Export option AtomBIOS to xrandr for bug tracking?
Hi, My programming skills are quite low but I try to test everyting that may be helpful. We currently have some bug in setting PLL for M82 (RV620): http://bugs.freedesktop.org/show_bug.cgi?id=18016 It's not sure where we are bugged: setting PPL2 or swithing off PLL1. Here goes my idea: If someone could export option of enabling AtomBIOS usage to xrandr I could test both things. Does it make any sense? Does such a exporting AtomBIOS to xrandr need much of work? Or is this completly stupid idea? ;) -- Rafał Miłecki -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On May 12, 09 13:18:56 +0200, Rafał Miłecki wrote:
Here goes my idea: If someone could export option of enabling AtomBIOS usage to xrandr I could test both things. Does it make any sense? Does such a exporting AtomBIOS to xrandr need much of work? Or is this completly stupid idea? ;)
It's not stupid, but I'm unsure ATM whether this is actually doable.
Somebody would need to dig into the code and see whether the
initialization phase relies on the setting (and does something different
in each case) or not.
Implementing the RandR property would be a breeze.
Matthias
--
Matthias Hopf
2009/5/12 Matthias Hopf
On May 12, 09 13:18:56 +0200, Rafał Miłecki wrote:
Here goes my idea: If someone could export option of enabling AtomBIOS usage to xrandr I could test both things. Does it make any sense? Does such a exporting AtomBIOS to xrandr need much of work? Or is this completly stupid idea? ;)
It's not stupid, but I'm unsure ATM whether this is actually doable. Somebody would need to dig into the code and see whether the initialization phase relies on the setting (and does something different in each case) or not.
If my logic is right, we could... just test that :) Currently bug appears after following 3 steps: 1) Starting X without AtomBIOS 2) Enabling DVI-D without AtomBIOS 3) Disabling PANEL without AtomBIOS So to test if initialization does matter in case of this bug we could: 1) Starting X without AtomBIOS 2) Enabling DVI-D _with_ AtomBIOS 3) Disabling PANEL _with_ AtomBIOS That test would let us know if initialization matters or not. In case this would work, we could try doing 2 or 3 without AtomBIOS. -- Rafał Miłecki -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On May 12, 09 23:02:49 +0200, Rafał Miłecki wrote:
It's not stupid, but I'm unsure ATM whether this is actually doable. Somebody would need to dig into the code and see whether the initialization phase relies on the setting (and does something different in each case) or not.
If my logic is right, we could... just test that :)
Ok, try the patchbomb attached - I found a number of randr property bugs
while coding this, so even if it does not work at all, not all is in
vain ;-)
But be aware that I get an assert in some cases, e.g.
../../src/rhd_crtc.c:586: DxModeRestore: Assertion '!RHD_CHECKDEBUGFLAG(rhdPtr, VGA_SETUP)' failed.
The _AtomBIOS property is always added to the *first* output (in my case
it's the TV...), because RandR has no support whatsoever for per-CRTC,
per-GPU, or per-board properties... :-(
Good luck
Matthias
--
Matthias Hopf
On May 13, 09 16:49:21 +0200, Matthias Hopf wrote:
Ok, try the patchbomb attached - I found a number of randr property bugs while coding this, so even if it does not work at all, not all is in vain ;-)
Rafal, have you had a chance at looking into this?
Matthias
--
Matthias Hopf
W dniu 5 czerwca 2009 18:39 użytkownik Matthias Hopf
On May 13, 09 16:49:21 +0200, Matthias Hopf wrote:
Ok, try the patchbomb attached - I found a number of randr property bugs while coding this, so even if it does not work at all, not all is in vain ;-)
Rafal, have you had a chance at looking into this?
Yes, you can check my last comments in bug report: http://bugs.freedesktop.org/show_bug.cgi?id=18016 I now know that using AtomBIOS command SetPixelClock for powering down PANEL is enought. We can use registers access to start X, to enable DVI-D_1 and switch to AtomBIOS for powering off PANEL only. Right now I'm tracing SetPixelClock in my AtomBIOS (using AtomDis) to know exactly what it does differently. I dumped registers for this tracking 3 minues ago, I'm tracing command right now. -- Rafał Miłecki -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org
On Jun 05, 09 19:05:03 +0200, Rafał Miłecki wrote:
W dniu 5 czerwca 2009 18:39 użytkownik Matthias Hopf
napisał: On May 13, 09 16:49:21 +0200, Matthias Hopf wrote:
Ok, try the patchbomb attached - I found a number of randr property bugs while coding this, so even if it does not work at all, not all is in vain ;-) Rafal, have you had a chance at looking into this?
Yes, you can check my last comments in bug report: http://bugs.freedesktop.org/show_bug.cgi?id=18016
So switching atombios usage on the fly somehow works for you? Do you think this is of general use (so we should commit all patches), or should this remain a patch? In the later case I would commit those parts that are bug fixes.
Right now I'm tracing SetPixelClock in my AtomBIOS (using AtomDis) to know exactly what it does differently. I dumped registers for this tracking 3 minues ago, I'm tracing command right now.
Sounds like you're having fun ;-P
Happy weekend
Matthias
--
Matthias Hopf
participants (2)
-
Matthias Hopf
-
Rafał Miłecki