by having a vmid <-> pciid mapping in /var/run i did not check if the vm has the pci device really in the config, but we should not need that, since we remove the reservation again in the cleanup step.
if wanted we can of course parse the target vms config and check if the pci device is still configured, or alternatively, ask qmp and or parse the /proc/PID/cmdline for the pcidevice, but both options seem too expensive? Dominik Csapak (3): pci: to not capture first group in PCIRE pci: add helpers to (un)reserve pciids for a vm fix #3258: block vm start when pci device is already in use PVE/QemuServer.pm | 8 ++++ PVE/QemuServer/PCI.pm | 91 ++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 98 insertions(+), 1 deletion(-) -- 2.30.2 _______________________________________________ pve-devel mailing list pve-devel@lists.proxmox.com https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
