FS#31264 - [linux] 3.5.x - 3.7.x ati radeon, modesetting black screen

Attached to Project: Arch Linux
Opened by Alexandre Bique (babali) - Friday, 24 August 2012, 12:02 GMT
Last edited by Tobias Powalowski (tpowa) - Wednesday, 27 February 2013, 13:33 GMT
Task Type Bug Report
Category Kernel
Status Closed
Assigned To Tobias Powalowski (tpowa)
Thomas Bächler (brain0)
Architecture All
Severity High
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 2
Private No

Details

Description:

When I boot if I enable modesetting, then the radeon driver will try to switch to some resolution, which is not supported by my screen, and I can't tell him to use 2560*1600. So I end up with a black screen which is worse than anything.

If I disable modesettings, then I have my console working at boot, and I can start X. But If I use the radeon driver, then black screen, and kernel panic I guess, as the keyboard is not responsive anymore.

So I am actually using xf86-video-vesa-2.3.2-1 but I can't enjoy 2560*1600.

I have this bug since 8 months but I was too lazy to report. Just catalyst-firepro is not working anymore and I'd love to see the free driver working and use it.

Additional info:
* 03:00.0 VGA compatible controller: Advanced Micro Devices [AMD] nee ATI RV620 [FirePro 2260]
* linux-3.5.2-1
* xorg-server-1.12.3.901-1
* xf86-video-vesa-2.3.2-1
* xf86-video-ati-1:6.14.6-1
This task depends upon

Closed by  Tobias Powalowski (tpowa)
Wednesday, 27 February 2013, 13:33 GMT
Reason for closing:  Upstream
Additional comments about closing:  No chance to debug this anymore
Comment by George (gh403) - Monday, 27 August 2012, 20:37 GMT
Do you have PowerXpress or something like that? My laptop does the same sort of thing; I have to just disable the discrete card because it won't work with Linux.
Comment by Alexandre Bique (babali) - Monday, 27 August 2012, 21:18 GMT
I don't know what is PowerXpress, I think that my problem is mostly that the driver switch to a mode unsupported by my screen, do you know if I can tell to radeon driver to set 2560*1600?
Comment by Laurent Carlier (lordheavy) - Tuesday, 28 August 2012, 05:05 GMT Comment by Alexandre Bique (babali) - Tuesday, 28 August 2012, 18:10 GMT
I don't know what strings to pass to <conn>: video=<conn>:2560x1600-24D

I use DisplayPort, and there is two outputs on my video card.
Comment by Alexandre Bique (babali) - Tuesday, 28 August 2012, 18:13 GMT
And why isn't radeon driver able to detect the right settings?
Comment by Laurent Carlier (lordheavy) - Tuesday, 28 August 2012, 18:35 GMT
You can get the differents ouputs and name with:
xrandr
=> DisplayPort-0 disconnected (normal left inverted right x axis y axis)
HDMI-0 disconnected (normal left inverted right x axis y axis)
DVI-0 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 510mm x 290mm

So put the connected ouput in <conn>

Failed screen detection can come from different things:
* buggy driver
* misconfigured driver/package
* broken edid (screen)

Can you post the output of dmesg (with/without nomodeset) ?
Comment by Alexandre Bique (babali) - Wednesday, 29 August 2012, 06:35 GMT
sure as soon as i get to work office.
Comment by Alexandre Bique (babali) - Wednesday, 29 August 2012, 09:32 GMT
Actually as I am using vesa driver, this is the output of xrandr:

xrandr: Failed to get size of gamma for output default
Screen 0: minimum 640 x 400, current 1600 x 1200, maximum 1600 x 1200
default connected 1600x1200+0+0 0mm x 0mm
1600x1200 0.0*
1280x1024 0.0
1152x864 0.0
1024x768 0.0
800x600 0.0
640x480 0.0
720x400 0.0
Comment by Alexandre Bique (babali) - Wednesday, 29 August 2012, 10:02 GMT
The interesting part:

[ 37.079347] [drm] PCIE GART of 512M enabled (table at 0x0000000000040000).
[ 37.079472] radeon 0000:03:00.0: WB enabled
[ 37.079478] radeon 0000:03:00.0: fence driver on ring 0 use gpu addr 0x0000000010000c00 and cpu addr 0xffff88030547cc00
[ 37.112701] [drm] ring test on 0 succeeded in 1 usecs
[ 37.112987] [drm] ib test on ring 0 succeeded in 0 usecs
[ 37.114286] [drm] Radeon Display Connectors
[ 37.114288] [drm] Connector 0:
[ 37.114289] [drm] DP-1
[ 37.114289] [drm] HPD2
[ 37.114291] [drm] DDC: 0x7e60 0x7e60 0x7e64 0x7e64 0x7e68 0x7e68 0x7e6c 0x7e6c
[ 37.114292] [drm] Encoders:
[ 37.114293] [drm] DFP1: INTERNAL_UNIPHY
[ 37.114294] [drm] Connector 1:
[ 37.114295] [drm] DP-2
[ 37.114295] [drm] HPD4
[ 37.114297] [drm] DDC: 0x7e20 0x7e20 0x7e24 0x7e24 0x7e28 0x7e28 0x7e2c 0x7e2c
[ 37.114297] [drm] Encoders:
[ 37.114298] [drm] DFP2: INTERNAL_UNIPHY
[ 37.114455] [drm] Internal thermal controller with fan control
[ 37.114498] [drm] radeon: power management initialized
[ 37.201787] [drm] fb mappable at 0xE0142000
[ 37.201789] [drm] vram apper at 0xE0000000
[ 37.201790] [drm] size 16384000
[ 37.201791] [drm] fb depth is 24
[ 37.201792] [drm] pitch is 10240
[ 37.201899] fbcon: radeondrmfb (fb0) is primary device
Comment by Laurent Carlier (lordheavy) - Wednesday, 29 August 2012, 22:06 GMT
It seems there is still a fglrx module installed:
[ 2.459272] fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel.
[ 2.459274] Disabling lock debugging due to kernel taint
[ 2.460835] fglrx: disagrees about version of symbol wake_up_process
[ 2.460838] fglrx: Unknown symbol wake_up_process (err -22)
[ 2.463461] input: AT Translated Set 2 keyboard as /devices/platform/i8042/serio0/input/input2

Perhaps it's the culprit......
Comment by Alexandre Bique (babali) - Thursday, 30 August 2012, 09:10 GMT
Yes I figured that, but I removed the catalyst-firepro package and rebuilt the initcpio and reboot. I had still the same issue.
Comment by Alexandre Bique (babali) - Thursday, 30 August 2012, 12:56 GMT
lordheavy: here a log without the fglrx, but I have still the black screen.
   dmesg3.xz (13.5 KiB)
Comment by Laurent Carlier (lordheavy) - Thursday, 30 August 2012, 13:19 GMT
Post the log without the two video entries (they seem useless) and with drm.debug=1 int the command line could give more clues
Comment by Alexandre Bique (babali) - Thursday, 30 August 2012, 13:27 GMT
As I don't know how are numbered my outputs, it may be better to let them?

I can reboot with drm.debug=1.
Comment by Laurent Carlier (lordheavy) - Thursday, 30 August 2012, 13:28 GMT
can you also post your xorg log file ?
Comment by Alexandre Bique (babali) - Thursday, 30 August 2012, 13:31 GMT
I did not try to start Xorg yet, as I have a black screen.

Maybe we can solve the kms problem first?
Comment by Laurent Carlier (lordheavy) - Thursday, 30 August 2012, 13:34 GMT
xorg log file also gives informations about the system
Comment by Alexandre Bique (babali) - Thursday, 30 August 2012, 14:00 GMT
Here is the dmesg with drm.debug=1 and no more video= options.
And then I add to reboot because of black screen with kms disabled and I started Xorg with no configuration file. See Xorg.log.
Comment by Laurent Carlier (lordheavy) - Thursday, 30 August 2012, 14:20 GMT
Just talked about your problem on IRC (#radeon) with an Ati developer (Alex Deucher):

[16:10] <agd5f> lordheavy: he's using UMS according to that log
[16:11] <agd5f> probably worth trying KMS on a recent kernel
[16:12] <agd5f> lordheavy: and the dmesg log indicates he also has KMS enabled so that's probably the problem.
[16:12] <agd5f> two drivers both trying to control the same hw

Comment by Alexandre Bique (babali) - Thursday, 30 August 2012, 14:25 GMT
You have to note that I dit two boot:

1) with kms enabled, and I could not start Xorg because I had a blackscreen.
2) I restarted with kms disabled and I started Xorg.
Comment by Alexandre Bique (babali) - Thursday, 30 August 2012, 16:23 GMT
Here is the bug on freedesktop.org https://bugs.freedesktop.org/attachment.cgi
Comment by Laurent Carlier (lordheavy) - Thursday, 30 August 2012, 16:27 GMT Comment by Alexandre Bique (babali) - Tuesday, 04 September 2012, 14:49 GMT
Do we have some nice PKGBUILD to checkout and build a linux kernel from a radeon developer's branch?

Thanks 8)
Comment by Laurent Carlier (lordheavy) - Thursday, 06 September 2012, 12:17 GMT
There is this repo with a precompiled rc kernel:
[miffe]
Server = http://arch.miffe.org/$arch

package name is linux-mainline

Comment by Alexandre Bique (babali) - Thursday, 06 September 2012, 12:23 GMT
Ok thanks, I adapted the linux packages from abs to use http://cgit.freedesktop.org/~agd5f/linux/log/?h=drm-fixes-3.6.
Comment by Tobias Powalowski (tpowa) - Monday, 05 November 2012, 15:07 GMT
Fixed in 3.6.x?
Comment by Alexandre Bique (babali) - Monday, 05 November 2012, 15:12 GMT
Not yet.
Comment by Tobias Powalowski (tpowa) - Wednesday, 23 January 2013, 14:53 GMT
Fixed in 3.7.x?
Comment by Alexandre Bique (babali) - Wednesday, 23 January 2013, 14:58 GMT
No it isn't.
Comment by Tobias Powalowski (tpowa) - Wednesday, 27 February 2013, 11:10 GMT
Status on 3.8?
Comment by Alexandre Bique (babali) - Wednesday, 27 February 2013, 12:47 GMT
I don't know this kernel is still in testing, and I'm quitting my job and so this desktop tomorrow.
Comment by Tobias Powalowski (tpowa) - Wednesday, 27 February 2013, 13:18 GMT
This means you cannot test this anymore?
Comment by Alexandre Bique (babali) - Wednesday, 27 February 2013, 13:26 GMT
yes :(

Loading...