Strange time behavior (domU migration)

[George Shuklin <george.shuklin@gmail.com>]

Using 2.6.34 from Gentoo (from SUSE), xen 3.4.2 (within XCP).

Found strange time behaviour on domU after migration - it looks like
time frozen in some counter.

PING 109.234.152.2 (109.234.152.2) 56(84) bytes of data.
64 bytes from 109.234.152.2: icmp_seq=1 ttl=64 time=0.000 ms
64 bytes from 109.234.152.2: icmp_seq=1 ttl=64 time=0.000 ms
64 bytes from 109.234.152.2: icmp_seq=1 ttl=64 time=0.000 ms
... (and so on, it's not so fast network, it's strange kernel behavior)

Some output displays on screen only after input (like TCP retransmission
occur).

DMESG displays following after migration:

RCU detected CPU stalls: 1 (detected by 0, t=47514 jiffies)

I tried to test both workarounds from debian wiki: 

xen.independent_wallclock=1
set /sys/devices/system/clocksource/clocksource0/current_clocksource to
jiffes

and 

xen.independent_wallclock=0
set /sys/devices/system/clocksource/clocksource0/current_clocksource to
jiffes


But bug continues... I don't really understand what happens and how to
fix it...