| View previous topic :: View next topic |
| Author |
Message |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Sun Jul 01, 2012 8:44 pm Post subject: |
|
|
| wcg wrote: | Another datapoint:
This all works with a Quadro NVS 290 (NV86 (G86) chip, NV50 family,
according to http://nouveau.freedesktop.org/wiki/CodeNames ),
using
| Code: |
kernel 3.2.12 (with drm, kms, and nouveau; FB_VESA, FB_UVESA, FB_NVIDIA not enabled)
x11-base/xorg-server-1.12.2
xf86-video-nouveau-0.0.16_pre20120322
|
Meaning that the vga to framebuffer console switch works at boot,
X works, and switching back and forth between X and virtual
consoles works. (Not tested with the closed source nvidia drivers.) |
Hello wcg.
That is what this specific forum post is about. It pertains (I think exclusively) to Nvidia drivers only.
An Nvidia driver higher than 290.* means there are no consoles for users of GeForce8800GT cards, and I presume cards of around the same age.
Users with newer "nvidia" cards are not affected. Likewise, users with nouveau drivers can switch between consoles without a problem. _________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Sun Jul 01, 2012 9:05 pm Post subject: |
|
|
Hey Princess, I might be barking up the wrong tree but is "reroute broken IRQ" enabled in your kernel?
I only say this because of the following lines in your code.
| Code: | | Jun 26 00:25:14 localhost kernel: irq 16: nobody cared (try booting with the "irqpoll" option) |
| Code: | | Jun 26 00:25:14 localhost kernel: [<c108cc54>] __report_bad_irq+0x24/0xc0 |
_________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
Princess Nell Guru


Joined: 15 Apr 2005 Posts: 465
|
Posted: Mon Jul 02, 2012 9:11 pm Post subject: |
|
|
Let's hope that bark is worse than the bite
Well spotted, I did not have this option set. I have now, and it makes no difference wrt the subject of this discussion (with 302.17). Going back to 290.10, so I won't have an opportunity to check whether any of those logs disappear (which I suspect). |
|
| Back to top |
|
 |
Princess Nell Guru


Joined: 15 Apr 2005 Posts: 465
|
Posted: Mon Jul 02, 2012 10:31 pm Post subject: |
|
|
What can I say. Even the brief period running under 302.17 crashed the machine with a log very similar to what I posted. Again, that's for CONFIG_X86_REROUTE_FOR_BROKEN_BOOT_IRQS=y.
| Code: |
Jul 2 22:53:26 localhost kernel: NVRM: GPU at 0000:01:00.0 has fallen off the bus.
Jul 2 22:53:26 localhost kernel: NVRM: GPU at 0000:01:00.0 has fallen off the bus.
Jul 2 22:53:30 localhost kernel: irq 16: nobody cared (try booting with the "irqpoll" option)
Jul 2 22:53:30 localhost kernel: Pid: 0, comm: swapper/1 Tainted: P O 3.2.12-gentoo #2
Jul 2 22:53:30 localhost kernel: Call Trace:
Jul 2 22:53:30 localhost kernel: [<c13f38be>] ? printk+0x18/0x1a
Jul 2 22:53:30 localhost kernel: [<c108cc54>] __report_bad_irq+0x24/0xc0
Jul 2 22:53:30 localhost kernel: [<c108cead>] note_interrupt+0x10d/0x1b0
Jul 2 22:53:30 localhost kernel: [<f9d56c4e>] ? nv_kern_isr+0x2e/0x80 [nvidia]
Jul 2 22:53:30 localhost kernel: [<c108af4b>] handle_irq_event_percpu+0x6b/0x140
Jul 2 22:53:30 localhost kernel: [<c101cc87>] ? io_apic_modify_irq.clone.26+0x87/0xa0
Jul 2 22:53:30 localhost kernel: [<c108b057>] handle_irq_event+0x37/0x60
Jul 2 22:53:30 localhost kernel: [<c108d6d0>] ? unmask_irq+0x20/0x20
Jul 2 22:53:30 localhost kernel: [<c108d72e>] handle_fasteoi_irq+0x5e/0xf0
Jul 2 22:53:30 localhost kernel: <IRQ> [<c100428d>] ? do_IRQ+0x3d/0xb0
Jul 2 22:53:30 localhost kernel: [<c105d101>] ? notifier_call_chain+0x41/0x60
Jul 2 22:53:30 localhost kernel: [<c13f72e9>] ? common_interrupt+0x29/0x30
Jul 2 22:53:30 localhost kernel: [<c105007b>] ? sys_setfsgid+0x1b/0xb0
Jul 2 22:53:30 localhost kernel: [<c1279698>] ? acpi_idle_enter_bm+0x232/0x26e
Jul 2 22:53:30 localhost kernel: [<c132090f>] ? cpuidle_idle_call+0x7f/0xd0
Jul 2 22:53:30 localhost kernel: [<c1001787>] ? cpu_idle+0x47/0x90
Jul 2 22:53:30 localhost kernel: [<c13f006b>] ? start_secondary+0x1ae/0x1b5
Jul 2 22:53:30 localhost kernel: handlers:
Jul 2 22:53:30 localhost kernel: [<f9d56c20>] nv_kern_isr
Jul 2 22:53:30 localhost kernel: Disabling IRQ #16
|
|
|
| Back to top |
|
 |
philip Guru

Joined: 10 Jun 2003 Posts: 404 Location: Sweden
|
Posted: Tue Jul 03, 2012 5:35 pm Post subject: |
|
|
I do have tty after login. But my problem is that the GUI just freezes and the sound gets stuck i a loop.
My /var/log/messages at the time of the event:
| Code: | Jul 3 19:25:42 phils_pc kernel: NVRM: GPU at 0000:01:00.0 has fallen off the bus.
Jul 3 19:25:42 phils_pc kernel: NVRM: os_pci_init_handle: invalid context!
Jul 3 19:25:42 phils_pc kernel: NVRM: os_pci_init_handle: invalid context!
Jul 3 19:25:42 phils_pc kernel: NVRM: os_pci_init_handle: invalid context!
Jul 3 19:25:42 phils_pc kernel: NVRM: os_pci_init_handle: invalid context!
Jul 3 19:25:42 phils_pc kernel: irq 24: nobody cared (try booting with the "irqpoll" option)
Jul 3 19:25:42 phils_pc kernel: Pid: 0, comm: kworker/0:1 Tainted: P 2.6.39-gentoo-r2 #7
Jul 3 19:25:42 phils_pc kernel: Call Trace:
Jul 3 19:25:42 phils_pc kernel: [<c1053132>] __report_bad_irq+0x30/0x89
Jul 3 19:25:42 phils_pc kernel: [<c1053436>] note_interrupt+0x119/0x17b
Jul 3 19:25:42 phils_pc kernel: [<c10520ea>] handle_irq_event_percpu+0xfc/0x10d
Jul 3 19:25:42 phils_pc kernel: [<c1052127>] handle_irq_event+0x2c/0x45
Jul 3 19:25:42 phils_pc kernel: [<c1053957>] ? handle_simple_irq+0x54/0x54
Jul 3 19:25:42 phils_pc kernel: [<c10539bd>] handle_fasteoi_irq+0x66/0x84
Jul 3 19:25:42 phils_pc kernel: <IRQ> [<c10039a7>] ? do_IRQ+0x37/0x90
Jul 3 19:25:42 phils_pc kernel: [<c134d329>] ? common_interrupt+0x29/0x30
Jul 3 19:25:42 phils_pc kernel: [<c1202d1d>] ? acpi_idle_enter_simple+0xfe/0x131
Jul 3 19:25:42 phils_pc kernel: [<c12a8dfd>] ? cpuidle_idle_call+0x6a/0x9b
Jul 3 19:25:42 phils_pc kernel: [<c10015b7>] ? cpu_idle+0x3a/0x63
Jul 3 19:25:42 phils_pc kernel: [<c1346e99>] ? start_secondary+0x19a/0x19f
Jul 3 19:25:42 phils_pc kernel: handlers:
Jul 3 19:25:42 phils_pc kernel: [<fa16384f>] (nv_kern_isr+0x0/0x54 [nvidia])
Jul 3 19:25:42 phils_pc kernel: Disabling IRQ #24
Jul 3 19:25:43 phils_pc kernel: irq 16: nobody cared (try booting with the "irqpoll" option)
Jul 3 19:25:43 phils_pc kernel: Pid: 0, comm: kworker/0:1 Tainted: P 2.6.39-gentoo-r2 #7
Jul 3 19:25:43 phils_pc kernel: Call Trace:
Jul 3 19:25:43 phils_pc kernel: [<c1053132>] __report_bad_irq+0x30/0x89
Jul 3 19:25:43 phils_pc kernel: [<c1053436>] note_interrupt+0x119/0x17b
Jul 3 19:25:43 phils_pc kernel: [<c10520ea>] handle_irq_event_percpu+0xfc/0x10d
Jul 3 19:25:43 phils_pc kernel: [<c1052127>] handle_irq_event+0x2c/0x45
Jul 3 19:25:43 phils_pc kernel: [<c1053957>] ? handle_simple_irq+0x54/0x54
Jul 3 19:25:43 phils_pc kernel: [<c10539bd>] handle_fasteoi_irq+0x66/0x84
Jul 3 19:25:43 phils_pc kernel: <IRQ> [<c10039a7>] ? do_IRQ+0x37/0x90
Jul 3 19:25:43 phils_pc kernel: [<c134d329>] ? common_interrupt+0x29/0x30
Jul 3 19:25:43 phils_pc kernel: [<c1202d1d>] ? acpi_idle_enter_simple+0xfe/0x131
Jul 3 19:25:43 phils_pc kernel: [<c12a8dfd>] ? cpuidle_idle_call+0x6a/0x9b
Jul 3 19:25:43 phils_pc kernel: [<c10015b7>] ? cpu_idle+0x3a/0x63
Jul 3 19:25:43 phils_pc kernel: [<c1346e99>] ? start_secondary+0x19a/0x19f
Jul 3 19:25:43 phils_pc kernel: handlers:
Jul 3 19:25:43 phils_pc kernel: [<fa616360>] (azx_interrupt+0x0/0x112 [snd_hda_intel])
Jul 3 19:25:43 phils_pc kernel: Disabling IRQ #16 |
My version of Nvidia-drivers:
| Code: | # emerge --search nvidia-drivers
Searching...
[ Results for search key : nvidia-drivers ]
[ Applications found : 1 ]
* x11-drivers/nvidia-drivers
Latest version available: 295.49
Latest version installed: 295.49
Size of files: 33,903 kB
Homepage: http://www.nvidia.com/
Description: NVIDIA X11 driver and GLX libraries
License: NVIDIA |
_________________ /Phil |
|
| Back to top |
|
 |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Tue Jul 03, 2012 8:01 pm Post subject: |
|
|
Hello Philip
I noticed you have kernel:
Which could be the reason why you are getting a tty.
But not so good for the frozen gui.
I am using a 3.2.12 kernel.
I would guess if you had a more recent kernel then you would not have a tty?
Could you try that with your current nvidia-driver and a newer kernel?
I will let the nvidia forums know when you get back to me with the results. _________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
TomWij Developer

Joined: 04 Jul 2012 Posts: 51
|
Posted: Wed Jul 04, 2012 6:56 pm Post subject: |
|
|
Same problem here, different set-up though:
| Code: | 02:00.0 VGA compatible controller: NVIDIA Corporation G92 [GeForce GTX 285M] (rev a2)
x11-drivers/nvidia-drivers 302.17 (acpi kernel_linux multilib tools -kernel_FreeBSD) |
|
|
| Back to top |
|
 |
wcg Guru

Joined: 06 Jan 2009 Posts: 556
|
Posted: Fri Jul 06, 2012 8:51 pm Post subject: |
|
|
My comments were merely for the benefit of anyone reading the thread
who has one of the affected chips and is wondering whether they can
get a working desktop or do they need a different gpu. (They can get
a working deskop if they can live without the features not supported
by the nouveau driver, in the context of these kernel and xorg-server
versions, and if they decline to enable frame buffer drivers not needed
with the nouveau driver, because it implements its own framebuffer driver.)
Other data (used by xorg):
| Code: |
libdrm-2.4.33
mesa-8.0.3
|
I did not see a dmesg in this thread. The nvidia drivers developers
may be able to guess what IRQ 16 and 24 are assigned to, but a
dmesg from the kernels where these problems crop up would probably
help them diagnose those problems. _________________ TIA |
|
| Back to top |
|
 |
Princess Nell Guru


Joined: 15 Apr 2005 Posts: 465
|
Posted: Fri Jul 06, 2012 9:50 pm Post subject: |
|
|
| Code: |
# grep -i "irq 16" dmesg
pci 0000:00:01.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
pci 0000:00:1c.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
ata_piix 0000:00:1f.1: PCI INT A -> GSI 16 (level, low) -> IRQ 16
nvidia 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
#
|
|
|
| Back to top |
|
 |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Fri Jul 06, 2012 10:12 pm Post subject: |
|
|
| wcg wrote: | My comments were merely for the benefit of anyone reading the thread
who has one of the affected chips and is wondering whether they can
get a working desktop or do they need a different gpu. (They can get
a working deskop if they can live without the features not supported
by the nouveau driver, in the context of these kernel and xorg-server
versions, and if they decline to enable frame buffer drivers not needed
with the nouveau driver, because it implements its own framebuffer driver.)
Other data (used by xorg):
| Code: |
libdrm-2.4.33
mesa-8.0.3
|
I did not see a dmesg in this thread. The nvidia drivers developers
may be able to guess what IRQ 16 and 24 are assigned to, but a
dmesg from the kernels where these problems crop up would probably
help them diagnose those problems. |
An Nvidia driver higher than 290.* means there are no consoles for users of GeForce8800GT cards, and I presume cards of around the same age. _________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
Kururu n00b

Joined: 09 Jul 2012 Posts: 1
|
Posted: Mon Jul 09, 2012 1:56 pm Post subject: |
|
|
I think this bug is related to another bug that cause random crashes and freezes to some card model of the 8000/9000 series, GT200 series (like mine GT250) and some Quadro.
The nvidia developers are aware of this bug but they haven't release a valid fix until now.
My system is:
-intel Core i5, 4G RAM
-NVIDIA GT250 1G RAM
I tested all the drivers version from 290.10 to 302.17, with various kernels from 3.1 to 3.4.0, and xorg-server from 1.10 to 1.12.2: with every drivers version >290.10 I have no console and experience random crashes and freezes.
Now I'm testing the gentoo-kernel-3.4.0, xorg-server-1.12.2, and nvidia-drivers 302.17, and I tried a fix to the console problem that I found in another forum: i added the parameter "nomodeset" to the kernel line in grub like this:
| Quote: | title Gentoo Linux 3.4.0
root (hd0,0)
kernel /boot/vmlinuz-3.4.0 root=/dev/sda1 video=uvesafb:1280x1024-32,mtrr:1,ywrap console=tty1 nomodeset |
or, without vesafb
| Quote: | title Gentoo Linux 3.4.0(nofb)
root (hd0,0)
kernel /boot/vmlinuz-3.4.0 root=/dev/sda1 console=tty1 nomodeset |
With this i can again switch from X to console and back for a couple of times but then the console goes black again. |
|
| Back to top |
|
 |
UncleVan n00b

Joined: 08 Feb 2011 Posts: 30
|
Posted: Tue Jul 10, 2012 3:00 am Post subject: Link to the bug at NVIDIA ? |
|
|
Did someone post this to the nvidia guys ? A link here highly appreciated - Thanks in advance !
Your UncleVan. |
|
| Back to top |
|
 |
Gusar Advocate

Joined: 09 Apr 2005 Posts: 2557 Location: Slovenia
|
Posted: Tue Jul 10, 2012 10:23 am Post subject: Re: Link to the bug at NVIDIA ? |
|
|
| UncleVan wrote: | | Did someone post this to the nvidia guys ? |
As I already mentioned earlier in the thread, there are at least half a dozen threads about this at nvnews.net. The last time a Nvidia dev commented, he said they haven't figured out the root cause of the issue yet.
You also have to understand that the console is an extremely low priority for Nvidia, their focus is X. |
|
| Back to top |
|
 |
UncleVan n00b

Joined: 08 Feb 2011 Posts: 30
|
Posted: Tue Jul 10, 2012 5:06 pm Post subject: Small suggestion |
|
|
| Gusar wrote: | | You also have to understand that the console is an extremely low priority for Nvidia, their focus is X. | ... cant be serious... What I HAVE understood is - no more nvidia hardware into the box...
Anyway, would you everyone with this problem try following:
Put this line in your xorg.conf - Device Section:
| Code: | Section "Device"
....
Option "NvAGP" "0"
....
EndSection
|
and tell me how it works in regard to the console issue. This disables AGP support and seems to cachè that bug.
Please let us know . - Your UncleVan. |
|
| Back to top |
|
 |
UncleVan n00b

Joined: 08 Feb 2011 Posts: 30
|
Posted: Wed Jul 11, 2012 11:06 am Post subject: Option "NvAGP" "0" |
|
|
Console lost after 36 hrs.
| Code: | vbetool vbemode get
| reports 16741 again.... Frustrating. |
|
| Back to top |
|
 |
bulletmark n00b

Joined: 14 Jul 2012 Posts: 2
|
Posted: Sat Jul 14, 2012 4:43 am Post subject: Re: Small suggestion |
|
|
| UncleVan wrote: |
Put this line in your xorg.conf - Device Section:
|
I am running Ubuntu 12.04 (sorry guys!) but see this same problem. Just tried adding this xorg option but nvidia driver exited with "unsupported option".
I have seen this bug on every nvidia driver since 295.33, i.e. also 295.40, 295.49, 295.53, 302.07, 302.11, 302.17, and today's 304.22. Also have tried kernel 3.2 -> 3.5.0-4.
Virtual consoles work fine if I boot with nouveau instead of nvidia. Note that this new nvidia driver 304.22 still suffers the segmentation fault that all the other versions I list above exhibit.
Aborted after 3 hours with:
| Code: |
10057.732360] Xorg[1514]: segfault at b4f25000 ip b4948ec0 sp bfb341a0 error 7 in nvidia_drv.so[b4865000+64d000]
|
PS: I will add that I have a 8600 GT card with 2 by 1920x1200 screens. |
|
| Back to top |
|
 |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Mon Jul 16, 2012 12:09 pm Post subject: Re: Link to the bug at NVIDIA ? |
|
|
| UncleVan wrote: | Did someone post this to the nvidia guys ? A link here highly appreciated - Thanks in advance !
Your UncleVan. |
UncleVan, I have posted the link to the nvidia devs and the gentoo bugzilla on the 1st page of this forum post.
They are fully aware of the regression but have not nailed it yet.
Some guys on nvnews say it can take nvidia a year to nail these problems.
My solution for the time being is to stick with the latest stable gentoo-sources, and use the nvidia-drivers-290.* series.
At least that way, the user has tty's.  _________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
UncleVan n00b

Joined: 08 Feb 2011 Posts: 30
|
Posted: Tue Jul 17, 2012 10:15 pm Post subject: Thanks Thistled ! - but even better : ... |
|
|
.. . - nouveau !
- Extremly fast mode switching
- superior frame buffer console
- true color videos in the VC - no X at all ! (needs DirectFB with video_cards_nvidia USE flag)
- 100% suspend/hibernate/resume
- minimal memory requirements
- 3D ??
etc.
Needs vdpau, g3dvl and gallium enabled in mesa a.o. for better performance.
And think twice next time you buying hardware .
Yor UncleVan. |
|
| Back to top |
|
 |
Princess Nell Guru


Joined: 15 Apr 2005 Posts: 465
|
Posted: Wed Aug 22, 2012 8:29 pm Post subject: |
|
|
I'm using nouveau on my 64-bit system. But 32-bit is still my main productivity platform.
Finally some progress. The list of bug fixes for 302.17 and 304.37 is quite long, and while nothing immediately looked related, I unmasked and built 304.37, and it looks promising. Console switching seems to work again, and I was finally able to unmask xorg-server-1.12 and glibc-2.15. |
|
| Back to top |
|
 |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Wed Aug 22, 2012 10:08 pm Post subject: |
|
|
This looks like good news Princess Nell.
But just out of interest, which graphics card are you using?
I am using a GeForce8800 GT
I might just try the latest nvidia if things are on the up. _________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
bulletmark n00b

Joined: 14 Jul 2012 Posts: 2
|
Posted: Thu Aug 23, 2012 12:36 am Post subject: |
|
|
| @Thistled, I have an 8600 GT and find nvidia v304.37 driver finally fixes the console ttys for me. It also, more importantly, fixes the xorg segmentation fault crash. Both these bugs have existed with my card since early nvidia v295.*. The funny thing is that after all this time now using the latest nouveau driver (which I had never used prior to these problems with nvidia) I find that I seem to get slightly better graphics performance on nouveau so I am staying with it. |
|
| Back to top |
|
 |
Princess Nell Guru


Joined: 15 Apr 2005 Posts: 465
|
Posted: Thu Aug 23, 2012 7:32 pm Post subject: |
|
|
| Code: |
[ 9687.178] (II) NVIDIA(0): NVIDIA GPU GeForce 8600M GT (G84) at PCI:1:0:0 (GPU-0)
|
|
|
| Back to top |
|
 |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Fri Aug 24, 2012 10:22 pm Post subject: |
|
|
I switched to Nouveau, as expected the consoles returned and I was very happy. Hey, the console switching was instant! I have never had that with Nvidia.
But, and it's a big but, (oo'er) I installed wine and tried to play a windows game "S.T.A.L.K.E.R." and the jittering rendered the game unplayable.
Incidentally, I am not really a PC gamer, more of a tweaker geek. lol.
So, I switched back to nvidia-drivers-2.90.* so I could get a console, and jitter free gaming.
I am just about to try the nvidia driver you mention, and see if I can "game" and use the consoles.
If that is the case, then it's a major hats off to nvidia devs, as it was mentioned on the nv-forums that it might take a year or so to fix this regression. _________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
Thistled Guru


Joined: 06 Jan 2011 Posts: 433 Location: Scotland
|
Posted: Sat Aug 25, 2012 2:04 pm Post subject: |
|
|
** UPDATE **
The new driver does indeed work!
Yes!
At last, the consoles and all the lovely fbterm decor is back.
I have put a big thank you to the nvidia devs on their forums.
 _________________ Whatever you do, do it properly! |
|
| Back to top |
|
 |
rubik-wuerfel n00b

Joined: 29 Sep 2004 Posts: 43 Location: Goettingen, Germany
|
Posted: Sat Mar 16, 2013 8:46 pm Post subject: |
|
|
FYI, the behaviour mentioned here just happened to me with the Nvidia module 304.84, with this device:
| Code: | # lspci | grep VGA
00:05.0 VGA compatible controller: NVIDIA Corporation C51 [Quadro NVS 210S/GeForce 6150LE] (rev a2) |
x11-drivers/nvidia-drivers-306.64 still works fine.
| Code: | # uname -a
Linux whatnow 3.6.11-gentoo-r1 #2 SMP Thu Mar 14 00:44:13 CET 2013 x86_64 AMD Processor model unknown AuthenticAMD GNU/Linux |
|
|
| Back to top |
|
 |
|