FC13 2..33.5-124.fc13.x86_64 locks up on boot

[dpawson]

I've an Nvidia graphics card, and with fc12 (32 bit) I had the regular dance
each time the kernel was updated. Since moving to FC13 that was (up until today)
apparently resolved. I *think* the system is using the nouveau driver.

yum update last night, booted this morning and the boot sequence hangs
at 'starting atd' which I think is the predecessor to starting X11, i.e. exactly
the same point it used to hang with FC12 when the livna drivers for the graphics
card were out of sync with the kernel. Guessing it was a similar problem, I
rebooted to mode 3, ran yum update but this time no joy. It still hangs
at 'starting atd'.

Any suggestions please?

TIA DaveP

[Keldorn]

atd is daemon that runs "at" commands, you can easily disable it.
chkconfig atd off
and i hope everything will be ok.
To prevent "dances" when kernel is updated try to install dkms package/

[dpawson]

Originally Posted by Keldorn

atd is daemon that runs "at" commands, you can easily disable it.
chkconfig atd off
and i hope everything will be ok.
To prevent "dances" when kernel is updated try to install dkms package/

From previous experience, I think atd is fine. It is just the last app that shows in
the boot sequence?

In the boot sequence the X server
is next 'in the queue' to start, which is what is failing?

Tried disabling atd, just hangs at the previous 'last' app load.

Tks for the reply

DaveP

[Keldorn]

Try to see what in the /var/log/Xorg.0.log

[dpawson]

Originally Posted by Keldorn

Try to see what in the /var/log/Xorg.0.log

No error messages - though no 'timestamped' info!
Assuming the end of the file is most recent,
it all seems pretty normal, a dump of modelines?
All from Nouveau.

Hundreds of lines earlier I see
AIGLX error: dlopen of /usr/lib64/dir/nouveau_dri.so failed ... no such file

True, no such file.

Yet the log shows loading modelines thereafter?

Any help?

DaveP

[Keldorn]

dmesg as i remeber will show you loading modules. Seems to me, that you had to recompile / reinstall nvidia modules, yesterday was xorg update (may be i am wrong)....

[dpawson]

Originally Posted by Keldorn

dmesg as i remeber will show you loading modules. Seems to me, that you had to recompile / reinstall nvidia modules, yesterday was xorg update (may be i am wrong)....

Yes.. except I'm using nouveau rather than the nvidia code?
I used to do that for 12. Not necessary (as yet) with fc13?

I haven't actively installed any drivers for nvidia.


DaveP

[Keldorn]

Ah, sorry. Lets try to do it other way. Boot into runlevel 3 and try to run startx. Also, you can switch to diffrent tty using ctrl+alt+f<number>

[dpawson]

Originally Posted by Keldorn

Ah, sorry. Lets try to do it other way. Boot into runlevel 3 and try to run startx. Also, you can switch to diffrent tty using ctrl+alt+f<number>

Just locks up.

Tail end of dmesg shows
[drm] nouveau nnnn allocating fifo number 2
....
None of which show any error messages?

DaveP

---------- Post added at 12:17 PM CDT ---------- Previous post was at 12:13 PM CDT ----------

Originally Posted by Keldorn

Ah, sorry. Lets try to do it other way. Boot into runlevel 3 and try to run startx. Also, you can switch to diffrent tty using ctrl+alt+f<number>

In the tty I used for 'startx'

(EE) No input driver/identifer specified (ignoring)
3 times
4 messages relating to GNOME_KEYRING_CONTROL

^c out of that and it shows
waiting for X server to shutdown gonme-settings-daemon: Fatal IO error 11 (Resource temp unavailable o X srever 0:0

No graphics though.

DaveP

---------- Post added at 02:04 PM CDT ---------- Previous post was at 12:17 PM CDT ----------

Originally Posted by Keldorn

dmesg as i remeber will show you loading modules. Seems to me, that you had to recompile / reinstall nvidia modules, yesterday was xorg update (may be i am wrong)....

If xorg was in yesterdays update, perhaps that is (again) the reason for
the failure? Perhaps nouveau needs the update as the nvidia--kmod package
used to do?

Not sure which repo nouveau comes from, but it seems livna is down today.

DaveP

[Keldorn]

livna has gone. rpmfusion repo instead of it, and it is working fine.
may be you should try to get backtrace of startx command?
I haven't more ideas about your problem now

[dpawson]

OK, thanks Keldorn.
I've removed livna.



Dave

---------- Post added at 04:53 PM CDT ---------- Previous post was at 02:39 PM CDT ----------

OK, thanks Keldorn.
I've removed livna.



Dave

---------- Post added at 04:55 PM CDT ---------- Previous post was at 04:53 PM CDT ----------

Quite confused.
How to tell which video driver I'm using?
I see the kmod-nvidia-... .x86_64 package is available. (I don't have it installed)
I do have xorg-x11-drv-nouveau 0.0.16 installed.

Should I infer that I am using the nouveau driver?
hence, as with FC12, that needs an update to match the nvidia card?
Is that what I should be waiting for?

With the kmod package installed, I used to be able to revert
to an older kernel and X server worked again.
Two old kernels still won't start the x server so something is different
from previously?


TIA Dave

[Keldorn]

Please show
ls /etc/X11

[dpawson]

Originally Posted by Keldorn

Please show
ls /etc/X11

applink fontath.d prefdm xinit Xmodmap xorg.conf.d Xresources

[dpawson]

Ah well.
Thanks for your help Keldorn.
I tried to log it as a bug at Fedora Bugzilla, but failed to make sense of the login sequence.

I guess I'll wait another 24 hours, see if anything else turns up, then try another distro.

Dave

---------- Post added at 02:48 PM CDT ---------- Previous post was at 08:44 AM CDT ----------

Problem .. not solved, worked around.
Installed an ATI 4770 Graphics card.
Booted sweetly, I'm back in business.
Tried Debian stable. Full install, Guess what? No graphics. Sigh.
Seems Nvidia is not flavour of the month, or Fedora doesn't like them