|
Description
Ludovic Rousseau
2007-06-16 04:36:22 UTC
This is due to a bug in pll computation, i really don't see how it sneak in in the first place anyway current git master should not exhibit such things. I just used "git pull" (latest commit is 3708713e0510a424fb70edfd92c0e84f69377932) to have a up-to-date version and the situation is worse. The image is still slanted (included in the X cursor) and the image is very unstable (and more unstable when I move the mouse). How can I help? Please attach xorg log (do not paste it). And grab radeondump git://people.freedesktop.org/~glisse/radeondump log in throught ssh while avivo is runing and issue radeondump -d avivo relaunch x server with fglrx ddx and issue radeondump -d fglrx attach here the two generated by this two command. Created attachment 10343 [details]
output of "X -verbose" with avivo driver
Created attachment 10344 [details]
output of "X -verbose" with driver fglrx
Created attachment 10345 [details]
radeondump with avivo driver running
Created attachment 10346 [details]
radeondump when fglrx driver running
Created attachment 10347 [details]
radeondump with fglrx running after stopping the avivo driver
The display is corrupted the same way I have with the avivo driver.
The fglrx driver does not "repair" the problem created by the avivo.
When I quit the X server using the avivo driver (using control-alt-backspace) the screen is completely corrupted (but stable) and I do not go back to a normal text console. I also tested with version 5e73799265df9e6984a241fa593902b52252c162. The bug is I describe here is still present. the radeondump generated files I attached today where generated using version 3708713e0510a424fb70edfd92c0e84f69377932 from yesterday 15 June. Scanout buffer need to be 256bits aligned (have to check with depth 16). I will commit latter a patch to fix this and other issues. git commit 8d01c28844dd5c434a648dfa7ec4da798224a8cb should have fixed your issue please test with git master and report. I also have a MacBook Pro with what I believe to be the same video card (because it also prints out "PCI:*(1:0:0) ATI Technologies Inc M56P [Radeon Mobility X1600] rev 0, Mem @ 0x80000000/28, 0x98300000/16, I/O @ 0x3000/8") and I don't have these problems with 8d01c28844dd5c434a648dfa7ec4da798224a8cb. (In reply to comment #12) > I also have a MacBook Pro with what I believe to be the same video card > (because it also prints out "PCI:*(1:0:0) ATI Technologies Inc M56P [Radeon > Mobility X1600] rev 0, Mem @ 0x80000000/28, 0x98300000/16, I/O @ 0x3000/8") and > I don't have these problems with 8d01c28844dd5c434a648dfa7ec4da798224a8cb. > What matter here is the screen size. I myself got a mbp but with 15" screen. Created attachment 10377 [details]
X -verbose output
I upgraded the driver to commit b9771c7e26a1dbdcb7dd5cf3c525dad8f196eb6e and my video card is no more recognised :-(
Here is lspci -v of my card:
01:00.0 VGA compatible controller: ATI Technologies Inc M56P [Radeon Mobility X1600] (prog-if 00 [VGA])
Subsystem: Apple Computer Inc. Unknown device 0080
Flags: bus master, fast devsel, latency 0, IRQ 169
Memory at 80000000 (32-bit, prefetchable) [size=256M]
I/O ports at 3000 [size=256]
Memory at 98300000 (32-bit, non-prefetchable) [size=64K]
Expansion ROM at 98320000 [disabled] [size=128K]
Capabilities: [50] Power Management version 2
Capabilities: [58] Express Legacy Endpoint IRQ 0
Capabilities: [80] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0
Enable-
(In reply to comment #13) > (In reply to comment #12) > > I also have a MacBook Pro with what I believe to be the same video card > > (because it also prints out "PCI:*(1:0:0) ATI Technologies Inc M56P [Radeon > > Mobility X1600] rev 0, Mem @ 0x80000000/28, 0x98300000/16, I/O @ 0x3000/8") and > > I don't have these problems with 8d01c28844dd5c434a648dfa7ec4da798224a8cb. > > > > What matter here is the screen size. I myself got a mbp but with 15" screen. Ah, me too. Created attachment 10396 [details]
X -verbose output
I updated libpciaccess to commit a7faac3d2dc49a2130906fbeea3298f32bb9349f and I dot not have the "No devices detected." error any more.
I now have a completely black screen. I can use Ctrl-Alr-Backspace to kill the X server and go back to text mode. No crash but just a black screen.
I will send the result of radeondump tomorrow when I have access to a remote computer.
Created attachment 10399 [details]
radeondump -d
radeondump when using the avivo driver and a black screen
Created attachment 10400 [details]
radeondump
radeondump of the avivo driver black screen after moving the mouse
Created attachment 10401 [details]
radeondump of fglrx driver
radeondump of fglrx driver
Created attachment 10402 [details]
radeondump of the fglrx driver after moving the mouse
Created attachment 10403 [details]
radeondump of with avivo driver executed after fglrx
After that I moved the mouse and retarsted radeondump. But the computer crashed when executing radeondump.
Created attachment 10420 [details]
xorg.conf for avivo driver
I upgraded xf86-video-avivo to git commit 9decf97be5f1aa5ae2001160fcc726236b398104 and I still have a black screen.
I tried with different DefaultDepth.
- 8 : not supported
- 16 : black screen
- 24 : black screen
- 32 : non supported
What else can I do?
Let's checkout you got everythings right: -cold boot -no fglrx kernel module -with apple bootcamp -lastest git avivo (please provide lastest xorg log) And you got black screen ? (In reply to comment #23) > Let's checkout you got everythings right: > -cold boot yes > -no fglrx kernel module yes > -with apple bootcamp I do not use bootcamp. But I upgraded my firmware to have a legacy BIOS. I use refit + lilo to boot a Debian official kernel 2.6.18-4-686. > -lastest git avivo (please provide lastest xorg log) yes. From git-log: commit 9decf97be5f1aa5ae2001160fcc726236b398104 Author: Jerome Glisse <glisse@freedesktop.org> Date: Fri Jun 22 16:41:04 2007 +0200 avivo: post randr 1.2 cleanup + always assignd LFP to TMDS2 commit 2440ac567ce721378395c341cf0322e383bb4888 Author: Jerome Glisse <glisse@freedesktop.org> Date: Fri Jun 22 15:47:39 2007 +0200 avivo: LFP should always be driven by second TMDS. > And you got black screen ? yes :-( In the output of X (also available at https://bugs.freedesktop.org/attachment.cgi?id=10396) I see: (EE) Avivo(0): Couldn't initialise I2C bus for DVI-I connector 0 Maybe the driver sends the video signal to the DVI connector or something like that. Please pull lastest git, launch and attach new xorg.log (i made the driver bit more verbose in order to track done this problem). Also once X server is running with avivo do a single radeondump -d avivo and attach output. Created attachment 10423 [details]
output of X -verbose
xf86-video-avivo version 9f40535f75b3dfb08efb6121294377e758070e56
Created attachment 10424 [details]
radeondump -d from text console
I don't know if that is important. My text console is using vga=794 (1280x1024 x64k colors).
from the X log I have:
(==) ServerLayout "Default Layout"
(**) |-->Screen "Default Screen" (0)
(**) | |-->Monitor "Écran générique"
(**) | |-->Device "Carte vidéo générique"
(**) |-->Input Device "Generic Keyboard"
(**) |-->Input Device "Configured Mouse"
(**) |-->Input Device "Synaptics Touchpad"
[...]
(II) Avivo(0): I2C bus "DVI-I connector 0" initialized.
(II) Avivo(0): Output DVI-I using monitor section Écran générique
(II) Avivo(0): added DVI-I connector 0 (0x7E40)
(II) Avivo(0): I2C bus "LFP connector 2" initialized.
(II) Avivo(0): Output LFP has no monitor section
(II) Avivo(0): added LFP connector 2 (0x7E60)
(EE) Avivo(0): Couldn't initialise I2C bus for DVI-I connector 0
"Écran générique" is the monitor I have configured in /etc/X11/xorg.conf:
[...]
Section "Monitor"
Identifier "Écran générique"
Option "DPMS"
EndSection
Section "Screen"
Identifier "Default Screen"
Device "Carte vidéo générique"
Monitor "Écran générique"
DefaultDepth 24
[...]
So maybe the error about DVI-I connector 0 is important?
i need the radeondump while xorg with avivo is running so do it from ssh. A dump while you are in the console do not give any usefull informations for finding out what is wrong. Created attachment 10428 [details]
radeondump while avivo driver is running
Sorry. I forgot to attach the file.
I pushed a patch which should properly set sync polarity. So, does lastest git make any change for you ? Please also attach lastest Xorg output. Created attachment 10479 [details] X -verbose output version: commit fc0fd1494feeaa48703ca47597d865d2f93c56cb Author: Jerome Glisse <glisse@freedesktop.org> Date: Wed Jun 27 00:34:07 2007 +0200 avivo: print horizontal & vertical sync polarity flags. No progress. I still have a black screen. I noticed the line: VirtualX,Y 1680, 1680 in the logs. I guess the second 1680 should be 1050 I also notice that xorg/avivo_crtc.c uses "vsc %di" instead of "vsc %d" (extra i from a vi user?) (In reply to comment #32) > I noticed the line: > VirtualX,Y 1680, 1680 > in the logs. I guess the second 1680 should be 1050 That is normal its in case of rotation. > I also notice that xorg/avivo_crtc.c uses "vsc %di" instead of "vsc %d" (extra > i from a vi user?) I am vi newbie :) I changed the way i compute PLL which should be better now can you try with lastest git and do radeondump if this still does not work. Created attachment 10482 [details]
X -verbose
YES. IT WORKS!
Scrolling in galeon is very slow (but not in gnome-terminal) but that is another problem.
Thansk a lot
Ok, for performance i will soon add shadow fb which will match performance of the vesa driver, and work for EXA acceleration will likely give a boost too. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.