Hi,
I'm using gpl PV drivers 0.8.1 drivers for HVM winxp and sometimes
it boots forever.
From the iostat in dom0 with xen-3.1.2 is see that there is NO access to
the underlying lvm volume of this winxp HVM.
From strace i see that qemu-dm indefinetly loops with:
read(14, "\34\0\0\0", 4) = 4
write(14, "\34\0\0\0", 4) = 4
clock_gettime(CLOCK_MONOTONIC, {53840, 172326473}) = 0
ioctl(14, EVIOCGKEYCODE, 0x7fff012f0fc0) = 0
clock_gettime(CLOCK_MONOTONIC, {53840, 172398473}) = 0
clock_gettime(CLOCK_MONOTONIC, {53840, 172436473}) = 0
clock_gettime(CLOCK_MONOTONIC, {53840, 172474473}) = 0
select(18, [7 10 12 14 17], [], [], {0, 10000}) = 1 (in [14], left {0, 10000})
and
lsof -p 6602 | awk '$4 ~ /(7|10|12|14|17)/'
qemu-dm 6602 root 7u CHR 10,200 1853
/dev/net/tun
qemu-dm 6602 root 10u CHR 5,2 2035
/dev/ptmx
qemu-dm 6602 root 12r FIFO 0,5 14646
pipe
qemu-dm 6602 root 14u CHR 10,63 1819
/dev/xen/evtchn
qemu-dm 6602 root 17u IPv4 19833 TCP
clustrix20.cs.bgu.ac.il:5911->lesbinux.cs.bgu.ac.il:51837 (ESTABLISHED)
and qemu-dm takes about 10% of cpu
while others HVM qemu-dm, which boot ok, loop indefinetly with:
select(14, [6 8 9 11 13], [], [], {0, 10000}) = 0 (Timeout)
clock_gettime(CLOCK_MONOTONIC, {54627, 313284473}) = 0
clock_gettime(CLOCK_MONOTONIC, {54627, 313366473}) = 0
clock_gettime(CLOCK_MONOTONIC, {54627, 313450473}) = 0
clock_gettime(CLOCK_MONOTONIC, {54627, 313558473}) = 0
if i reboot this HVM it boots ok , and may boot ok next few times untill
it gets stuck the same way on some reboot/start
I'm not 100% sure it never happens without the gpl PV rivers.
Any ideas?
Alex
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
Home