| View previous topic :: View next topic |
| Author |
Message |
jagdpanther
Apprentice
Joined: 22 Nov 2003
Posts: 260
|
 Posted: Fri Feb 20, 2015 6:06 pm Post subject: [Solved] How To Not AutoLoad Nvidia-Drivers During Boot? |
 |
|
Because I have been having an intermittent boot-time freezing issue that seems to be caused by the nvidia-driver ( [url] https://forums.gentoo.org/viewtopic-t-1011112-highlight-.html[/url] ) I do NOT want the nvidia driver to load at boot time. Instead I want to autoload the nvidia kernel module when I type 'startx' from the console.
I have removed xdm from rc-update. I have switched OpenCL implimintation to intel via 'eselect opencl'. I have removed 'nvidia' from /etc/conf.d/modules.
The nvidia kernel module is still being loaded at boot-time. From dmesg:
| Code: | [ 2.947372] xhci_hcd 0000:05:00.0: new USB bus registered, assigned bus
number 3
[ 2.950221] random: nonblocking pool is initialized
[ 2.952612] nvidia: module license 'NVIDIA' taints kernel.
[ 2.952613] Disabling lock debugging due to kernel taint
[ 2.953599] AVX2 version of gcm_enc/dec engaged.
[ 2.955237] alg: No test for __gcm-aes-aesni (__driver-gcm-aes-aesni)
[ 2.958816] alg: No test for crc32 (crc32-pclmul)
[ 2.959888] vgaarb: device changed decodes:
PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=io+mem
[ 2.960605] [drm] Initialized nvidia-drm 0.0.0 20130102 for 0000:01:00.0 on
minor 0
[ 2.960618] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 343.36 Mon Dec
1 16:18:58 PST 2014
[ 3.008128] hub 3-0:1.0: USB hub found |
Is it autoloading because 'eselect opengl' is set to nvidia?
Does it have something to do with /etc/conf.d/nvidia-persistenced ?
Grep shows that 'nvidia' is in the binary file /etc/udev/hwdb.bin.
How do I keep the nvidia driver module from loading at boot-time while keeping it available for later use?
ps. (I tried moving the acutal nvidia module, /lib/modules/3.17.8-gentoo-r1/video/nvidia.ko, and that worked, but I had to move it back before 'startx' and that is a horrible hack.)
Last edited by jagdpanther on Fri Feb 20, 2015 10:09 pm; edited 1 time in total |
|
| Back to top |
|
 |
Naib
Advocate
Joined: 21 May 2004
Posts: 4339
Location: Removed by Neddy
|
| Posted: Fri Feb 20, 2015 6:39 pm Post subject: |
|
|
Do as that thread suggests, blacklist the nvidia module. You than write an alias that insmod nvidia and startx
Odd that there is a kernel panic with the nvidia driver. Might be worth fixing that
I have been using Linux and nvidia for ... 15years or so without such a problem, so it is possible
_________________
the table is made from wood. forget what you learnt, the table is made from carbon. forget what you learnt, the table is made from protons. forget what you learnt, the table is made from quarks. forget what you learnt, the table is good for shagging on |
|
| Back to top |
|
|
Jaglover
Advocate
Joined: 29 May 2005
Posts: 4954
Location: Saint Amant, Acadiana
|
| Posted: Fri Feb 20, 2015 6:42 pm Post subject: |
|
|
Actually I just tried it on my laptop, didn't even need an alias, when X starts it loads the module (which is blacklisted for kernel and friends).
_________________
Please learn how to denote units correctly! |
|
| Back to top |
|
|
depontius
Advocate
Joined: 05 May 2004
Posts: 2788
|
| Posted: Fri Feb 20, 2015 7:11 pm Post subject: |
|
|
This is the basic piece of the process for quickly switching between nouveau and nvidia. Blacklist both. All that means is that neither will load a boot time. When X starts it will demand-load he module called for in xorg.conf.
You still need to reboot to switch - I wouldn't trust rmmod then modprobe to switch between kernel modules. But at least you only have to tweak xorg.conf and not a lot of other junk in order to switch.
_________________
.sigs waste space and bandwidth |
|
| Back to top |
|
|
jagdpanther
Apprentice
Joined: 22 Nov 2003
Posts: 260
|
| Posted: Fri Feb 20, 2015 7:55 pm Post subject: |
|
|
Thanks for the suggestions. I need to look up how to 'blacklist' the module. I thought that was a concept not an actual procedure.
| Quote: | Odd that there is a kernel panic with the nvidia driver. Might be worth fixing that
I have been using Linux and nvidia for ... 15years or so without such a problem, so it is possible |
Yes, I have been using Gentoo for a decade and Linux longer and have rarely had nvidia driver issues. I think that this time it is a combination of a new motherboard (Asus X99-A) and new video card (GTX 970). I expect that with newer nvidia-drivers these issues will disappear. |
|
| Back to top |
|
|
jagdpanther
Apprentice
Joined: 22 Nov 2003
Posts: 260
|
| Posted: Fri Feb 20, 2015 10:09 pm Post subject: |
|
|
blacklisting the nvidia kernel module solved the problem.
Thanks |
|
| Back to top |
|
|
ulenrich
Veteran
Joined: 10 Oct 2010
Posts: 1219
|
| Posted: Sat Feb 21, 2015 10:03 pm Post subject: |
|
|
Just let override /lib/udev/rules.d/99-nvidia.rules by this file
/etc/udev/rules.d/99-nvidia.rules
| Code: | # ACTION=="add", DEVPATH=="/module/nvidia", SUBSYSTEM=="module", RUN+="nvidia-udev.sh $env{ACTION}"
# Previously the ACTION was "add|remove" but one user on bug #376527 had a
# problem until he recompiled udev-171-r5, which is one of the versions I
# tested with and it was fine. I'm breaking the rules out just to be safe
# so someone else doesn't have an issue
ACTION=="remove", DEVPATH=="/module/nvidia", SUBSYSTEM=="module", RUN+="nvidia-udev.sh $env{ACTION}" |
I added an additional comment sign in the first line!
_________________
fun2gen2 |
|
| Back to top |
|
|
|
Portage & Programming
