Every time I try to start a VM with this gpu passed through I get the attached error message and the VM fails to start, I have attempted to implement the vendor-reset (https://github.com/gnif/vendor-reset) and I am still receiving the same error, completely lost on what to do, any help would be very appreciated.
Trouble passing through my Rx 550
- Thread starter Voleris
- Start date
The message is not an error (or should not be) and to be expected as FLR does not work on that generation of AMD GPUs. You can ignore that (when your Proxmox is up to date).
Did you explicitly enable vendor-reset by writing
Sometimes passthrough just does not work with a device or motherboard. What log messages do you see in journalctl when starting the VM (please copy-paste the text in CODE-tags for readability and search-ability).
Did you explicitly enable vendor-reset by writing
device_specific to the reset_method (after each boot of the Proxmox host)? See https://github.com/gnif/vendor-reset/issues/46#issuecomment-992282166 for an example (and do try this first if you haven't).Sometimes passthrough just does not work with a device or motherboard. What log messages do you see in journalctl when starting the VM (please copy-paste the text in CODE-tags for readability and search-ability).
After writing device_specific I get a different error "Error: unable to read tail (got 0 bytes)"
Code:
Mar 14 22:23:43 wyse5070e kernel: oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=qemu.slice,mems_allowed=0,global_oom,task_memcg=/qemu.slice/100.scope,task=kvm,pid=3344,uid=0
Mar 14 22:23:43 wyse5070e kernel: Out of memory: Killed process 3344 (kvm) total-vm:8526664kB, anon-rss:7467508kB, file-rss:1792kB, shmem-rss:0kB, UID:0 pgtables:14928kB oom_score_adj:0
Mar 14 22:23:43 wyse5070e kernel: fwbr100i0: port 2(tap100i0) entered disabled state
Mar 14 22:23:43 wyse5070e kernel: tap100i0 (unregistering): left allmulticast mode
Mar 14 22:23:43 wyse5070e kernel: fwbr100i0: port 2(tap100i0) entered disabled state
Mar 14 22:23:43 wyse5070e systemd[1]: 100.scope: A process of this unit has been killed by the OOM killer.
Mar 14 22:23:43 wyse5070e systemd[1]: 100.scope: Failed with result 'oom-kill'.
Mar 14 22:23:43 wyse5070e systemd[1]: 100.scope: Consumed 7.959s CPU time.
Mar 14 22:23:43 wyse5070e systemd[1]: Stopping user@0.service - User Manager for UID 0...
Mar 14 22:23:43 wyse5070e systemd[2910]: Activating special unit exit.target...
Mar 14 22:23:43 wyse5070e systemd[2910]: Stopped target default.target - Main User Target.
Mar 14 22:23:43 wyse5070e systemd[2910]: Stopped target basic.target - Basic System.
Mar 14 22:23:43 wyse5070e systemd[2910]: Stopped target paths.target - Paths.
Mar 14 22:23:43 wyse5070e systemd[2910]: Stopped target sockets.target - Sockets.
Mar 14 22:23:43 wyse5070e systemd[2910]: Stopped target timers.target - Timers.
Mar 14 22:23:43 wyse5070e systemd[2910]: Closed dbus.socket - D-Bus User Message Bus Socket.
Mar 14 22:23:43 wyse5070e systemd[2910]: Closed dirmngr.socket - GnuPG network certificate management daemon.
Mar 14 22:23:43 wyse5070e systemd[2910]: Closed gpg-agent-browser.socket - GnuPG cryptographic agent and passphrase cache (access for web browsers).
Mar 14 22:23:43 wyse5070e systemd[2910]: Closed gpg-agent-extra.socket - GnuPG cryptographic agent and passphrase cache (restricted).
Mar 14 22:23:43 wyse5070e systemd[2910]: Closed gpg-agent-ssh.socket - GnuPG cryptographic agent (ssh-agent emulation).
Mar 14 22:23:43 wyse5070e systemd[2910]: Closed gpg-agent.socket - GnuPG cryptographic agent and passphrase cache.
Mar 14 22:23:43 wyse5070e systemd[2910]: Closed snapd.session-agent.socket - REST API socket for snapd user session agent.
Mar 14 22:23:43 wyse5070e systemd[2910]: Removed slice app.slice - User Application Slice.
Mar 14 22:23:43 wyse5070e systemd[2910]: Reached target shutdown.target - Shutdown.
Mar 14 22:23:43 wyse5070e systemd[2910]: Finished systemd-exit.service - Exit the Session.
Mar 14 22:23:43 wyse5070e systemd[2910]: Reached target exit.target - Exit the Session.
Mar 14 22:23:43 wyse5070e systemd[1]: user@0.service: Deactivated successfully.
Mar 14 22:23:43 wyse5070e systemd[1]: Stopped user@0.service - User Manager for UID 0.
Mar 14 22:23:43 wyse5070e systemd[1]: Stopping user-runtime-dir@0.service - User Runtime Directory /run/user/0...
Mar 14 22:23:43 wyse5070e pvestatd[1078]: VM 100 qmp command failed - VM 100 qmp command 'query-proxmox-support' failed - unable to connect to VM 100 qmp socket - Connection refused
Mar 14 22:23:43 wyse5070e pvedaemon[1090]: VM 100 qmp command failed - VM 100 qmp command 'query-proxmox-support' failed - unable to connect to VM 100 qmp socket - Connection refused
Mar 14 22:23:43 wyse5070e systemd[1]: run-user-0.mount: Deactivated successfully.
Mar 14 22:23:43 wyse5070e systemd[1]: user-runtime-dir@0.service: Deactivated successfully.
Mar 14 22:23:43 wyse5070e systemd[1]: Stopped user-runtime-dir@0.service - User Runtime Directory /run/user/0.
Mar 14 22:23:43 wyse5070e systemd[1]: Removed slice user-0.slice - User Slice of UID 0.
Mar 14 22:23:43 wyse5070e systemd[1]: qemu.slice: A process of this unit has been killed by the OOM killer.
Mar 14 22:23:43 wyse5070e pvedaemon[3332]: start failed: QEMU exited with code 1
Mar 14 22:23:44 wyse5070e pvestatd[1078]: status update time (8.647 seconds)
Mar 14 22:23:44 wyse5070e pvedaemon[1091]: <root@pam> end task UPID:wyse5070e:00000D04:0000F42F:67D4AC65:qmstart:100:root@pam: start failed: QEMU exited with code 1
Mar 14 22:23:58 wyse5070e pvedaemon[3477]: starting termproxy UPID:wyse5070e:00000D95:0000FE10:67D4AC7E:vncshell::root@pam:
Mar 14 22:23:58 wyse5070e pvedaemon[1092]: <root@pam> starting task UPID:wyse5070e:00000D95:0000FE10:67D4AC7E:vncshell::root@pam:
Mar 14 22:23:58 wyse5070e pvedaemon[1091]: <root@pam> successful auth for user 'root@pam'
Mar 14 22:23:58 wyse5070e login[3480]: pam_unix(login:session): session opened for user root(uid=0) by root(uid=0)
Mar 14 22:23:58 wyse5070e systemd-logind[657]: New session 9 of user root.
Mar 14 22:23:58 wyse5070e systemd[1]: Created slice user-0.slice - User Slice of UID 0.
Mar 14 22:23:59 wyse5070e systemd[1]: Starting user-runtime-dir@0.service - User Runtime Directory /run/user/0...
Mar 14 22:23:59 wyse5070e systemd[1]: Finished user-runtime-dir@0.service - User Runtime Directory /run/user/0.
Mar 14 22:23:59 wyse5070e systemd[1]: Starting user@0.service - User Manager for UID 0...
Mar 14 22:23:59 wyse5070e (systemd)[3486]: pam_unix(systemd-user:session): session opened for user root(uid=0) by (uid=0)
Mar 14 22:23:59 wyse5070e systemd[3486]: Queued start job for default target default.target.
Mar 14 22:23:59 wyse5070e systemd[3486]: Created slice app.slice - User Application Slice.
Mar 14 22:23:59 wyse5070e systemd[3486]: Reached target paths.target - Paths.
Mar 14 22:23:59 wyse5070e systemd[3486]: Reached target timers.target - Timers.
Mar 14 22:23:59 wyse5070e systemd[3486]: Starting dbus.socket - D-Bus User Message Bus Socket...
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on dirmngr.socket - GnuPG network certificate management daemon.
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on gpg-agent-browser.socket - GnuPG cryptographic agent and passphrase cache (access for web browsers).
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on gpg-agent-browser.socket - GnuPG cryptographic agent and passphrase cache (access for web browsers).
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on gpg-agent-extra.socket - GnuPG cryptographic agent and passphrase cache (restricted).
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on gpg-agent-ssh.socket - GnuPG cryptographic agent (ssh-agent emulation).
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on gpg-agent.socket - GnuPG cryptographic agent and passphrase cache.
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on snapd.session-agent.socket - REST API socket for snapd user session agent.
Mar 14 22:23:59 wyse5070e systemd[3486]: Listening on dbus.socket - D-Bus User Message Bus Socket.
Mar 14 22:23:59 wyse5070e systemd[3486]: Reached target sockets.target - Sockets.
Mar 14 22:23:59 wyse5070e systemd[3486]: Reached target basic.target - Basic System.
Mar 14 22:23:59 wyse5070e systemd[3486]: Reached target default.target - Main User Target.
Mar 14 22:23:59 wyse5070e systemd[3486]: Startup finished in 260ms.
Mar 14 22:23:59 wyse5070e systemd[1]: Started user@0.service - User Manager for UID 0.
Mar 14 22:23:59 wyse5070e systemd[1]: Started session-9.scope - Session 9 of User root.
Mar 14 22:23:59 wyse5070e login[3502]: ROOT LOGIN on '/dev/pts/0'I don't see any of the lines that I expect from vendor-reset in your log fragment. Did you load the vendor-reset module? How did you enable device_specific reset? A hookscript or cron-job/rc.locl might help.
I do see an out of memory (OOM) kill of VM 100. Know that PCI(e) passthrough requires all VM memory to be pinned into actual host RAM becuase of DMA. Try starting the VM with (much) less memory (or make more memory free beforehand).
You have not shown the VM configuration or much information about your host hardware, which might help to give more specific advice.
I do see an out of memory (OOM) kill of VM 100. Know that PCI(e) passthrough requires all VM memory to be pinned into actual host RAM becuase of DMA. Try starting the VM with (much) less memory (or make more memory free beforehand).
You have not shown the VM configuration or much information about your host hardware, which might help to give more specific advice.