Radeon mem map fix (#2): Need regression tests please

[Andreas Haumer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi!

Roland Scheidegger schrieb:
> Andreas Haumer wrote:
> 
>> I'm testing the patch on three different ATI based graphic hardware
>> for a couple of days now and for me the patch works fine. This patch
>>  is even the only way to let Xorg 6.9.0 run on this hardware, as the
>>  original radeon driver (without the patch) seems to be very unstable
>>  (X server freezes in a couple of minutes)
> 
> On which system doesn't it run without the patch? All three?
> 

The original Xorg 6.9.0 radeon driver (without the patch by Benjamin
Herrenschmidt) freezes on the Fujitsu Siemens Celsius H210 notebook
(ATI FireGL Mobility T2 chip) within a few seconds or minutes after
the start of the X server.
On the Fujitsu Siemens Lifebook E2010 (ATI Radeon IGP 330M/340M/350M)
it takes longer, but I also had a few freezes with the original Xorg
6.9.0 server after several hours of work on the device.
On the Asus Pundit-R (ATI Radeon 9100 PRO IGP) I did not see any crash
or X server freeze so far, even without the patch.

On all three devices I have running the patched Xorg 6.9.0 server
for several days now without problems.

Sorry for not beeing precise enough in my last mail!

>> I do not see any strange effects with the mouse cursor or icons.
>> Switching between text and graphics mode as well as between different
>>  X11 sessions works. Watching videos (using xine) works fine, too.
> 
> This is very different to xawtv overlay. xawtv grabdisplay which is
> comparable works too here even with the patch.
> 
Ok. I only could test with xine so far. I do not have any TV hardware.

>> I do not use DRI though, as I get kernel panics (oops) when activating
>> the dri module (this is not radeon specific as I get DRI related
>> kernel panics with other graphic drivers too)
> 
> That's a very big difference - not using dri is almost like using a
> different driver. Not sure how you can stand that though, for me it
> always feels a bit too sluggish. You shouldn't get kernel panics when
> using DRI though obviously (on all 3 systems?).

Yes. All crashes occur at more or less the same place (see
attached file xorg.ksymoops)
It might be some problem with devfs which we use here.
I seem to remember some mails on this list stating that DRI
is not really supported on 2.4 kernels anymore. For the time
being I can live with that. If anyone wants me to try some
patches I can do that, though.

> Your hardware is interesting though and all of them quite different to
> mine (your r100-derived chip is an igp, the igp9100 and M10 are quite
> different beasts too).
> 
>> glxgears gives me about 223 FPS
> 
> glxgears scores aren't all that interesting anyway, but without dri they
> are only measuring cpu performance :-).
> 
> Because I was curious I tried without dri too, and the results were even
> worse. It didn't crash, but I couldn't see enough to even log in with
> kdm, instead I could see fragments of the previous X session, though
> they were "noisy" - obviously a result of having powered off for some
> seconds the pc after switching drivers/changing xorg config...
> So it's not only drm interaction which causes this to not work here.
> 
This is strange.
YMMV, of course, but for me X is quite useable on all hardware
we have here even without DRI (we usually only do software
development, system administration and typical office tasks
here at work, no 3D ego shooters, though ;-)

Regards,

- - andreas

- --
Andreas Haumer                     | mailto:andreas at xss.co.at
*x Software + Systeme              | http://www.xss.co.at/
Karmarschgasse 51/2/20             | Tel: +43-1-6060114-0
A-1100 Vienna, Austria             | Fax: +43-1-6060114-71
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDw67AxJmyeGcXPhERAqe/AJ9WfKpF1es0Dw1Ud+nM2AEsRHUsqwCePITh
dANHdjqL0RBo6YTgfkSrLl8=
=5qxY
-----END PGP SIGNATURE-----
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: xorg.ksymoops
URL: <http://lists.x.org/archives/xorg/attachments/20060110/7497ae1d/attachment.ksh>