virtio block devices names are not recycled

[Scott Moser]

I'm playing with eucalyptus, and attaching and detaching a block device as a virtio (bug 588410)
$ euca-attach-volume --instance ${inst} --device /dev/vdc ${vol}
VOLUME vol-5EC30651

Then,

$ euca-detach-volume ${vol}

The device names start as /dev/vda and go through to /dev/vdz and then to /dev/vdaa ...

The names are not recycled it would appear. In other buses (scsi) I'm used to the device name being recycled.

This can most likely be re-created without eucalyptus, using only kvm, but I dont' have those steps right at hand.

ProblemType: Bug
DistroRelease: Ubuntu 10.10
Package: linux-image-2.6.34-4-virtual 2.6.34-4.11
Regression: No
Reproducible: Yes
ProcVersionSignature: User Name 2.6.34-4.11-virtual 2.6.34
Uname: Linux 2.6.34-4-virtual x86_64
AlsaDevices: Error: command ['ls', '-l', '/dev/snd/'] failed with exit code 2: ls: cannot access /dev/snd/: No such file or directory
AplayDevices: Error: [Errno 2] No such file or directory
Architecture: amd64
ArecordDevices: Error: [Errno 2] No such file or directory
Date: Tue Jun 8 21:29:41 2010
Lsusb: Bus 001 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
MachineType: Bochs Bochs
PciMultimedia:

ProcCmdLine: root=/dev/sda1 console=ttyS0
ProcEnviron:
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: linux
dmi.bios.date: 01/01/2007
dmi.bios.vendor: Bochs
dmi.bios.version: Bochs
dmi.chassis.type: 1
dmi.chassis.vendor: Bochs
dmi.modalias: dmi:bvnBochs:bvrBochs:bd01/01/2007:svnBochs:pnBochs:pvr:cvnBochs:ct1:cvr:
dmi.product.name: Bochs
dmi.sys.vendor: Bochs

[Jeremy Foshee (jeremyfoshee)]

Hi Scott,

If you could also please test the latest upstream kernel available that would be great. It will allow additional upstream developers to examine the issue. Refer to https://wiki.ubuntu.com/KernelMainlineBuilds . Once you've tested the upstream kernel, please remove the 'needs-upstream-testing' tag. This can be done by clicking on the yellow pencil icon next to the tag located at the bottom of the bug description and deleting the 'needs-upstream-testing' text. Please let us know your results.

Thanks in advance.

    [This is an automated message. Apologies if it has reached you inappropriately; please just reply to this message indicating so.]

[Scott Moser (smoser)]

I do not really have the time to test the upstream kernel. However, looking at the source of drivers/block/virtio_blk.c , I would guess that this bug does exist upstream. My (possibly to quick) read of virtblk_probe and virtblk_remove seem to me to be indicating the static variable "index" and never decrementing it. That variable is what is used to determine the next name in virtblk_probe.

[Scott Moser (smoser)]

2 things.
a.) i just realized that the simple 'index--' wouldn't suffice as that would result in:
- add vda
- add vdb
- add vdc
- remove vdb
- add vdc !!
b.) i did reproduce with the upstream build linux-image-2.6.34-020634-generic .

[John Johansen (jjohansen)]

Scott,

your reading of the code is correct. Currently virtio_blk is not designed to recycle names, and this won't be possible until upstream changes this.

closing as invalid as this is the expected behavior.

[Scott Moser (smoser)]

On Mon, 14 Jun 2010, John Johansen wrote:

> your reading of the code is correct. Currently virtio_blk is not
> designed to recycle names, and this won't be possible until upstream
> changes this.

So if I wanted that behavior, my option is basically to write a patch and
send to lkml ?

[John Johansen (jjohansen)]

Yes unfortunately. Though even if you had a patch they may not take it. The correct next step would be an email to the kvm development list (<email address hidden>) to discuss the behavior and see if such a change has any chance of going upstream.