Re: Xen dom0 freeze after domU exits (was Re: Zombie kernel thread)

[manu%netbsd.org@localhost (Emmanuel Dreyfus)]

Emmanuel Dreyfus <manu%netbsd.org@localhost> wrote:

> I follow up on my previous report: On XEN3_DOM0 kernel I very often freeze the
> kernel after a domU exits. 

I think I found a second problem, which seems to be reproductible if I
run a domU with two block devices configured and this stress test:
while true; xl create -x domu.conf; sleep 60; xl shutdown -w  domu; done

After a while the system will freeze. Here are significant processes:

PID    LID S CPU     FLAGS       STRUCT LWP *               NAME WAIT
6309     1 3   0         0   ffffa00002093700           vnconfig biowait
6846     1 3   0        80   ffffa00000d1a080           vnconfig fstcnt
0      243 3   0       200   ffffa0000213e340               vnd1 fstchg
0      242 3   0       200   ffffa000024168a0               vnd0 vndbp


vnconfig of PID 6309 backtrace:
sleepq_block 
cv_wait 
biowait
convertdisklabel
validate_label
readdisklabel 
vndopen
spec_open
VOP_OPEN
vn_open
do_open
do_sys_openat
sys_open
syscall()

That process is waiting for an I/O inside a fs transaction (started
where?)

db> call fstrans_dump
Fstrans locks by lwp:
6309.1   (/) shared 1 cow 0
Fstrans state by mount:
/                state suspended


But the I/O should be performed by vndthread of PID 0.243 which is
itself waiting to start a fs transaction. Here is the backtrace:
sleepq_block
cv_wait
fstrans_start
genfs_do_putpages
VOP_PUTPAGES
vndthread

That looks like a deadlock, but for now I have no idea how to fix it.

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu%netbsd.org@localhost