303795 - Firefox causes error in ATI2DRAG.DRV, then crashes

Status: RESOLVED INVALID

(Firefox :: General, defect)

Description

[Nick Howitt]

User-Agent:       Mozilla/5.0 (Windows; U; Win 9x 4.90; en-US; rv:1.8b4) Gecko/20050730 Firefox/1.0+
Build Identifier: Mozilla/5.0 (Windows; U; Win 9x 4.90; en-US; rv:1.8b4) Gecko/20050730 Firefox/1.0+

If I leave the PC for a while, it goes into powersaving mode and shuts the
display down. Bringing it out of powersaving by moving the mouse normally works.
Sometimes, though, I only see the mouse and the rest of the display stays black.
Sometimes I have to hard reboot. Sometimes hitting some keys brings the system
back to normal. Occasionally, hitting some keys brings the system back to life,
but windows flashes an error message then DPA terminates. A couple of times the
error message "Firefox has caused an error in ATI2DRAG.DRV" ahs remained on
screen. When I then close the error message Ff terminates.

This is not a new problem, but certainly was happening with DPA1 and possibly
previous nightlies. I do not know if it happens with 1.0.x.

Reproducible: Sometimes

Steps to Reproduce:
1.Leave the PC with Ff running.
2.Wait for it to go into power saving mode
3.Move the mouse to bring it to life

Actual Results:  
Display does not always resume correctly, giving a black screen with only the
mouse showing. Sometimes hitting some keys brings it back to life.

Expected Results:  
PC should resume normally

My graphics card is a Sapphire Radeon 9550 with 256Mb RAM. The driver is the
latest on ATI's site, 8-03-98-3-050117a-021000e, but it also happened with the
drivers supplied by Sapphire, and the latest drivers on their site (Catalyst
5.4). Note Sapphire also point you to ATI's site to get updated drivers.

Comment 1

[Matthias Versen [:Matti]]

I don't think that we can much do in this case because the ATI driver caused the
crash (ATI2DRAG.DRV).....

Comment 2

[Jaime Mitchell (use [e-mail] for email)]

Sounds like a display driver issue (as per comment 1) , what makes you believe
this is at all related to firefox and not your display drivers?

Comment 3

[Nick Howitt]

It could well be a display driver problem. I will not dispute that. The error
message says Ff caused the problem with the display driver. Is Ff doing
something naughty when it calls the display driver or is the display driver
doing something naughty when it tries to process a call from Ff? I do not know
the inner workings of either, or of Windows so I do not know the answer. I am
also submitting a report to ATI/Sapphire.

If anyone is 100% sure it is not a Ff fault I am happy for the bug to be marked
invalid.

Comment 4

[timeless]

reporter: could you at least install talkback and get a talkback report? there
seem to be people from ati who use bugzilla, but you really haven't given us
enough information to pass it along to them.

Comment 5

[zug_treno]

See also Suite bug 198395 and Suite bug 241088.

Comment 6

[Nick Howitt]

Talkback is installed. So far no log has been produced. I will post if I get one.

The bug is not manifesting itself in quite the same way as bug 198395 or bug
241088 as I have not seen a GPF message but another one. It still could be
linked. My symptoms are possibly different in that when I move the mouse to
bring the PC back to life I can see the mouse, but otherwise the screen is
black. It sometimes comes to life properly if I then type something. It is very
intermittent.

In response to comment 4, what other info would you like? Screen resolution is
1024x768, 32 bit colour, DVI interface. All driver settings are default except
show ATI icon in task bar. Let me know what other info would help.

Comment 7

[Nick Howitt]

I have had four crashes in the last couple of days and each time DPA has closed.
QFA is installed, but no talkback report is produced.

I have filed a ticket (737-854726) with ATI, but so far only had an automated
response. I am working my way down their list of things to try (bios changes
etc.), but it still crashes so far, even with DPA2 as the only open application.

Comment 8

[timeless]

hrm, ok, if you're willing to run firefox somewhat (possibly very) slowly, you 
could install windbg, and run firefox from there.

windbg is Microsoft's Windows Debugger (free, powerful, small, not intrusive), 
http://www.microsoft.com/whdc/DevTools/Debugging/default.mspx

as of this comment, the current version is available from: 
http://msdl.microsoft.com/download/symbols/debuggers/dbg_x86_6.5.3.7.exe

anyway, run windbg, there's lots of help for it (if you have time, i encourage 
you to read it), 

1. file>symbol path
srv*%appdata%\symbols*http://msdl.microsoft.com/download/symbols

(i'm not sure it'll appreciate %%, if it doesn't, pick something you like 
better -- you can find the value of %appdata% by entering it into windows 
explorer's location bar)

2. file>open executable
3. browse to firefox.exe
4. get properties for firefox.exe from the general tab, copy the location
5. close the properties dialog
6. paste that into the start directory field
7. check [x] debug child processes also (this is because i don't want to worry 
about NO_EM_RESTART)
8. select open
9. debug>go

hopefully you'll crash eventually.
if you do:
10. view>command
11. !analyze -v
12. copy the output from that to a text file and attach it to this bug.

--
alternatively, one thing to check is start>settings>control 
panel>administrative tools>event view>application
view>filter
source: application error

see if you happen to have anything like:
Faulting application seamonkey.exe, version 1.0.0.0, faulting module , version 
0.0.0.0, fault address 0x00000000.

For more information, see Help and Support Center at 
http://go.microsoft.com/fwlink/events.asp.

[that's one of mine for seamonkey, yours would say firefox....], you can sort 
by time or use whatever else from thje ui to try to make finding an incident 
you know of easier.

Comment 9

[Nick Howitt]

It looks like the Microsoft debugger you have pointed to does not support WinME.
There is a link to a WinME debugger at the bottom of
http://www.microsoft.com/whdc/devtools/debugging/installx86.mspx, but this
appears to be radically different from the tool you referred to, and I do not
see any obvious instructions to use it. I am willing to try if you can give me
pointers.

Comment 10

[Nick Howitt]

I have tried changing all the settings ATI suggested. I also reduced the colour
depth from 32 bit to 16 bit, but no change - random crashing coming out of power
saving.
On a completely different tack, I tried changing my usage. If I just load a
number of tabs and leave them, then I had no crashes for 24 hours. If I use the
extension ReloadEvery 0.6.1 with a version number bump (my normal usage,
reloading all 6 pages every 15 minutes), then I often get crashes. Also, instead
of using ReloadEvery, if I use the Continuous Reload bookmarklet from
http://subsimple.com/bookmarklets/collection_navigation.asp I also get the same
crashing. As a last check I found a site, http://www.nu.nl/, with a meta-refresh
every 4 minutes and this has not produced the same crashing so far.

Comment 11

[Nick Howitt]

I've changed the status to invalid as both my hardware setup and software setup have changed and I no longer see the issue.