → Jan 2011
Version: 1.9.2 Branch → 10.x
Crash Signature: [@ FlashPlayer-10.4-10.5@0x410380 ] [@ FlashPlayer-10.4-10.5@0x40fa00 ]
Version and milestone values are being reset to defaults as part of product refactoring.
Target Milestone: Jan 2011 → ---
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.