Thinkpad T60 [1680x1050], ATIx1400, no X on Gutsy, Hardy

Bug #184992 reported by harry.mangalam
2
Affects Status Importance Assigned to Milestone
linux-restricted-modules-2.6.24 (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

When using 8.45.4 driver direct from ATI ( Build Date: Jan 16 2008 10:42:57). This happens on both of the following distros (both on a Thinkpad T60, ATI Technologies Inc Radeon Mobility X1400)

===== separator =====

1 - on a dist-upgrade from feisty to gutsy, using kernel2.6.22-14-generic, the login screen comes up fine, but when I try to log on, the screen goes black then drops me back to the graphical login screen. When going to console. Running startx, X will come up briefly (can see the X cursor on the grey background, but then dies back to the console. When that happens, the screen reports some XKEYBOARD/xkbcomp errors (which explicitly say they aren't fatal), but those are the only errors showing up. I'm not really getting any errors now, except for the ones mentioned below.

The log has given me a number of errors, none of which appear to be fatal, but clearly something is.

Currently, the warnings or errors from X are as follows:

(WW) The directory "/usr/share/X11/fonts/misc" does not exist.
(WW) The directory "/usr/share/X11/fonts/100dpi" does not exist.
(WW) The directory "/usr/share/X11/fonts/75dpi" does not exist.
(WW) The directory "/usr/share/X11/fonts/Type1" does not exist.
(WW) `fonts.dir' not found (or not valid) in "/var/lib/defoma/fontconfig.d/a".
(WW) Including the default font path /usr/share/fonts/X11/misc,/usr/share/fonts/X11/cyrillic,/usr/share/fonts/X11/100dpi/:unscaled,/usr/share/fonts/X11/75dpi/:unscaled,/usr/share/fonts/X11/Type1,/usr/share/fonts/X11/100dpi,/usr/share/fonts/X11/75dpi,/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType.
(WW) fglrx(0): GetVBEMode failed
(WW) fglrx(0): board is an unknown third party board, chipset is supported
(WW) fglrx(0): Only one display is connnected,so single mode is enabled
(WW) fglrx(0): Option "UseFBDev" is not used
(WW) fglrx(0): Option "VendorName" is not used
(WW) fglrx(0): Option "ModelName" is not used

so no Errors and the warnings do not seem to be all that horrible.

Some errors in the past dealt with not being able to load 'glcore' and 'font1' modules, but I've commented those out without complaint.

syslog gives me the following fglrx-related errs

 [fglrx:firegl_free_mutex] *ERROR* mutex id 0x20000008 not found in mutex list
 [fglrx:firegl_init_pcie] *ERROR* Invalid GART type 0.
 [fglrx:firegl_init_pcie] *ERROR* Invalid GART type 0.

but then on rerunning startx, I only get these in a repeating fglrx message block:

[fglrx] PCIe has already been initialized. Reinitializing ...
[fglrx] Reserve Block - 0 offset = 0X7ffb000 length = 0X5000
[fglrx] Reserve Block - 1 offset = 0X0 length = 0X1000000
[fglrx] Reserve Block - 2 offset = 0X7fbb000 length = 0X40000
[fglrx] interrupt source 10000000 successfully enabled
[fglrx] enable ID = 0x00000008
[fglrx] Receive enable interrupt message with irqEnableMask: 10000000
[fglrx] interrupt source 10000000 successfully disabled!
[fglrx] enable ID = 0x00000000
[fglrx] Receive disable interrupt message with irqEnableMask: 10000000; dwIRQEnableId:

===== separator =====

2 - on Hardy (apt-get current as of Jan 21, 2008). (same result with the fglrx patch released today [fglrx 8.43.2] as well as the brand new one direct from ATI [8.45.4] ).
The conditions are just about identical: Same behavior for the initial login, then when I try to login via console, I get a brief flicker of the base X11 screen (movable X cursor for a few seonds and then a crash back to the console.

I have the following additions to my xorg.conf based on other postings:
-----
 Section "Extensions"
         Option "Composite" "0" # or "Disable"
 EndSection

 Section "ServerFlags"
         Option "AIGLX" "off"
 EndSection
 -----

With those additions, the warnings or errors from X are as follows:

(WW) `fonts.dir' not found (or not valid) in "/var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType".
(WW) fglrx(0): GetVBEMode failed
(WW) fglrx(0): board is an unknown third party board, chipset is supported
(WW) fglrx(0): Only one display is connnected,so single mode is enabled
(EE) fglrx(0): === [swlDalHelperAddCustomizeMode] === CWDDEDI_DisplayGetSetModeTimingOverride failed: 2
(WW) fglrx(0): could not detect X server version (query_status=-3)
(EE) fglrx(0): [drm] failed to remove DRM signal handler

The last EE is embedded in the last few lines of the log:
...
(II) fglrx(0): Shutdown CMMQS
(EE) fglrx(0): [drm] failed to remove DRM signal handler
(II) fglrx(0): [drm] removed 1 reserved context for kernel
(II) fglrx(0): [drm] unmapping 8192 bytes of SAREA 0x35000 at 0xb7fb0000
(II) fglrx(0): [drm] Closed DRM master.
FreeFontPath: FPE "/usr/share/fonts/X11/misc" refcount is 2, should be 1; fixing.

Neither dmesg nor syslog has any indications of errors.

Revision history for this message
harry.mangalam (hjm) wrote :

Solved (mostly). Sorry about the confusion. This bug (?) apparently was caused by a misconfiguration somewhere in the depths of my '.kde' directory tree. My kmail had begun to act strangely which is why I tried to upgrade to gutsy. The mistake was revealed when I created a new user and logged in to test this possibility and all of a sudden, X started to work perfectly. I experimented with copying various suspicious files and directories and the .kde dir caused the glitch. Somewhere in there is a file that corrupts X, but I haven't found it yet.

ALSO!! Power suspend-to-RAM / recover now works work, and trying to run google earth causes a crash, but I'll deal with that later. Just glad to be seeing my laptop alive again. The Synaptics touch pad works as it's supposed to.

Changed in linux-restricted-modules-2.6.24:
status: New → Invalid
Revision history for this message
harry.mangalam (hjm) wrote :

This problem reappeared recently with Hardy and the current situation has nothing to do with X, but DOES have to do with the permissions in /tmp. Creating a new user which solved the previous problem also creates new entries in /tmp for that user and that would have addressed a possible permissions problem with starting X and KDE as that user.

In the current case, with Hardy, after about 3 weeks without reboot on a laptop with multiple suspend/wake sessions a day, the battery finally ran out and the machine rebooted. X would not restart, so I went thru 2 hours of xorg.conf gymnastics, almost all of which gave the same error as above:

(EE) fglrx(0): [drm] failed to remove DRM signal handler

This error message is almost Microsoftian in its passive-aggressively accurate misdirection.

Whatever happened, the /tmp dir was marked root write only, which prevented any user from writing to it (and therefore I'm guessing failed to remove a previously written file somewhere in the /tmp tree form getting removed). So X refused to start up as that user but would start as root.

The solution was to make /tmp writable for everyone.

I still don't have the exact file that was at fault, but the fix is easy enough:

sudo chmod ugo+rwx /tmp

Best
hjm

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.