|
|
#1 | |
|
Registered User
Join Date: Oct 2006
Location: Germany
Posts: 77
|
Hi all.
I already posted briefly about this here but this really deserves its own thread because it makes composition unstable and useless. Again, I just had a hang when switching from the text console to Xorg while running KDE 4 w/ composition on. The mouse cursor was fully rendered and movable but that's it. I managed to restart the machine with the magic keys and recovered the following infos: Code:
Nov 29 15:26:38 kernel: [ 5693.743545] NVRM: Xid (0000:01:00): 13, 0004 00000000 00009097 00002484 00000206 00000000 Nov 29 15:26:38 kernel: [ 5693.750761] NVRM: Xid (0000:01:00): 31, Ch 00000004, engmask 00000101, intr 10000000 Nov 29 15:26:38 kernel: [ 5693.762515] NVRM: Xid (0000:01:00): 31, Ch 00000004, engmask 00000101, intr 10000000 Nov 29 15:26:38 kernel: [ 5693.769236] NVRM: Xid (0000:01:00): 31, Ch 00000004, engmask 00000101, intr 10000000 Nov 29 15:26:45 kernel: [ 5701.247069] NVRM: Xid (0000:01:00): 8, Channel 00000004 Nov 29 15:26:47 kernel: [ 5703.247185] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context Nov 29 15:26:51 kernel: [ 5707.251346] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context Nov 29 15:26:53 kernel: [ 5709.251493] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context Nov 29 15:26:55 kernel: [ 5711.251980] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context Nov 29 15:26:57 kernel: [ 5713.252125] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context Nov 29 15:28:49 kernel: [ 5825.129774] SysRq : SAK To sum it up, every time one switches from the text console to X, the screen is either there immediately or can take up to several seconds to appear... and sometimes, not all. Sometimes, a window is initially there and the rest comes a second or two later. This is really annoying. If there is anything I can do to help locate this bug, please let me know. So long matthias. PS. Just in case someone actually tells me I have the composition extension disabled: Yes, I have now and I did this _after_ the hang. :-) |
|
|
|
|
|
|
#2 | |
|
Registered User
Join Date: Oct 2006
Location: Germany
Posts: 77
|
@nvidia: An official word on this would be very much appreciated. ;-) Using composition has some real advantages besides the not so interesting eye candy and I'd really like to be able to use it withoung having to worry about data loss.
|
|
|
|
|
|
|
#3 |
|
Registered User
Join Date: May 2004
Location: Norway
Posts: 117
|
Long standing issue, at least it has been for me and my NVS 3100M (Dell Laptop).
Thread here: http://www.nvnews.net/vbulletin/showthread.php?t=157041 May have improved slightly in later versions (running latest nvidia driver now), but it still happens from time to time. (And when it does, it wrecks havoc with stability.) |
|
|
|
|
|
#4 | |
|
Developer
Join Date: Sep 2011
Location: Australia
Posts: 5
|
Quick question: does /usr/bin/X occasionally show high CPU usage for you when the delay occurs? Because if so then we share the same problem.
|
|
|
|
|
|
|
#5 |
|
Registered User
Join Date: Oct 2006
Location: Germany
Posts: 77
|
Hi @all.
First of all, I think the recent support NVIDIA has given us is rather shameful and kinda looks like "we are only dealing with issues we are comfortable with". So I am pinging this up and I hope that a dev has something useful to say. After all, everyone in this forum paid good money for their cards and can expect them to work more or less flawlessly - and that includes the composite extension. Saying one should deactivate it (the standard recently), is really absolutely not helpful, to say the least. And to be totally blunt, if NVIDIA wants to cut back on the resources allocated to the Linux driver development, maybe they should stop it altogether, release the specs and help nouveau become a serious alternative/standard. Or people might think twice next time they are out to purchase a new card - after all, the AMD open source stack is maturing slowly but it is evolving steadily. @RobertMassaioli: Unfortunately I cannot test it easily as I cannot login to my box remotely while it hangs due to a lack of another system. ![]() Merry Christmas to @all... |
|
|
|
|
|
#6 | |
|
Registered User
Join Date: Oct 2006
Location: Germany
Posts: 77
|
@NVIDIA: Still the same with 295.09. sigh Any official word please?
|
|
|
|
|
|
|
#7 |
|
Registered User
Join Date: Nov 2011
Posts: 19
|
Same here, I also tried xorg-server 1.11.99.2 and it doesn't work either. I guess I'm going back to nouveau because those freezes are starting to be really annoying.
I would also like to read at least a little about this from NVIDIA. Could you say at least if you are trying to fix this bug, please? |
|
|
|
|
|
#8 |
|
Registered User
Join Date: Oct 2004
Posts: 118
|
Seems to be fixed in 295.33. Thanks.
|
|
|
|
|
|
#9 |
|
Registered User
Join Date: Oct 2006
Location: Germany
Posts: 77
|
@Belegdol: Sorry but I have to disagree unfortunately.
The bug is still present w/ 295.33, GTX470, kernel 3.2.11, Xorg 1.12 and Gnome 3.4 (even though it doesn't matter if its Gnome or KDE or whatever). It seems a little bit harder to trigger but still happens, is extremly annoying and renders any compositing useless which also means no Gnome 3 for example. I don't think NVIDIA will ever fix this. This thread gets no attention whatsoever. And recently their Linux support has really started to drop imho. Bugs don't get the attention they deserve (like this one). Honestly, I am more than disappointed. Maybe they should finally start to support nouveau if they don't have the man power or the will to continue their (previously) superb support. If this does not get fixed and the AMD open source drivers are finally a valuable alternative, I know what my next card will be. And note here I was talking about the open source drivers. I know that the AMD proprietary drivers suck in terms of support. Sorry but this should have been fixed ages ago... really disappointing! |
|
|
|
|
|
#10 |
|
Registered User
Join Date: May 2004
Location: Norway
Posts: 117
|
Yes, typically. Sometimes it's possible to recover from hangs. However, when things are really stuck, Xorg process spins with 100% on one core.
|
|
|
|
|
|
#11 | |
|
Registered User
Join Date: Oct 2004
Posts: 118
|
Quote:
GeForce GTX 485M xorg-x11-server-common-1.11.4-2.fc16.x86_64 xorg-x11-drv-nvidia-295.33-3.fc16.x86_64 kernel-3.3.0-4.fc16.x86_64 gnome-shell-3.2.2.1-1.fc16.x86_64 I am not disagreeing with you, just wanted to show what works for me. |
|
|
|
|
![]() |
| Thread Tools | |
|
|