Bug#597429: [radeon] firefox often crashes whole system

[Jonathan Nieder <jrnieder@gmail.com>]

fixed 597429 linux-2.6/2.6.35~rc4-1~experimental.1
quit

sdfsdfsdf ewfwdfsdf wrote:

> Ok, i tried all the kernels backwards down from 3.1 to 2.6.33.
> The result is the following:
>
> In all kernels -  including 2.6.35 - the bug seems fixed; i couldnt
> reproduce it.

Thanks much.  That includes v2.6.35-rc4, I assume.

> 2.6.33 & 34 had some bugs which prevented further testing: Both did
> not run with activated KMS. And i couldnt even force them to do so.
> Another thing was, that i coulnd not even switch to text console
> (ctrl+alt+F1). I only got a blank screen. But i still could switch
> back to graphical mode again. But there were no compositing effects
> and i couldnt enable them at all. So it seems both kernels have some
> very weird problems too.
>
> If you insist i could send you the logfiles from them.

Yes, I suppose a log from one of them would be useful.

Moving forward:

 - Please also try getting a log from 2.6.32-40 with drm.debug=0x6 on
   the kernel command line.

 - Was this a regression?  Do you remember any older kernel version
   that worked?  (Well, I'm guessing the problems started with KMS,
   but it's worth knowing anyway.)

 - Can you reproduce this without KMS activated?  (I.e., "recovery mode"
   with drm.modeset=0 on the kernel command line, and then starting
   X with the "vesa" driver.)

 - Can you reproduce it with KMS activated but without X?  (I.e.,
   ordinary "recovery mode" on the kernel command line; you might
   need to run "modprobe radeon" to trigger symptoms, not sure if
   it gets loaded automatically.)

 - How about with KMS activated and the X fbdev driver instead of the
   radeon driver?

 - What is your userspace like?  Please attach output from
   "reportbug --template xserver-xorg-video-radeon" with an affected
   version loaded.

 - Any other weird symptoms or clues?

Hope that helps,
Jonathan