Problem - ATI Radeon X1950XTX + Dell 3007WFPHC = no boot

Ancoron Luciferis ancoron at chaoslayer.de
Thu Dec 10 21:02:41 UTC 2009


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stefan Bader wrote:
> Ancoron Luciferis wrote:
>> Hi *,
>>
>> As the subject says it already I have a problem regarding booting a
>> previously running Kubuntu 9.10 installation.
>>
>> My system:
>>
>> - Gigabyte GA-MA785GM-US2H motherboard
>> - AMD Athlon(tm) II X4 620 Quad-Core
>> - 4GB RAM
>> - ATI Radeon X1950XTX PEG
>>
>> The kernel I tried first is the latest in Karmic: 2.6.31-15.50 (generic,
>> amd64). I also tried the vanilla 2.6.32 without change.
>>
>> The situation is the following:
>>
>> I wanted to change from a dual-headed CRT setup into a single-headed LCD
>> setup using the mentioned Dell 3007WFP-HC as I successfully run another
>> Dell 3007WFP-HC at work with a Radeon HD2400PRO. So I changed the
>> xorg.conf accordingly, turned off the system, plugged out the CRT's and
>> plugged in the Dell. Then I turned on the machine which brought me to
>> the usual grub menu but after that the display remained black and the
>> system doesn't boot any further. The backlight of the display is turned
>> on but nothing (not even a blinking cursor or something) is displayed
>> and the system sits there forever.
>>
>> I can reboot the system using CTRL+ALT+DEL and if I plug off the monitor
>> cable than the kernel continues to boot. When the system is booted
>> without the monitor plugged in I plug it in, restart KDM and voila, all
>> things are fine and the display comes up properly. In addition there is
>> no warning and no error in the logs that would point me to any direction.
>>
>> In the logs I see this:
>>
>> ...
>> [    2.250193] Freeing unused kernel memory: 728k freed
>> [    2.250388] Write protecting the kernel read-only data: 7792k
>> [    2.395052] r8169 Gigabit Ethernet driver 2.3LK-NAPI loaded
>> [    2.395080] r8169 0000:02:00.0: PCI INT A -> GSI 18 ...
>> [    2.395143] r8169 0000:02:00.0: setting latency timer to 64
>> [    2.395176]   alloc irq_desc for 27 on node 0
>> [    2.395179]   alloc kstat_irqs on node 0
>> [    2.395191] r8169 0000:02:00.0: irq 27 for MSI/MSI-X
>> [    2.395660] eth0: RTL8168c/8111c at 0xffffc90000672000, ...
>> [    2.395890] usbcore: registered new interface driver hiddev
>> [    2.400429] input: Microsoft Microsoft 5-Button Mouse with ...
>> [    2.400502] generic-usb 0003:045E:0039.0001: input,hidraw0: ...
>> [    2.400536] usbcore: registered new interface driver usbhid
>> [    2.400538] usbhid: v2.6:USB HID core driver
>> [    2.403855] ohci1394 0000:03:0e.0: PCI INT A -> GSI 22 ...
>> [    2.470131] ohci1394: fw-host0: OHCI-1394 1.1 (PCI): ...
>> [    3.780218] ieee1394: Host added: ID:BUS[0-00:1023] ...
>>
>> ...wait, wait, wait ... nothing happens at all...
>> ...pluggin off the Dell, and whoa:
>>
>> [  178.848332] PM: Starting manual resume from disk
>> [  178.848336] PM: Resume from partition 8:17
>> [  178.848338] PM: Checking hibernation image.
>> [  178.848579] PM: Error -22 checking image file
>> [  178.848583] PM: Resume from disk failed.
>> [  178.879493] EXT4-fs (sda5): mounted filesystem with ...
>> [  183.587543] Adding 3903752k swap on /dev/sda1.  ...
>> [  183.594043] udev: starting version 147
>> [  183.712255] ip_tables: (C) 2000-2006 Netfilter Core Team
>> [  183.872782] lp: driver loaded but no devices found
>> [  183.915753] EDAC MC: Ver: 2.1.0 Dec  3 2009
>> ...
>>
>> Using a different graphics card (a Radeon HD4770) works as expected
>> during boot (although other problems occur later which are caused by the
>> young driver support). And as my problem occurs in a very early stage of
>> the kernel boot process I don't see that it should be a driver problem.
>>
>> Recovery modes work as expected too.
>>
>> Thanx for any help,
>>
>> Ancoron
> 
> Hi Ancoron,
> 
> as the boot is normally done with "quiet splash" on the grub arguments
> you might try to hit left-shift on boot to get to the grub screen and
> then replace the quiet with debug. This will make the dmesg more detailed
> and by that allows to better see where the hang is.
> Experimentally you might also try removing the splash parameter, to see whether
> this has any influence there.
> It still could be a driver problem as some drivers are started from initrd.
> 
> Hope this helps,
> Stefan
> 
> 

Hi Stefan,

thanx for your reply.

As I have a very similar issue at another machine I filed a bug report
in launchpad: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/493834

As I wrote there the "splash" option is causing the boot hang.
Additionally the two kern.log excerpts I provided there from a hung boot
and boot with "splash" option removed yields to a small set of
differences ("diff -w splash-enabled splash-disabled"):

663,665c663,664


<   Magic number: 5:611:889


< ata_piix 0000:00:1f.2: hash matches


< rtc_cmos 00:05: setting system clock to 2009-12-10 14:51:57 UTC
(1260456717)

- ---


>   Magic number: 5:164:940


> rtc_cmos 00:05: setting system clock to 2009-12-10 14:55:33 UTC
(1260456933)

...and...

764a764,773


> mtrr: type mismatch for c0000000,10000000 old: write-back new:
write-combining

> mtrr: type mismatch for c0000000,10000000 old: write-back new:
write-combining

> mtrr: type mismatch for c0000000,10000000 old: write-back new:
write-combining

> mtrr: type mismatch for c0000000,10000000 old: write-back new:
write-combining

> mtrr: type mismatch for c0000000,10000000 old: write-back new:
write-combining

> [drm] Setting GART location based on new memory map


> [drm] Loading RV610 CP Microcode


> [drm] Loading RV610 PFP Microcode


> [drm] Resetting GPU


> [drm] writeback test succeeded in 1 usecs

So the drm gets initialized in both cases:

[drm] Initialized drm 1.1.0 20060810

and also the radeon things seem to get initialized correctly:

[drm] Initialized radeon 1.31.0 20080528 for 0000:01:00.0 on minor 0


And if I only disable the "quiet" option I get some output but as soon
as the progress splash screen should be initialized I see nothing (LCD's
going blank and heavy flickering on CRT but also black, just like a
completely wrong timing).

I hope this helps.


Thanx for any other help,

Ancoron








-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkshYegACgkQHwxOsqv2bG3JZwCeP8HaX5aWBEzZCUGIjWDCUfe4
FqoAoIm/kgNaKkeZkZWsOJ2aKdmeg4LF
=eSHp
-----END PGP SIGNATURE-----




More information about the kernel-team mailing list