qemu-nbd kthread becomes defunct on disconnect
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
linux (Ubuntu) |
Fix Released
|
High
|
Andy Whitcroft | ||
Natty |
Fix Released
|
High
|
Andy Whitcroft |
Bug Description
When I do
qemu-nbd -n --connect /dev/nbd0 tmpXYZ.qcow2
I'm able to subsequently mount /dev/nbd0p1 as I expect, but the process listing then includes:
root 2753 1 0 17:47 ? 00:00:00 qemu-nbd -n --connect /dev/nbd0 tmpEXX8vS.qcow2
root 2754 2753 0 17:47 ? 00:00:00 [qemu-nbd] <defunct>
When I then try to disconnect with
qemu-nbd -d /dev/nbd0
that process hangs, and syslog shows it is hanging on a mutex in nbd_ioctl:
Jan 7 14:47:18 localhost kernel: [24459.139094] INFO: task nbd-client:29113 blocked for more than 120 seconds.
Jan 7 14:47:18 localhost kernel: [24459.139101] "echo 0 > /proc/sys/
Jan 7 14:47:18 localhost kernel: [24459.139106] nbd-client D 0000000000000005 0 29113 29112 0x00000000
Jan 7 14:47:18 localhost kernel: [24459.139115] ffff8801d1215d98 0000000000000082 ffff8801d1215fd8 ffff8801d1214000
Jan 7 14:47:18 localhost kernel: [24459.139123] 0000000000013a80 ffff8801b1cd47e0 ffff8801d1215fd8 0000000000013a80
Jan 7 14:47:18 localhost kernel: [24459.139131] ffff88021f5996c0 ffff8801b1cd4440 ffffea00068737f8 ffffffffa001a020
Jan 7 14:47:18 localhost kernel: [24459.139139] Call Trace:
Jan 7 14:47:18 localhost kernel: [24459.139176] [<ffffffff815b9
Jan 7 14:47:18 localhost kernel: [24459.139189] [<ffffffff81132
Jan 7 14:47:18 localhost kernel: [24459.139198] [<ffffffff815b9
Jan 7 14:47:18 localhost kernel: [24459.139211] [<ffffffffa0018
Jan 7 14:47:18 localhost kernel: [24459.139219] [<ffffffff812cb
Jan 7 14:47:18 localhost kernel: [24459.139227] [<ffffffff81196
Jan 7 14:47:18 localhost kernel: [24459.139234] [<ffffffff81175
Jan 7 14:47:18 localhost kernel: [24459.139242] [<ffffffff81164
Jan 7 14:47:18 localhost kernel: [24459.139248] [<ffffffff81175
Jan 7 14:47:18 localhost kernel: [24459.139255] [<ffffffff8100c
This does NOT happen with lucid userspace on lucid kernel (with recent updates, but not completely uptodate)
This does happen in plain natty.
It does happen with the latest mainline kernel-ppa.
It does happen in a lucid userspace chroot on natty kernel - almost. The qemu-nbd process does not hang in that case, but defunct qemu-nbd kthread stays around, as does the user qemu-nbd connect thread.
tags: | added: iso-testing |
Confirmed. I'm seeing this also on natty.