Thank you amd, now I can't use Linux

[gradinaruvasile]

Well. I may have found a crude workaround for the freezing issue. It seems that if i completely disable screensaver (it was turning the monitor off with dpms after 4 minutes) and leave the monitor on permanently works. I now have 1 day uptime.

Does it crash while using Chrome? I had the same problem, and it was a problem with Chrome. Disable all hardware acceleration for videos and stuff, reset all the chrome flags to default.

It sounds like it might be, since you mentioned chrome before. Also, check that the DPM is actually working, it isnt 100% there yet.

I use Seamonkey mainly (uses the Firefox engine). Which does have hardware acceleration ("when needed"). But hardware acceleration AFAIK is used when displaying WebGL content. Maybe Chrome uses it for rendering everything or something.
But Chrome cannot display WebGL with fglrx or with radeon on A8-5500 - the accelerated GPU thread crashes every time it attempts. So i ditched it in favor of Seamonkey which works wonderfully with webgl too.

50c is really hot for an idle temp. I have a PhII 955BE that it's load temp is 54c. Idle temp is 36-37c

No no. Its not idle. I meant loaded. Idle is reported 14C, which isnt real, but these chips dont report accurately when idle.

[duby229]

Well. I may have found a crude workaround for the freezing issue. It seems that if i completely disable screensaver (it was turning the monitor off with dpms after 4 minutes) and leave the monitor on permanently works. I now have 1 day uptime.

I use Seamonkey mainly (uses the Firefox engine). Which does have hardware acceleration ("when needed"). But hardware acceleration AFAIK is used when displaying WebGL content. Maybe Chrome uses it for rendering everything or something.
But Chrome cannot display WebGL with fglrx or with radeon on A8-5500 - the accelerated GPU thread crashes every time it attempts. So i ditched it in favor of Seamonkey which works wonderfully with webgl too.

No no. Its not idle. I meant loaded. Idle is reported 14C, which isnt real, but these chips dont report accurately when idle.

Oh OK, Sorry I just misunderstood.

It seems that you have been able to narrow it down somewhat though. That's good news. Are you using xscreensaver? There are other screensaver apps you can use, I wonder if any of them trigger the same bug?

[gradinaruvasile]

Oh OK, Sorry I just misunderstood.

It seems that you have been able to narrow it down somewhat though. That's good news. Are you using xscreensaver? There are other screensaver apps you can use, I wonder if any of them trigger the same bug?

Well, unfortunately i got a hang after 36 hours or so (which is far more than the 4-5 hours i got before).
It occured sometimes after i restarted my browser - which was listed in /sys/kernel/debug/dri/0/clients as active renderer after i started (and stopped) a webgl demo on it right after boot and remained there until i restarted the browser. Now this can be a coincidence of course. Now i reactivated the screensaver and did the demo thing again.

[gradinaruvasile]

Over the weekend i tried disabling screensaver, enabling it. It seems though this doesnt really have any impact - i get hangs anyway. So at this point i have hangs no matter what, idle or not, after a few hours of use (which can be as much as 1.5 days, i wish i know why was that session so long).
I got only one instance of hang while gaming, a few minutes after changing to the performance cpu governor from ondemand (IF thats the trigger).

To sum it up, on my hardware (A8-5500 APU, Gigabyte F2A85X-D3H mobo, 4 GB RAM) and my OS (Debian 64-bit testing, self-compiled kernel from git, mesa from git, xf86-ati from git, libdrm from git) i get random hangs (hard lockup, nothing in logs, computer totally unresponsive, magic keys not working). Tried various kernel compile options regarding schedulers and whatnot, tried kernel command line options regarding iommu, acpi, tried R600_DEBUG=nodma, R600_VA=0 in /etc/environment. Applied the patch mentioned before by agd5f.
Nothing worked so far. Although while running opengl applications this happened only once (after changing cpu schedulers).

@agd5f
I saw that you have some patches for 3.12 (the -wip branch). Can those be of any use in this case?

[pandev92]

For now I'm staying with kernel 3.10 this is more stable( averidge...), I wil try dpm when kernel 3.11.7 will out.

[gradinaruvasile]

I already tried rc7 + the wip 3.12 radeon gcc patches from the agd5f git. It still hangs after a few hours.

[asdfblah]

I already tried rc7 + the wip 3.12 radeon gcc patches from the agd5f git. It still hangs after a few hours.

did you try setting up netconsole? https://wiki.archlinux.org/index.php/Netconsole

[dante]

I tried the opensource drivers with archlinux, chakra linux, ubuntu 13.04 , ubuntu 13.10, averigde mesa, mesa git, kernel 3.9,3.10, 3.11 git....;(
All this distros,reboot suddenly after 30/40 minutes of usage if I use opensource drivers...

With catalyst drivers, by ppa for example, on kubuntu had slow perfomance, flash is soooo slow, vlc xv or mplayer xv are so slow compared to opengl output, and opengl ouptut have artifacts when you switch to fullscren at window mode...
On archlinux , is completely sloooooooooooooooooooooooooow if I use gnome shell, and the shell crash every 2 or 3 hours of use...

Really it's a shame

The opensource driver is the only with good 2d performance but it restart my computer after a while for no reason

Have you tried some distros with long term stable kernel versions like Debian or Slackware ?

[gradinaruvasile]

did you try setting up netconsole? https://wiki.archlinux.org/index.php/Netconsole

I just did. Lets see what happens...

[gradinaruvasile]

..Nothing it seems. It froze while i was playing Dod, but nothing new on the netconsole capture. Will try with other loglevels, lets see if it helps.