Re: [libvirt] [PATCH] qemu: Forcibly mknod() even if it exists

On 11/20/19 11:05 AM, Daniel P. Berrangé wrote: > On Wed, Nov 20, 2019 at 10:51:47AM +0100, Michal Privoznik wrote: > > Another weird bug appeared concerning qemu namespaces. Basically > > the problem is as follows: > > > > 1) Issue an API that causes libvirt to create a node in domain's > > namespace, say /dev/sda with 8:0 as major:minor (the API can > > be attach-disk for instance). Or simply create the node from > > a console by hand. > > > > 2) Detach the disk from qemu. > > > > 3) Do something that makes /dev/sda change it's minor number. > > Wait, what ? > > major/minor numbers for SCSI disks are a defined standard > IIUC > > $ grep SCSI ./admin-guide/devices.txt | grep block > 8 block SCSI disk devices (0-15) > 11 block SCSI CD-ROM devices > 65 block SCSI disk devices (16-31) > 66 block SCSI disk devices (32-47) > 67 block SCSI disk devices (48-63) > 68 block SCSI disk devices (64-79) > 69 block SCSI disk devices (80-95) > 70 block SCSI disk devices (96-111) > 71 block SCSI disk devices (112-127) > 128 block SCSI disk devices (128-143) > 129 block SCSI disk devices (144-159) > 130 block SCSI disk devices (160-175) > 131 block SCSI disk devices (176-191) > 132 block SCSI disk devices (192-207) > 133 block SCSI disk devices (208-223) > 134 block SCSI disk devices (224-239) > 135 block SCSI disk devices (240-255) > > > IOW, /dev/sda should always be 8,0 > > There is, however, the possibiity of dynamically assign > major/minor numbers so its possible we'll see a block > device with a changable number, but AFAIK, such a block > device should never be call /dev/sda !??! > > IOW, the commit is fine, but is this commit message > really accurate ? Ah, the bug talks about /dev/nvmeN that has changed MIN number. I haven't found corresponding docs on NVMe though. Is s/sda/nvmeN/ on the commit message enough?