NVidia drivers: Some hard freezes on -current...
I have tried a variety of NVidia drivers, using the NVidia installer, finally settling on 313.09. With all of these versions I experience a hard freeze every couple of days. By 'hard freeze' I mean the display freezes and there is no response from mouse and keyboard, a reboot from the power switch is required. This with a GTS 450.
Has anybody else experienced this issue? |
Blind guess: maybe the nvidia drivers you tried are not fully compatible with mesa 9.1? See recent entry in the Changelog:
Code:
x/mesa-9.1-i486-1.txz: Upgraded. |
I'm using an Nvidia optimus laptop and have yet to experience these freezes (when I use the Nvidia card).
|
Quote:
Seriously, do you use the (same) proprietary driver? |
Yes, I wouldn't have posted otherwise. I tested 313.09 for awhile (to see any if there were any improved FPS results in certain programs), and when I didn't see any I reverted back to 310.32, but no problems on either of them.
|
Quote:
https://wiki.archlinux.org/index.php...ze_using_flash So I have disabled hardware acceleration with Flash and I will install the latest 310 drivers and then cross my fingers... |
Quote:
|
Perhaps not, I did all of the above and the system again froze :). This time opening a new tab in Firefox directed from an URL in mutt...
|
Spending far too much time with this one :(. Tried nouveau for a while and it looks like a reasonable backup if all else fails. Disabled compositing in xorg.conf as it seems to be a villain in a number of bug reports...
|
I don’t know if this helps, but I've also been experiencing problems with the nvidia driver and I suspect that the numerous random freezes that have been reported in the net arise from different bugs. Hence, there is no simple solution for the problem. Sometimes, the suggested fixes work, but quite often they don’t. It all depends on the combination of the driver and the kernel. In my case, crashes start becoming frequent when I upgraded to nvidia 310.32. The symptoms were always the same: a frozen video display, but the computer was still working. Loging in via ssh allowed me to restart the machine after some house-cleaning, but interestingly I couldn’t find any traces KDE or even X!
dmesg consistently displayed this message irq 16: nobody cared (try booting with the "irqpoll" option) Pid: 1773, comm: kwin Tainted: P O 3.6.03.6 #4 Call Trace: <IRQ> [<ffffffff8108ac9f>] ? __report_bad_irq+0x2f/0xd0 [<ffffffff8108afb5>] ? note_interrupt+0x1d5/0x220 [<ffffffff810889ae>] ? handle_irq_event_percpu+0x7e/0x140 [<ffffffff81088aaa>] ? handle_irq_event+0x3a/0x70 [<ffffffff8108b9ac>] ? handle_fasteoi_irq+0x4c/0xe0 [<ffffffff81004335>] ? handle_irq+0x15/0x20 [<ffffffff81004242>] ? do_IRQ+0x52/0xd0 [<ffffffff8146bca7>] ? common_interrupt+0x67/0x67 <EOI> [<ffffffff8146c3a2>] ? system_call_fastpath+0x16/0x1b handlers: [<ffffffff812ef200>] usb_hcd_irq [<ffffffffa07c3b30>] nv_kern_isr [nvidia] Disabling IRQ #16 After googling for this, I’ve tried all different sorts of suggested fixes (“irqpoll” in lilo header, “/usr/bin/nvidia-smi -pm 1” in rc.local, etc) but freezes still occurred randomly. I was about to change my old nvidia 8600GT when I upgraded to kernel 3.8.03.6, keeping nvidia 310.32, “irqpoll” in lilo and “/usr/bin/nvidia-smi -pm 1” in rc.local, and I haven’t seen a frozen screen since then! Antonio PS: I'm not running -current. It's a slack64 multilib setup. Only the kernel, nvidia blob, and standard patches were upgraded |
Thanks for that Sairum. So far I have:
So now I will sit tight and see if I have beaten the freezes... |
Quote:
|
Quote:
|
Still getting the odd hard freeze so for the moment I have removed the NVidia drivers and I have decided to use nouveau for a couple of months. Then perhaps try newer NVidia drivers and see if the issue is resolved. Nouveau seems to work well enough, I miss the 3d acceleration in my vms and d2x-rebirth will not run but a stable system may very well be worth the price...
|
Running -Current doesn't always mean that everything will work perfectly with the system.
Proprietary drivers are no exception. The Nvidia driver shouldn't even have top worry about LibMesa at all as it contains it's own OpenGL libraries. Because you have an Optimus setup... Try this website for help getting the Optimus muxer support known as Bumblebee working. Code:
http://bumblebee-project.org/ |
Optimus? Not here :)
|
Looks like with nouveau my system is again solid as a rock, but not quite as pretty as it was with the NVidia blob. I am a little surprised that the problem is not so widespread in this Slackware forum but I know that Google has seen the issue many times and not just recently. Waiting for newer NVidia drivers before I try again, sigh....
|
The long-term driver branch has been updated to 310.40
http://www.nvidia.com/object/unix.html |
Quote:
|
I've had freezing up: -current (kernel 3.7.1) using the nouveau driver on a motherboard with integrated Geforce 7025 graphics. Within a few seconds of logging into to a user account using KDE, or even attempting to launch Firefox from within FVWM. No Ctrl-Alt-Bksp, Alt-SysRq combinations worked.
I've since installed nVidia 304.84 drivers, but need to see if that improves matters. |
304.84 seems ok so far but black screen when logout. I can see a fix for kde but I am using Fluxbox...
|
I am marking this thread Solved as my own situation has stabilised completely. Currently running kernel 3.8.4 and NVidia 313.26 and no more freezing or black windows on logout. I can see by these Forums that the NVidia drivers are going through a rough patch at the moment so I am glad to have found a solution in my own case :).
|
All times are GMT -5. The time now is 05:07 PM. |