* Microsoft Azure NV-series instances with NVidia GRID drivers started to experience xserver crashes while following Microsoft's official guide to installing Nvidia drivers [1].
* Root cause analysis showed that it was due to having a device with BusID "PCI:0@<domain_id>:0:0", where domain id is >= 32767 while the hyperv_drm kernel module is loaded.
* Removing either the BusID specification or unloading the hyperv_drm kernel module seems to fix the crash.
* The crash is happening while X.server is trying to enumerate PCI devices. X.server dereferences a NULL pointer while trying to access to the PCI device info.
* The reason why it only happens while the hyperv_drm kernel module is loaded is that the hyperv_drm module does not expose PCI hardware information since it's a virtual device.
* The upstream patch [2] addresses the issue and it's confirmed that the xserver with the patch does not experience the crash.
* Ubuntu Focal `xorg-server` package does not include the patch [2] at the moment (xserver-xorg-core=2:1.20.13-1ubuntu1~20.04.6).
Part (a) is quoted from Microsoft's official guide [1].
Part (a):
* Spawn a Microsoft Azure NV-series instance with an NVidia GRID-supported GPU
- e.g. `NV36adms A10`
* Install updates, required tooling, and the desktop environment:
- sudo apt-get update
- sudo apt-get upgrade -y
- sudo apt-get dist-upgrade -y
- sudo apt-get install build-essential ubuntu-desktop -y
- sudo apt-get install linux-azure -y
* Disable nouveau kernel driver:
# Create a blacklist file /etc/modprobe.d/nouveau.conf with following contents:
blacklist nouveau
blacklist lbm-nouveau
* Reboot the VM, re-connect, and then stop X server:
- sudo reboot
# wait for the reboot, reconnect, and continue:
- sudo systemctl stop lightdm.service
* Download and install the NVidia GRID driver:
- wget -O NVIDIA-Linux-x86_64-grid.run https://go.microsoft.com/fwlink/?linkid=874272
- chmod +x NVIDIA-Linux-x86_64-grid.run
- sudo ./NVIDIA-Linux-x86_64-grid.run
- # When the setup asks whether you want to run the nvidia-xconfig utility to update your X configuration file, select Yes.
* Copy /etc/nvidia/gridd.conf.template to /etc/nvidia/gridd.conf
- sudo cp /etc/nvidia/gridd.conf.template /etc/nvidia/gridd.conf
* Edit /etc/nvidia/grid.conf
- sudo nano /etc/nvidia/grid.conf
# Append the following lines:
IgnoreSP=FALSE
EnableUI=FALSE
# Remove this line if present:
FeatureType=0
# And save.
* Reboot the VM
Part (b):
* Ensure that the hyperv_drm kernel module is loaded:
- sudo modprobe hyperv_drm
* Use the attached xorg.conf file to override /etc/X11/xorg.conf file
* try to start the `xserver`:
- sudo startx
* `xserver` should crash with a similar output to the following:
X.Org X Server 1.20.13
X Protocol Version 11, Revision 0
Build Operating System: linux Ubuntu
Current Operating System: Linux a10test 5.15.0-1031-azure #38~20.04.1-Ubuntu SMP Mon Jan 9 18:23:48 UTC 2023 x86_64
Kernel command line: BOOT_IMAGE=/boot/vmlinuz-5.15.0-1031-azure root=PARTUUID=4cac852b-afba-447b-b3e7-c002155c1305 ro console=tty1 console=ttyS0 earlyprintk=ttyS0 panic=-1
Build Date: 07 February 2023 12:48:13PM
xorg-server 2:1.20.13-1ubuntu1~20.04.6 (For technical support please see http://www.ubuntu.com/support)
Current version of pixman: 0.38.4
Before reporting problems, check http://wiki.x.org
to make sure that you have the latest version.
Markers: (--) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
(==) Log file: "/var/log/Xorg.1.log", Time: Sat Feb 18 10:54:26 2023
(==) Using config file: "/etc/X11/xorg.conf"
(==) Using system config directory "/usr/share/X11/xorg.conf.d"
(EE)
(EE) Backtrace:
(EE) 0: /usr/lib/xorg/Xorg (OsLookupColor+0x13c) [0x55e7787c5ecc]
(EE) 1: /lib/x86_64-linux-gnu/libpthread.so.0 (funlockfile+0x60) [0x7f9576cac420]
(EE) 2: /usr/lib/xorg/Xorg (xf86PlatformDeviceCheckBusID+0xa7) [0x55e7786c4db7]
(EE) 3: /usr/lib/xorg/Xorg (xf86PlatformMatchDriver+0x700) [0x55e7786bf1b0]
(EE) 4: /usr/lib/xorg/Xorg (xf86CallDriverProbe+0x5c) [0x55e7786971dc]
(EE) 5: /usr/lib/xorg/Xorg (xf86BusConfig+0x43) [0x55e778697b23]
(EE) 6: /usr/lib/xorg/Xorg (InitOutput+0x90b) [0x55e7786a59eb]
(EE) 7: /usr/lib/xorg/Xorg (InitFonts+0x1d4) [0x55e778667ea4]
(EE) 8: /lib/x86_64-linux-gnu/libc.so.6 (__libc_start_main+0xf3) [0x7f9576ac8083]
(EE) 9: /usr/lib/xorg/Xorg (_start+0x2e) [0x55e778651ace]
(EE)
(EE) Segmentation fault at address 0x124
(EE)
Fatal server error:
(EE) Caught signal 11 (Segmentation fault). Server aborting
(EE)
(EE)
Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
(EE) Please also check the log file at "/var/log/Xorg.1.log" for additional information.
(EE)
(EE) Server terminated with error (1). Closing log file.
^Cxinit: giving up
xinit: unable to connect to X server: Connection refused
xinit: unexpected signal 2
[ Where problems could occur ]
* The regression risk is low, given that the patch is well-isolated and basically adds a null check that is already assumed to be there in the first place.
[ Impact ]
* Microsoft Azure NV-series instances with NVidia GRID drivers started to experience xserver crashes while following Microsoft's official guide to installing Nvidia drivers [1].
* Root cause analysis showed that it was due to having a device with BusID "PCI:0@ <domain_ id>:0:0" , where domain id is >= 32767 while the hyperv_drm kernel module is loaded.
* Removing either the BusID specification or unloading the hyperv_drm kernel module seems to fix the crash.
* The crash is happening while X.server is trying to enumerate PCI devices. X.server dereferences a NULL pointer while trying to access to the PCI device info.
* The reason why it only happens while the hyperv_drm kernel module is loaded is that the hyperv_drm module does not expose PCI hardware information since it's a virtual device.
* The upstream patch [2] addresses the issue and it's confirmed that the xserver with the patch does not experience the crash.
* Ubuntu Focal `xorg-server` package does not include the patch [2] at the moment (xserver- xorg-core= 2:1.20. 13-1ubuntu1~ 20.04.6) .
[1]: https:/ /learn. microsoft. com/en- us/azure/ virtual- machines/ linux/n- series- driver- setup#install- grid-drivers- on-nv-or- nvv3-series- vms /github. com/freedesktop /xorg-xserver/ commit/ 0d93bbfa2cfacbb 73741f8bed0e32f a1a656b928
[2]: https:/
[ Test Plan ]
Part (a) is quoted from Microsoft's official guide [1].
Part (a):
* Spawn a Microsoft Azure NV-series instance with an NVidia GRID-supported GPU d/nouveau. conf with following contents: Linux-x86_ 64-grid. run https:/ /go.microsoft. com/fwlink/ ?linkid= 874272 Linux-x86_ 64-grid. run Linux-x86_ 64-grid. run gridd.conf. template to /etc/nvidia/ gridd.conf gridd.conf. template /etc/nvidia/ gridd.conf grid.conf grid.conf
- e.g. `NV36adms A10`
* Install updates, required tooling, and the desktop environment:
- sudo apt-get update
- sudo apt-get upgrade -y
- sudo apt-get dist-upgrade -y
- sudo apt-get install build-essential ubuntu-desktop -y
- sudo apt-get install linux-azure -y
* Disable nouveau kernel driver:
# Create a blacklist file /etc/modprobe.
blacklist nouveau
blacklist lbm-nouveau
* Reboot the VM, re-connect, and then stop X server:
- sudo reboot
# wait for the reboot, reconnect, and continue:
- sudo systemctl stop lightdm.service
* Download and install the NVidia GRID driver:
- wget -O NVIDIA-
- chmod +x NVIDIA-
- sudo ./NVIDIA-
- # When the setup asks whether you want to run the nvidia-xconfig utility to update your X configuration file, select Yes.
* Copy /etc/nvidia/
- sudo cp /etc/nvidia/
* Edit /etc/nvidia/
- sudo nano /etc/nvidia/
# Append the following lines:
IgnoreSP=FALSE
EnableUI=FALSE
# Remove this line if present:
FeatureType=0
# And save.
* Reboot the VM
Part (b):
* Ensure that the hyperv_drm kernel module is loaded: /boot/vmlinuz- 5.15.0- 1031-azure root=PARTUUID= 4cac852b- afba-447b- b3e7-c002155c13 05 ro console=tty1 console=ttyS0 earlyprintk=ttyS0 panic=-1 13-1ubuntu1~ 20.04.6 (For technical support please see http:// www.ubuntu. com/support) wiki.x. org Xorg.1. log", Time: Sat Feb 18 10:54:26 2023 xorg.conf" X11/xorg. conf.d" 0x13c) [0x55e7787c5ecc] 64-linux- gnu/libpthread. so.0 (funlockfile+0x60) [0x7f9576cac420] viceCheckBusID+ 0xa7) [0x55e7786c4db7] tchDriver+ 0x700) [0x55e7786bf1b0] Probe+0x5c) [0x55e7786971dc] 0x43) [0x55e778697b23] 64-linux- gnu/libc. so.6 (__libc_ start_main+ 0xf3) [0x7f9576ac8083] wiki.x. org Xorg.1. log" for additional information.
- sudo modprobe hyperv_drm
* Use the attached xorg.conf file to override /etc/X11/xorg.conf file
* try to start the `xserver`:
- sudo startx
* `xserver` should crash with a similar output to the following:
X.Org X Server 1.20.13
X Protocol Version 11, Revision 0
Build Operating System: linux Ubuntu
Current Operating System: Linux a10test 5.15.0-1031-azure #38~20.04.1-Ubuntu SMP Mon Jan 9 18:23:48 UTC 2023 x86_64
Kernel command line: BOOT_IMAGE=
Build Date: 07 February 2023 12:48:13PM
xorg-server 2:1.20.
Current version of pixman: 0.38.4
Before reporting problems, check http://
to make sure that you have the latest version.
Markers: (--) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
(==) Log file: "/var/log/
(==) Using config file: "/etc/X11/
(==) Using system config directory "/usr/share/
(EE)
(EE) Backtrace:
(EE) 0: /usr/lib/xorg/Xorg (OsLookupColor+
(EE) 1: /lib/x86_
(EE) 2: /usr/lib/xorg/Xorg (xf86PlatformDe
(EE) 3: /usr/lib/xorg/Xorg (xf86PlatformMa
(EE) 4: /usr/lib/xorg/Xorg (xf86CallDriver
(EE) 5: /usr/lib/xorg/Xorg (xf86BusConfig+
(EE) 6: /usr/lib/xorg/Xorg (InitOutput+0x90b) [0x55e7786a59eb]
(EE) 7: /usr/lib/xorg/Xorg (InitFonts+0x1d4) [0x55e778667ea4]
(EE) 8: /lib/x86_
(EE) 9: /usr/lib/xorg/Xorg (_start+0x2e) [0x55e778651ace]
(EE)
(EE) Segmentation fault at address 0x124
(EE)
Fatal server error:
(EE) Caught signal 11 (Segmentation fault). Server aborting
(EE)
(EE)
Please consult the The X.Org Foundation support
at http://
for help.
(EE) Please also check the log file at "/var/log/
(EE)
(EE) Server terminated with error (1). Closing log file.
^Cxinit: giving up
xinit: unable to connect to X server: Connection refused
xinit: unexpected signal 2
[ Where problems could occur ]
* The regression risk is low, given that the patch is well-isolated and basically adds a null check that is already assumed to be there in the first place.
[ Other Info ]
* workaround #1: unload hyperv_drm kernel module: /etc/X11/ vdisplay. edid"
- sudo modprobe -r hyperv_drm
* workaround #2: Comment out BusID line in /etc/X11/xorg.conf [Device] section:
Section "Device"
Identifier "Device0"
Driver "nvidia"
VendorName "NVIDIA Corporation"
# BusID "PCI:0@32828:0:0"
Option "HardDPMS" "false"
Option "CustomEDID" "DFP-0:
EndSection