/dev/dri/card0 sometimes vanishes when I exit Xorg (Radeon RS780)

/dev/dri/card0 sometimes vanishes when I exit Xorg (Radeon RS780)

[Markus Trippelsdorf]

Hi,

since a few weeks the /dev/dri/card0 entry sometimes vanishes when I
exit Xorg. When I want to start Xorg again I get the following error
message:

...
[ 12711.547] (II) [KMS] Kernel modesetting enabled.
[ 12711.548] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support
[ 12711.548] (II) RADEON(0): Creating default Display subsection in Screen section
	"Default Screen Section" for depth/fbbpp 24/32
[ 12711.548] (==) RADEON(0): Depth 24, (--) framebuffer bpp 32
[ 12711.548] (II) RADEON(0): Pixel depth = 24 bits stored in 4 bytes (32 bpp pixmaps)
[ 12711.548] (==) RADEON(0): Default visual is TrueColor
[ 12711.548] (==) RADEON(0): RGB weight 888
[ 12711.548] (II) RADEON(0): Using 8 bits per RGB (8 bit DAC)
[ 12711.548] (--) RADEON(0): Chipset: "ATI Radeon HD 3300 Graphics" (ChipID = 0x9614)
[ 12711.548] (II) RADEON(0): PCI card detected
[ 12711.940] (EE) RADEON(0): [drm] Failed to open DRM device for pci:0000:01:05.0: No such file or directory
[ 12711.940] (EE) RADEON(0): Kernel modesetting setup failed
[ 12711.940] (II) UnloadModule: "radeon"
[ 12711.940] (EE) Screen(s) found, but none have a usable configuration.
[ 12711.940] 
Fatal server error:
[ 12711.940] no screens found

I have to manually run "mdev -s" to restore the /dev/dri/card0 device.

This happens with an Radeon RS780 and Xorg version 1.13.0 running the
latest git kernel.

/sys/class/drm/card0/device/power_profile is set to "auto"

Is this a known problem? Are there any workarounds?

-- 
Markus

Re: [Solved] /dev/dri/card0 sometimes vanishes when I exit Xorg (Radeon RS780)

[Markus Trippelsdorf]

On 2012.09.29 at 20:38 +0200, Markus Trippelsdorf wrote:
> since a few weeks the /dev/dri/card0 entry sometimes vanishes when I
> exit Xorg. When I want to start Xorg again I get the following error
> message:
> 
> ...
> [ 12711.547] (II) [KMS] Kernel modesetting enabled.
> [ 12711.548] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support
> [ 12711.548] (II) RADEON(0): Creating default Display subsection in Screen section
> 	"Default Screen Section" for depth/fbbpp 24/32
> [ 12711.548] (==) RADEON(0): Depth 24, (--) framebuffer bpp 32
> [ 12711.548] (II) RADEON(0): Pixel depth = 24 bits stored in 4 bytes (32 bpp pixmaps)
> [ 12711.548] (==) RADEON(0): Default visual is TrueColor
> [ 12711.548] (==) RADEON(0): RGB weight 888
> [ 12711.548] (II) RADEON(0): Using 8 bits per RGB (8 bit DAC)
> [ 12711.548] (--) RADEON(0): Chipset: "ATI Radeon HD 3300 Graphics" (ChipID = 0x9614)
> [ 12711.548] (II) RADEON(0): PCI card detected
> [ 12711.940] (EE) RADEON(0): [drm] Failed to open DRM device for pci:0000:01:05.0: No such file or directory
> [ 12711.940] (EE) RADEON(0): Kernel modesetting setup failed
> [ 12711.940] (II) UnloadModule: "radeon"
> [ 12711.940] (EE) Screen(s) found, but none have a usable configuration.
> [ 12711.940] 
> Fatal server error:
> [ 12711.940] no screens found
> 
> I have to manually run "mdev -s" to restore the /dev/dri/card0 device.

It turned out that this is apparently caused by local voltage spikes of my
power line. From time to time I see this message in dmseg:

hub 4-0:1.0: port 3 disabled by hub (EMI?), re-enabling...
usb 4-3: USB disconnect, device number 2
usb 4-3: new full-speed USB device number 3 using ohci_hcd

It it exactly at this point that /dev/dri/card0 vanishes.
Guess I will buy an UPS in the next days...

-- 
Markus