Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.
Sign upNVidia driver installation fails with provided instructions #2526
Comments
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
|
I formatted your report to make it readable. |
andrewdavidwong
added
C: doc
C: other
help wanted
labels
Dec 19, 2016
andrewdavidwong
added this to the Release 3.2 updates milestone
Dec 19, 2016
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
andrewdavidwong
Dec 19, 2016
Member
Have you already seen this thread?
https://groups.google.com/d/topic/qubes-users/v26zXkiNElg/discussion
|
Have you already seen this thread? https://groups.google.com/d/topic/qubes-users/v26zXkiNElg/discussion |
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
0spinboson
Dec 21, 2016
I've not had much luck getting it to work either, though for different reasons / at a different stage.
full recipe as I tried it:
- dom0: qubes-dom0-update gcc kernel-devel kmod kernel-headers grub2-tools
- appvm: download NVIDIA-Linux-375.26-install.run (any appvm)
- dom0: qvm-run --pass-io 'cat /PATH/TO/NVIDIA.run' > NVIDIA.run; chmod +x NVIDIA.run;
- dom0: ./NVIDIA.run --ui=none --no-x-check --keep --extract-only
5: dom0: cd NVIDIA-*/kernel/; make module IGNORE_XEN_PRESENCE=y CC="gcc -DNV_VMAP_4_PRESENT -DNV_SIGNAL_STRUCT_RLIM";
6: dom0: sudo cp nvidia.ko /lib/modules/4.4.31-11.pvops.qubes.x86_64/extra/
7: dom0: sudo depmod -a; modinfo nvidia (all working will show up fine)
8: dom0: EDIT GRUB or EFI; add rd.driver.blacklist=nouveau to kernel options; sudo grub2-mkconfig -o /boot/grub2/grub.cfg
9: dom0; reboot and enjoy nvidia support.
I've tried to install nv drivers (375.26, gtx 950) using the above method, but couldn't get them to load properly. xinit fails; startx actually hangs without allowing me to switch to a different tty.
Error message varied depending on how I try to init X, either:
"parse_vt_settings: Cannot open /dev/tty0 (No such file or directory)"
or something like:
(II) NVIDIA: Using ~16GB of virtual memory for indirect memory
(II) NVIDIA: access.
(EE) NVIDIA(GPU-0): Failed to initialize DMA.
(EE) *** Aborting ***
(EE)
Fatal server error:
(EE) AddScreen/ScreenInit failed for driver 0
or something about "iops" which I didn't write down. As such, I'm still/back on nouveau
0spinboson
commented
Dec 21, 2016
•
edited
Edited 1 time
-
0spinboson
edited Dec 21, 2016 (most recent)
edited
-
Dec 21, 2016 (most recent)
|
I've not had much luck getting it to work either, though for different reasons / at a different stage.
I've tried to install nv drivers (375.26, gtx 950) using the above method, but couldn't get them to load properly. xinit fails; startx actually hangs without allowing me to switch to a different tty. or something like: or something about "iops" which I didn't write down. As such, I'm still/back on nouveau |
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
marmarek
Dec 22, 2016
Member
|
On Wed, Dec 21, 2016 at 10:32:34AM -0800, 0spinboson wrote:
(II) NVIDIA: Using ~16GB of virtual memory for indirect memory
(II) NVIDIA: access.
(EE) NVIDIA(GPU-0): Failed to initialize DMA.
(EE) *** Aborting ***
(EE)
This one looks like incompatibility with Xen... Does 'nvidia' kernel
module have any options (`modinfo nvidia`)? Maybe there is some switch
for this?
Also, check if you have some related message in Xen log (`xl dmesg`). If
anything, it should be at the end (just after failed start of X server).
…--
Best Regards,
Marek Marczykowski-Górecki
Invisible Things Lab
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
|
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
0spinboson
Dec 24, 2016
The only thing modinfo gives me is this, none of which seems xen-related to me. I didn't get the message you quoted consistently, though. More usually it was the 'cannot open /dev/tty0'. modinfo output attached.
Rechecking the nvidia driver manual installation guide a 4th time, the only deviation I see is that the /x64/ directory doesn't exist, which I didn't quite notice because the kernel module built without reporting any error. Could that be behind it? (I.e., /usr/src/kernels/2.6.34.1-12.xenlinux.qubes.x86_64/arch/x64/include/mach-xen should be present (if not - take it from kernel sources) ) If so, where can I find its contents?
0spinboson
commented
Dec 24, 2016
|
The only thing modinfo gives me is this, none of which seems xen-related to me. I didn't get the message you quoted consistently, though. More usually it was the 'cannot open /dev/tty0'. modinfo output attached. Rechecking the nvidia driver manual installation guide a 4th time, the only deviation I see is that the /x64/ directory doesn't exist, which I didn't quite notice because the kernel module built without reporting any error. Could that be behind it? (I.e., /usr/src/kernels/2.6.34.1-12.xenlinux.qubes.x86_64/arch/x64/include/mach-xen should be present (if not - take it from kernel sources) ) If so, where can I find its contents? |
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
loeken
Dec 30, 2016
@andrewdavidwong thanks for formatting, sorry for my late response.
So I also tried the approach shown in that google groups post ( https://groups.google.com/d/topic/qubes-users/v26zXkiNElg/discussion ). And I successfully built the module and could load it, modinfo shows what @0spinboson posted in the last response:
filename: /lib/modules/4.4.31-11.pvops.qubes.x86_64/extra/nvidia.ko
alias: char-major-195-*
version: 375.26
supported: external
license: NVIDIA
srcversion: 7850503DF85713C6E0D88A3
alias: pci:v000010DEd00000E00sv*sd*bc04sc80i00*
alias: pci:v000010DEd*sv*sd*bc03sc02i00*
alias: pci:v000010DEd*sv*sd*bc03sc00i00*
depends:
vermagic: 4.4.31-11.pvops.qubes.x86_64 SMP mod_unload
parm: NVreg_Mobile:int
parm: NVreg_ResmanDebugLevel:int
parm: NVreg_RmLogonRC:int
parm: NVreg_ModifyDeviceFiles:int
parm: NVreg_DeviceFileUID:int
parm: NVreg_DeviceFileGID:int
parm: NVreg_DeviceFileMode:int
parm: NVreg_UpdateMemoryTypes:int
parm: NVreg_InitializeSystemMemoryAllocations:int
parm: NVreg_UsePageAttributeTable:int
parm: NVreg_MapRegistersEarly:int
parm: NVreg_RegisterForACPIEvents:int
parm: NVreg_CheckPCIConfigSpace:int
parm: NVreg_EnablePCIeGen3:int
parm: NVreg_EnableMSI:int
parm: NVreg_TCEBypassMode:int
parm: NVreg_UseThreadedInterrupts:int
parm: NVreg_MemoryPoolSize:int
parm: NVreg_RegistryDwords:charp
parm: NVreg_RmMsg:charp
parm: NVreg_AssignGpus:charp
I also disabled nouveau like by adding
rd.driver.blacklist=nouveau
to /etc/default/grub to the cmdline_linux line and regenerated grub.cfg with grub2-mkconf.
however when issuing lscpi -v -s 02:00.0 it shows me that both modules nouveau and nvidia are loaded but that nouveau is still in use.
[loeken@dom0 ~]$ lspci -v -s 02:00.0
02:00.0 VGA compatible controller: NVIDIA Corporation GM206 [GeForce GTX 950] (rev a1) (prog-if 00 [VGA controller])
Subsystem: Micro-Star International Co., Ltd. [MSI] Device 8c92
Physical Slot: 2
Flags: bus master, fast devsel, latency 0, IRQ 177, NUMA node 0
Memory at fa000000 (32-bit, non-prefetchable) [size=16M]
Memory at e0000000 (64-bit, prefetchable) [size=256M]
Memory at f0000000 (64-bit, prefetchable) [size=32M]
I/O ports at e000 [size=128]
Expansion ROM at fb000000 [disabled] [size=512K]
Capabilities: <access denied>
Kernel driver in use: nouveau
Kernel modules: nouveau, nvidia
I recorded my steps as a video and uploaded to youtube for clarification
https://www.youtube.com/watch?v=BvJYEkiyeoc
any ideas how to actually disable nouveau?
loeken
commented
Dec 30, 2016
|
@andrewdavidwong thanks for formatting, sorry for my late response. I also disabled nouveau like by adding however when issuing lscpi -v -s 02:00.0 it shows me that both modules nouveau and nvidia are loaded but that nouveau is still in use. I recorded my steps as a video and uploaded to youtube for clarification any ideas how to actually disable nouveau? |
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
0spinboson
Dec 30, 2016
I've tried to do so by uninstalling anything nouveau-related, rebuilding initramfs, then grub2-mkconfig. But the 'best' outcome I get afterwards is the nvidia driver failing to load properly, and X not initializing and/or hanging (while giving one of several different errors, with which Google could not help me). That said, I'm still pretty stupid when it comes to linux, so my being unable to get it to work is most likely due to me lacking experience.
0spinboson
commented
Dec 30, 2016
|
I've tried to do so by uninstalling anything nouveau-related, rebuilding initramfs, then grub2-mkconfig. But the 'best' outcome I get afterwards is the nvidia driver failing to load properly, and X not initializing and/or hanging (while giving one of several different errors, with which Google could not help me). That said, I'm still pretty stupid when it comes to linux, so my being unable to get it to work is most likely due to me lacking experience. |
This comment has been minimized.
Show comment
Hide comment
This comment has been minimized.
tonsimple
Jan 12, 2017
This might be relevant to anyone still looking into this
https://groups.google.com/forum/#!topic/qubes-users/v26zXkiNElg
Basically, I've tried, build and successfully install driver, but x dies due to a nefarious "Failed to allocate sofware rendering cache surface: out of memory" error that is notorious NVIDIA compatibility error. I dunno what to do next. Might try doing a no-igfx parameter, but I don't have much hope.
Doomed to use integrated graphics for now.
tonsimple
commented
Jan 12, 2017
|
This might be relevant to anyone still looking into this Basically, I've tried, build and successfully install driver, but x dies due to a nefarious "Failed to allocate sofware rendering cache surface: out of memory" error that is notorious NVIDIA compatibility error. I dunno what to do next. Might try doing a no-igfx parameter, but I don't have much hope. Doomed to use integrated graphics for now. |
loeken commentedDec 19, 2016
•
edited by andrewdavidwong
Edited 1 time
-
andrewdavidwong
edited Dec 19, 2016 (most recent)
Qubes OS version (e.g.,
R3.2):R3.2
Affected TemplateVMs (e.g.,
fedora-23, if applicable):I tested on fedora-23 template ( work, app vm)
Expected behavior:
https://www.qubes-os.org/doc/install-nvidia-driver/
Actual behavior:
Steps to reproduce the behavior:
following the installation instructions from https://www.qubes-os.org/doc/install-nvidia-driver/
General notes:
To me it looks like it fails at:
when it tries to access the non-existing file
/usr/share/nvidia-kmod-367.57/nvidia-kmod-367.57-x86_64.tar.xzhowever there is a .rpm file with the same "name" inside the folder i executed the yumdownloader commands in.
rpm fusion is enabled:
Related issues:
I also failed on using the manual method. there are kernel 2.6. files, so i guess this process is maybe outdated and not working anymore?