[arch-general] Linux-3.7.3-1-i686 blank screen
I ran pacmatic -Syu last night, which updated the kernel to the version in the subject. This afternoon, the system started to boot normally but the display went blank before the login prompt appeared (I boot to a text login) so I was unable to use the system.
I used my original installation disc to boot, and was able to follow the Wiki instructions (accessed using a different computer) to run chroot and downgrade the kernel to 3.6.11-1, and the system is now running again.
An 'unconfirmed' bug is listed at https://bugs.archlinux.org/ but this seems to relate to vlc segfaulting with the new kernel, so I don't know if that's the same problem affecting me.
On Tue, 22 Jan 2013 17:34:29 +0000 Whiskers catwheezel@operamail.com wrote:
I ran pacmatic -Syu last night, which updated the kernel to the version in the subject. This afternoon, the system started to boot normally but the display went blank before the login prompt appeared (I boot to a text login) so I was unable to use the system.
I used my original installation disc to boot, and was able to follow the Wiki instructions (accessed using a different computer) to run chroot and downgrade the kernel to 3.6.11-1, and the system is now running again.
An 'unconfirmed' bug is listed at https://bugs.archlinux.org/ but this seems to relate to vlc segfaulting with the new kernel, so I don't know if that's the same problem affecting me.
Linux 3.7.x has seen a lot of kms/drm changes, so it's hard to answer your question without knowing your graphics hardware. FWIW, for me all 3.7 kernel boot fine on machines with (not-so-new) intel and radeon cards. Also, can you boot with "nomodeset" in kernel boot line?
Best, L.
On Tue, 22 Jan 2013 12:20:28 -0600 Leonid Isaev lisaev@umail.iu.edu wrote:
On Tue, 22 Jan 2013 17:34:29 +0000 Whiskers catwheezel@operamail.com wrote:
I ran pacmatic -Syu last night, which updated the kernel to the version in the subject. This afternoon, the system started to boot normally but the display went blank before the login prompt appeared (I boot to a text login) so I was unable to use the system.
I used my original installation disc to boot, and was able to follow the Wiki instructions (accessed using a different computer) to run chroot and downgrade the kernel to 3.6.11-1, and the system is now running again.
An 'unconfirmed' bug is listed at https://bugs.archlinux.org/ but this seems to relate to vlc segfaulting with the new kernel, so I don't know if that's the same problem affecting me.
Linux 3.7.x has seen a lot of kms/drm changes, so it's hard to answer your question without knowing your graphics hardware. FWIW, for me all 3.7 kernel boot fine on machines with (not-so-new) intel and radeon cards. Also, can you boot with "nomodeset" in kernel boot line?
Best, L.
I'll keep the nomodeset option in mind but I'm reluctant to try Linux-3.7.3-1 again just yet!
Other people have posted problems with this kernel to the "Pacman & Package Upgrade Issues" web forum, so I'm not entirely on my own.
Hi,
Am 22.01.2013 18:34, schrieb Whiskers:
An 'unconfirmed' bug is listed at https://bugs.archlinux.org/ but this seems to relate to vlc segfaulting with the new kernel, so I don't know if that's the same problem affecting me.
No, that's almost certainly another issue.
You may want to try 3.7.4-1 from testing. If that doesn't fix your issue, you are probably best off to contact upstream directly. With Linux 3.7 there was quite some restructuring of the graphics stack going on, so it may well be that in some cases it breaks things.
Best regards, Karol Babioch
On Tue, 22 Jan 2013 19:25:45 +0100 Karol Babioch karol@babioch.de wrote:
Hi,
Am 22.01.2013 18:34, schrieb Whiskers:
An 'unconfirmed' bug is listed at https://bugs.archlinux.org/ but this seems to relate to vlc segfaulting with the new kernel, so I don't know if that's the same problem affecting me.
No, that's almost certainly another issue.
You may want to try 3.7.4-1 from testing. If that doesn't fix your issue, you are probably best off to contact upstream directly. With Linux 3.7 there was quite some restructuring of the graphics stack going on, so it may well be that in some cases it breaks things.
Best regards, Karol Babioch
I'm using the Nouveau video driver with an old laptop. As Linux-3.6.11-1 seems to be stable for me, I'll wait for the next kernel version to reach the core repository.
I'll also try to work out how to have the previous kernel still available to Grub-legacy after the next kernel upgrade.
For the record, this is what my video hardware says it is:
$ lspci | grep VGA 01:00.0 VGA compatible controller: NVIDIA Corporation NV31M [GeForce FX Go5600] (rev a1)
On Tue, 22 Jan 2013 21:07:56 +0000 Whiskers catwheezel@operamail.com wrote:
On Tue, 22 Jan 2013 19:25:45 +0100 Karol Babioch karol@babioch.de wrote:
Hi,
Am 22.01.2013 18:34, schrieb Whiskers:
An 'unconfirmed' bug is listed at https://bugs.archlinux.org/ but this seems to relate to vlc segfaulting with the new kernel, so I don't know if that's the same problem affecting me.
No, that's almost certainly another issue.
You may want to try 3.7.4-1 from testing. If that doesn't fix your issue, you are probably best off to contact upstream directly. With Linux 3.7 there was quite some restructuring of the graphics stack going on, so it may well be that in some cases it breaks things.
Best regards, Karol Babioch
I'm using the Nouveau video driver with an old laptop. As Linux-3.6.11-1 seems to be stable for me, I'll wait for the next kernel version to reach the core repository.
If you're talking about 3.7.4, then it's unlikely to fix your issue, because there are only radeon fixes in the changelog http://www.kernel.org/pub/linux/kernel/v3.0/ChangeLog-3.7.4.
I'll also try to work out how to have the previous kernel still available to Grub-legacy after the next kernel upgrade.
For the record, this is what my video hardware says it is:
$ lspci | grep VGA 01:00.0 VGA compatible controller: NVIDIA Corporation NV31M [GeForce FX Go5600] (rev a1)
On Tue, 22 Jan 2013 16:20:47 -0600 Leonid Isaev lisaev@umail.iu.edu wrote:
On Tue, 22 Jan 2013 21:07:56 +0000 Whiskers catwheezel@operamail.com wrote:
On Tue, 22 Jan 2013 19:25:45 +0100 Karol Babioch karol@babioch.de wrote:
Hi,
Am 22.01.2013 18:34, schrieb Whiskers:
[...]
I'm using the Nouveau video driver with an old laptop. As Linux-3.6.11-1 seems to be stable for me, I'll wait for the next kernel version to reach the core repository.
If you're talking about 3.7.4, then it's unlikely to fix your issue, because there are only radeon fixes in the changelog http://www.kernel.org/pub/linux/kernel/v3.0/ChangeLog-3.7.4.
I'll also try to work out how to have the previous kernel still available to Grub-legacy after the next kernel upgrade.
For the record, this is what my video hardware says it is:
$ lspci | grep VGA 01:00.0 VGA compatible controller: NVIDIA Corporation NV31M [GeForce FX Go5600] (rev a1)
I'm now up to Linux 3.7.6-1, and still the video card is going blank when booting. It happens reliably whenever I boot with a kernel later than linux-3.6.11-1-i686.pkg.tar.xz and happens at the point in the boot sequence where the print on the screen shrinks to a small size - I think that's when the 'frame buffer' is started.
I can log in 'blind' and issue console commands (eg shutdown -r now) but startx does nothing.
If I edit the 3.7.* boot entry in Grub-legacy to include nomodeset then I do get to a visible login prompt, but of course no X.
For now, I'm booting into linux-lts-3.0.60-1-i686.pkg.tar.xz which works fine, but may not be a long-term solution.
I don't know enough to interpret all the log entries, but I have found that when booting from a kernel that successfully starts the graphics, /var/log/kernel.log has a line such as this:
Feb 9 18:29:00 tavy kernel: [ 18.124619] [drm] Initialized nouveau 0.0.16 20090420 for 0000:01:00.0 on minor 0
whereas using a kernel that doesn't get the graphics working, the corresponding log entry is like this:
Feb 9 18:23:33 tavy kernel: [ 16.068483] [drm] Initialized nouveau 1.1.0 20120801 for 0000:01:00.0 on minor 0
Both entries are confusing, given that the current nouveau packages installed are nouveau-dri 9.0.2-1 and xf86-video-nouveau 1.0.6-1. Would I be right in guessing that xf86-video-nouveau 1.0.6-1 can handle the kernel initialising nouveau 0.0.16 but not 1.1.0?
On 09-02-2013 23:59, Whiskers wrote:
On Tue, 22 Jan 2013 16:20:47 -0600 Leonid Isaev lisaev@umail.iu.edu wrote:
On Tue, 22 Jan 2013 21:07:56 +0000 Whiskers catwheezel@operamail.com wrote:
On Tue, 22 Jan 2013 19:25:45 +0100 Karol Babioch karol@babioch.de wrote:
Hi,
Am 22.01.2013 18:34, schrieb Whiskers:
[...]
I'm using the Nouveau video driver with an old laptop. As Linux-3.6.11-1 seems to be stable for me, I'll wait for the next kernel version to reach the core repository.
If you're talking about 3.7.4, then it's unlikely to fix your issue, because there are only radeon fixes in the changelog http://www.kernel.org/pub/linux/kernel/v3.0/ChangeLog-3.7.4.
I'll also try to work out how to have the previous kernel still available to Grub-legacy after the next kernel upgrade.
For the record, this is what my video hardware says it is:
$ lspci | grep VGA 01:00.0 VGA compatible controller: NVIDIA Corporation NV31M [GeForce FX Go5600] (rev a1)
I'm now up to Linux 3.7.6-1, and still the video card is going blank when booting. It happens reliably whenever I boot with a kernel later than linux-3.6.11-1-i686.pkg.tar.xz and happens at the point in the boot sequence where the print on the screen shrinks to a small size - I think that's when the 'frame buffer' is started.
I can log in 'blind' and issue console commands (eg shutdown -r now) but startx does nothing.
If I edit the 3.7.* boot entry in Grub-legacy to include nomodeset then I do get to a visible login prompt, but of course no X.
For now, I'm booting into linux-lts-3.0.60-1-i686.pkg.tar.xz which works fine, but may not be a long-term solution.
I don't know enough to interpret all the log entries, but I have found that when booting from a kernel that successfully starts the graphics, /var/log/kernel.log has a line such as this:
Feb 9 18:29:00 tavy kernel: [ 18.124619] [drm] Initialized nouveau 0.0.16 20090420 for 0000:01:00.0 on minor 0
whereas using a kernel that doesn't get the graphics working, the corresponding log entry is like this:
Feb 9 18:23:33 tavy kernel: [ 16.068483] [drm] Initialized nouveau 1.1.0 20120801 for 0000:01:00.0 on minor 0
It seems you are using late KMS, try adding nouveau to the MODULES array in /etc/mkinitcpio.conf and see if you still have problems.
I've had a problem similar to this quite a long time ago when kms stated to be supported by the radeon driver, adding the module to the initramfs solved the problem reliably.
Both entries are confusing, given that the current nouveau packages installed are nouveau-dri 9.0.2-1 and xf86-video-nouveau 1.0.6-1. Would I be right in guessing that xf86-video-nouveau 1.0.6-1 can handle the kernel initialising nouveau 0.0.16 but not 1.1.0?
On Sun, 10 Feb 2013 00:15:50 +0000 Mauro Santos registo.mailling@gmail.com wrote:
On 09-02-2013 23:59, Whiskers wrote:
On Tue, 22 Jan 2013 16:20:47 -0600 Leonid Isaev lisaev@umail.iu.edu wrote:
On Tue, 22 Jan 2013 21:07:56 +0000 Whiskers catwheezel@operamail.com wrote:
On Tue, 22 Jan 2013 19:25:45 +0100 Karol Babioch karol@babioch.de wrote:
Hi,
Am 22.01.2013 18:34, schrieb Whiskers:
[...]
I'm using the Nouveau video driver with an old laptop. As Linux-3.6.11-1 seems to be stable for me, I'll wait for the next kernel version to reach the core repository.
If you're talking about 3.7.4, then it's unlikely to fix your issue, because there are only radeon fixes in the changelog http://www.kernel.org/pub/linux/kernel/v3.0/ChangeLog-3.7.4.
I'll also try to work out how to have the previous kernel still available to Grub-legacy after the next kernel upgrade.
For the record, this is what my video hardware says it is:
$ lspci | grep VGA 01:00.0 VGA compatible controller: NVIDIA Corporation NV31M [GeForce FX Go5600] (rev a1)
I'm now up to Linux 3.7.6-1, and still the video card is going blank when booting. It happens reliably whenever I boot with a kernel later than linux-3.6.11-1-i686.pkg.tar.xz and happens at the point in the boot sequence where the print on the screen shrinks to a small size - I think that's when the 'frame buffer' is started.
I can log in 'blind' and issue console commands (eg shutdown -r now) but startx does nothing.
If I edit the 3.7.* boot entry in Grub-legacy to include nomodeset then I do get to a visible login prompt, but of course no X.
For now, I'm booting into linux-lts-3.0.60-1-i686.pkg.tar.xz which works fine, but may not be a long-term solution.
I don't know enough to interpret all the log entries, but I have found that when booting from a kernel that successfully starts the graphics, /var/log/kernel.log has a line such as this:
Feb 9 18:29:00 tavy kernel: [ 18.124619] [drm] Initialized nouveau 0.0.16 20090420 for 0000:01:00.0 on minor 0
whereas using a kernel that doesn't get the graphics working, the corresponding log entry is like this:
Feb 9 18:23:33 tavy kernel: [ 16.068483] [drm] Initialized nouveau 1.1.0 20120801 for 0000:01:00.0 on minor 0
It seems you are using late KMS, try adding nouveau to the MODULES array in /etc/mkinitcpio.conf and see if you still have problems.
I've had a problem similar to this quite a long time ago when kms stated to be supported by the radeon driver, adding the module to the initramfs solved the problem reliably.
Both entries are confusing, given that the current nouveau packages installed are nouveau-dri 9.0.2-1 and xf86-video-nouveau 1.0.6-1. Would I be right in guessing that xf86-video-nouveau 1.0.6-1 can handle the kernel initialising nouveau 0.0.16 but not 1.1.0?
-- Mauro Santos
Have you tried the xf86-video-nv driver from [extra]. Also, if preloading the module in mkinitcpio.conf does not work. I would try appending both pci=nommconf and idle=poll to the kernel line of your boot loader..
On Sat, 9 Feb 2013 16:33:52 -0800 Federico Cinelli cinelli.federico@gmail.com wrote:
Have you tried the xf86-video-nv driver from [extra].
Not yet; if necessary I'll even try the proprietary nVidia driver.
Also, if preloading the module in mkinitcpio.conf does not work. I would try appending both pci=nommconf and idle=poll to the kernel line of your boot loader..
That makes no difference, unfortunately. Thanks for the idea though.
On Sun, Feb 10, 2013 at 09:55:17PM +0000, Whiskers wrote:
On Sat, 9 Feb 2013 16:33:52 -0800 Federico Cinelli cinelli.federico@gmail.com wrote:
Have you tried the xf86-video-nv driver from [extra].
Not yet; if necessary I'll even try the proprietary nVidia driver.
Is nv still supported and usable ?
I had a problem with a new install using nouveau recently. Looked up the archwiki page on nv, but this only mentioned the proprietary nvidia driver (which did solve the problem).
Ciao,
I've been trying again with the latest kernel, 3.7.9-1-i686. My logs are now showing what looks like a possible clue; here's what I think is the interesting part of /var/log/Xorg.0.log.old for my last attempt to boot using that kernel:
[ 1426.786] (II) NOUVEAU(0): NVLeaveVT is called. [ 1426.786] (II) NOUVEAU(0): Closed GPU channel 1 [ 1426.788] (WW) xf86CloseConsole: KDSETMODE failed: Input/output error [ 1426.788] (WW) xf86CloseConsole: VT_GETMODE failed: Input/output error [ 1426.788] Fatal server error: [ 1426.788] xf86CloseConsole: VT_ACTIVATE failed: Input/output error [ 1426.788] [ 1426.788] (EE) Please consult the The X.Org Foundation support at http://wiki.x.org for help. [ 1426.788] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information. [ 1426.788] (EE) [ 1426.788] (EE) [ 1426.788] (EE) Backtrace: [ 1426.789] (EE) 0: /usr/bin/X (xorg_backtrace+0x49) [0x81e4a59] [ 1426.789] (EE) 1: /usr/bin/X (0x8048000+0x1a0946) [0x81e8946] [ 1426.789] (EE) 2: linux-gate.so.1 (__kernel_rt_sigreturn+0x0) [0xb771640c] [ 1426.789] (EE) 3: /usr/bin/X (0x8048000+0x1a7180) [0x81ef180] [ 1426.789] (EE) 4: /usr/bin/X (CloseWellKnownConnections+0x34) [0x81e5744] [ 1426.789] (EE) 5: /usr/bin/X (0x8048000+0x1a8834) [0x81f0834] [ 1426.789] (EE) 6: /usr/bin/X (0x8048000+0x1a8f5e) [0x81f0f5e] [ 1426.789] (EE) 7: /usr/bin/X (0x8048000+0x9d0f5) [0x80e50f5] [ 1426.789] (EE) 8: /usr/bin/X (xf86CloseConsole+0x157) [0x80e5ad7] [ 1426.789] (EE) 9: /usr/bin/X (ddxGiveUp+0x8d) [0x80bee5d] [ 1426.789] (EE) 10: /usr/bin/X (0x8048000+0x1fa1f) [0x8067a1f] [ 1426.789] (EE) 11: /usr/lib/libc.so.6 (__libc_start_main+0xf5) [0xb736a825] [ 1426.789] (EE) 12: /usr/bin/X (0x8048000+0x1fae9) [0x8067ae9] [ 1426.789] (EE) [ 1426.789] (EE) Segmentation fault at address 0x44 [ 1426.789] FatalError re-entered, aborting [ 1426.789] Caught signal 11 (Segmentation fault). Server aborting [ 1426.789]
I have installed xf86-video-vesa 2.3.2-2 but while xf86-video-nouveau 1.0.6-1 is also installed that seems to be given preference in the boot procedure, so I'll remove it and keep my fingers crossed!
On Sun, 10 Feb 2013 00:15:50 +0000 Mauro Santos registo.mailling@gmail.com wrote:
It seems you are using late KMS, try adding nouveau to the MODULES array in /etc/mkinitcpio.conf and see if you still have problems.
Thanks for the suggestion; sadly, it made no difference this time.
participants (6)
-
Federico Cinelli
-
Fons Adriaensen
-
Karol Babioch
-
Leonid Isaev
-
Mauro Santos
-
Whiskers