Hard Lockups

[trotter1985]

I've tried both the 64bit and 32bit versions of Fedora 13 -Alpha
and keep getting hard lockups - where a reboot is the only
escape. I have an NVidia 6800 graphics card, and no doubt
that is the problem. I would disable the Nouveau driver
and build the NVidia kernel modules, or try something else
if I could do anything, but just touching a menu causes the
screen to freeze. Any ideas how to get past first base?

trotter1985

[jvillain]

Is this a brand new install or upgrade? If it is an upgrade start by booting your last known good kernel. After that try booting into single user mode and see if it is stable. While your in single user mode go look at your logs and see if there is any thing to indicate why things locked up.

When you say reboot are we talking ALT+CTRL+DEL works or are we talking about the power button is the only way? Can you try sshing in from another machine. If so you can get a better look at who is gumming up the works.

[newiLuvatar]

since a few days, I'm also getting screen freezes. strange thing is this also happens when using the live cd. windows is running fine, so I guess the hw is not the problem.
before a freeze, I get lots and lots of these:

Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): Enabling debugging
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): Changing user:group to gdm:gdm
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 1: signum=4 (nil)
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 4 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 2: signum=7 (nil)
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 7 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 3: signum=11 (nil)
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 11 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 4: signum=6 (nil)
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 6 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 5: signum=5 (nil)
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 5 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 6: signum=15 0x406090
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 15 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 7: signum=2 0x406090
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 2 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 8: signum=8 0x406090
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 8 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 9: signum=1 0x406090
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 1 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Adding handler 10: signum=10 0x406090
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSignalHandler: Registering for 10 signals
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmManager: GDM starting to manage displays
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmLocalDisplayFactory: Reserving X display: 0
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplayStore: Adding display /org/gnome/DisplayManager/Display1 to store
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplay: Managing display: /org/gnome/DisplayManager/Display1
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplay: manage display
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplay: Preparing display: /org/gnome/DisplayManager/Display1
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplay: prepare display
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplayAccessFile: creating xauth directory /var/run/gdm/auth-for-gdm-XXXXXX
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplayAccessFile: chowning /var/run/gdm/auth-for-gdm-k5FnQs to 42:478
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplayAccessFile: creating /var/run/gdm/auth-for-gdm-k5FnQs/database
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplayAccessFile: chowning /var/run/gdm/auth-for-gdm-k5FnQs/database to 42:478
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmDisplayAccessFile: chmoding /var/run/gdm/auth-for-gdm-k5FnQs to 0711
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmLocalDisplayFactory: static display status changed: 1
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSlaveProxy: Running command: /usr/libexec/gdm-simple-slave --display-id /org/gnome/DisplayManager/Display1 --force-active-vt
Apr 20 19:19:53 r4pt0r gdm-binary[1613]: DEBUG(+): GdmSlaveProxy: Started slave with pid 1653
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): Enabling debugging
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 1: signum=15 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 15 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 2: signum=2 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 2 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 3: signum=4 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 4 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 4: signum=7 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 7 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 5: signum=8 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 8 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 6: signum=1 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 1 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 7: signum=11 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 11 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 8: signum=6 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 6 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 9: signum=10 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 10 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Adding handler 10: signum=12 0x4073b0
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSignalHandler: Registering for 12 signals
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSlave: Registering /org/gnome/DisplayManager/Slave1
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSlave: starting slave
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSlave: Starting slave
Apr 20 19:19:53 r4pt0r gdm-simple-slave[1653]: DEBUG(+): GdmSlave: Creating proxy for /org/gnome/DisplayManager/Display1

and so on for about 3000 lines. then it says:

Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmSlaveProxy: slave (pid:2184) done (status:0)
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmDisplay: Slave exited: 0
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmDisplay: Finishing display: /org/gnome/DisplayManager/Display6
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmDisplay: Unmanaging display
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmDisplay: unmanage display
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmSlaveProxy: Killing slave
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmSlaveProxy: Disposing slave proxy
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmSlaveProxy: Killing slave
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: WARNING: GdmDisplay: display lasted 0.195054 seconds
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmLocalDisplayFactory: static display status changed: 4
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: DEBUG(+): GdmDisplayStore: Unreffing display: 0x19763f0
Apr 20 19:20:08 r4pt0r gdm-binary[2137]: WARNING: GdmLocalDisplayFactory: maximum number of X display failures reached: check X server log for errors
Apr 20 19:20:08 r4pt0r init: prefdm main process (2137) terminated with status 1
Apr 20 19:20:08 r4pt0r init: prefdm respawning too fast, stopped

but

cat /var/log/Xorg.0.log | grep EE

only returns

[    22.690] (EE) No input driver/identifier specified (ignoring)
[    22.758] (EE) No input driver/identifier specified (ignoring)

I guess gnome is not the problem since the freeze sometimes occurs during start up.
can anyone point me to the right direction? I don't really want to reinstall fedora...

[jpollard]

The problem is not Fedora.

Could you try booting into initmode 3, then login on a console terminal and try to
start X with "startx" ?

It sounds like something is wrong with graphics. The error "..respawning..." only
occurs when the X server aborts on its own. Unfortunately, it isn't reporting what
the fault is (there are several places where logs occur, not all of them in /var/log/Xorg..., one other place is in the home directory (.xsession-errors is another).

In all cases, there are some errors that cannot get logged (Signal 11 can be one of
them). This is partly due to running under a GUI, where stderr is "not all ways" redirected to a usable log output. This can happen during server startup (logs are
supposed to go to /var/log/Xorg....) but during startup, stderr gets switched at
certain times to the users home directory (.xsession-errors) - and if the failure occurs
between the two logs, it can get lost.

The hope with "startx" is that all errors go the console terminal used to start X - even
those that usually get missed.

Another error is when the physical graphics interface doesn't quite match the driver
being used. What do you get from "lspci | grep video"?

[newiLuvatar]

I suspect it has something to do with my graphics card. that's why I borrowed a gpu from a friend to see if the problem is still there.
a grep for video displays nothing, however a grep for vga shows

01:00.0 VGA compatible controller: nVidia Corporation G80 [GeForce 8800 GTS] (rev a2)

this is the borrowed card, I usually run a 9800gtx+.
I'll see if the problem is persists with this gpu. if it doesn't, I guess there's something wrong with my gpu (although it runs perfectly under windows). if it does, I'll keep investigating and see if I can find some reasonable error messages in my logs...
I have to add that the freeze sometimes occurs even during startup, before X gets started. so I don't know if X is doing anything wrong.

[domg472]

I have/had similar issues with intel gpu. This is how i (believe) it was fixed for me:

http://forums.fedoraforum.org/showpo...57&postcount=6

[newiLuvatar]

well, I have compiz disabled, no intel gpu though.
interestingly enough, the borrowed gpu seems to run normal until I start rendercheck, then it crashes. I will see if my gpu behaves the same. perhaps I can figure out what triggers the lockups...

---------- Post added at 11:00 PM CDT ---------- Previous post was at 06:19 PM CDT ----------

unfortunately the boot runlevel 3 and start x there revealed no additional information. .xsession-errors is also no good because it gets cleared at reboot.
not really sure what to do else...

[Jake]

I just got around to re-installing a second Fedora OS (Been sorting other stuff out)

it works partially Ok. But also getting hard lock-ups. but in the sense that it's completely gone, even pressing caps lock key does not turn on the LED, same as moving mouse does not move the courser. Think of it like, looking at a photograph .

Ironically this being a fresh install also means I haven't tampered with anything (Yet!). But I seem to have got odd lock-ups, first one I noticed, was when I opened nautilus first time to view files on an external drive, second, was opening Firefox for first time, and third was sending a message on empathy (After updates),

But, if I do all of these things now, it does not occur, seems to only happen on first time.

I'm not sure if the fact I currently have dual-monitors has anything to do with it. Though they are perfectly fine with Fedora 12 (as is everything else).

I'll have to wait off using that FC13 a while until a proper fix for this is found, I really hate holding that power button to halt my machine. I tried an old-trick of ctrl alt backspace (after enabling it in keyboard pref) that doesn't work either.

I've only ever seen these kinds of lock-ups with my system when I've literally exhausted all of my ram (eg used up all swap, and all 2GB ram). However on those occasions the system did resume after killing the app which caused it. I tried leaving Fedora 13 to see if it was the same thing, it never resumed :-(.

I hope this gets sorted :-)

And Yes, it's x86_64.

[newiLuvatar]

still nothing conclusive... borrowed gpu works fine... so it could be that my gpu is faulty, on the other hand it works fine under windows, so I'm not quite sure...
guess I'll have to try to reinstall...

[newiLuvatar]

hmm, fedora seems to running a little more stable than earlier...
before the last lock up, I got this:
May 1 11:59:59 r4pt0r kernel: [drm] nouveau 0000:01:00.0: fail wait other chan
May 1 11:59:59 r4pt0r kernel: [drm] nouveau 0000:01:00.0: validate vram_list
May 1 11:59:59 r4pt0r kernel: [drm] nouveau 0000:01:00.0: validate: -16
might this be the problem?

[Maryyy]

IIUC, you are still trying with nouveau loaded?
I would just boot in RL3, got rid of nouveau and try nvidia drivers. The card you borrowed may behave a bit differently, maybe because if nouveau is the cause, the bug in it can be triggered only by utilizing features specific for your previous GPU. Just try Nvidia drivers or even VESA. Hardlockups like this are not normal stuff.
BTW, even if the system seems to be totally dead, you can try magic keys, but they need to be enabled in /etc/sysctl.conf first. (if something hasn't changed from F12, I'm not on F13 atm). If for nothing else, if they work, your filesystem gets cleanly unmounted, unlike with cold reset.

[newiLuvatar]

well, I was having the issue with the nvidia driver first, so I changed to nouveau...
guess I'll try nvidia and vesa again
magic keys? couldn't find anything in /etc/sysctl.conf...

---------- Post added at 07:32 PM CDT ---------- Previous post was at 06:50 PM CDT ----------

ok, tried running nvidia, is crashing even faster... vesa doesn't seem to work at all...
nouveau is the most stable driver right now...

[Maryyy]

kernel.sysrq = 1

more info about it e.g. here. http://en.wikipedia.org/wiki/Magic_SysRq_key I can't imagine not having them enabled. (although I needed them only once and it was my fault). Basically, hold down Alt+SysRq and while holding these two down, type R S E I U B, with something like 1-2 sec delay between them. It should rebbot after that, if it doesn't, just hit ctrl+alt+del, should reboot now.

If nvidia crashes too, well then I guess you'll have to wait for Fedora to sort this out. But filing bug would be a good idea, maybe they even don't know about this issue, devs most of the time don't read forums.

[newiLuvatar]

thanks for the tip, I will try that...
I want to file a bug report, but I don't yet know if it is a hardware or software problem... and also not which component is faulty... perhaps I'll give f14 a try...

[newiLuvatar]

just installed ubuntu, runs perfect so far...
so, I'm kinda helpless:
my gpu works perfectly on windows and ubuntu. but fedora (and interestingly f12 and f13, installed and live) crashes randomly, sometimes at boot, sometimes in gnome. it's not a question of if, but when. no difference when using nouveau or nvidia driver. guess I'll try to put my gpu in another pc and boot a live cd. if it works, then either my gpu is faulty (which I doubt, win and ubuntu run without problems), or fedora is buggy...