Re: soft lockups of tapdisk2 process - xen 4.4, kernel 3.12.17

[Eugene Istomin <E.Istomin@edss.ee>]

We compare 3.12/3.13/3.14 kernels (keyword - PREEMPT_NEED_RESCHED), 3.12 
haven't any patches for related to PREEMPT_NEED_RESCHED:

3.12 - https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/log/?id=refs%2Ftags%2Fv3.12.17&qt=grep&q=PREEMPT_NEED_RESCHED 

3.13 - https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/log/?id=refs%2Ftags%2Fv3.13.10&qt=grep&q=PREEMPT_NEED_RESCHED 

3.14 - https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/log/?id=refs%2Ftags%2Fv3.14.1&qt=grep&q=kmemleak.c 


What do you think about commit ba1f14fbe70965ae0fb1655a5275a62723f65b77  
(https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=ba1f14fbe70965ae0fb1655a5275a62723f65b77 
)? May this problem related to interrupt handling?

[ 6624.928058] Call Trace:
[ 6624.928058]  [<ffffffff8036d192>] notify_remote_via_ipi+0x82/0xb0
..



---
Best regards,
Eugene Istomin


On Thursday, April 17, 2014 08:11:25 AM Eugene Istomin wrote:

Hello,

after updating from xen 4.3 to xen 4.4 we have strange soft lockups, here is 
log - https://gist.github.com/evis/10953677.

We already disabled c-states in BIOS, kernel (processor.max_cstate=1 
intel_idle.max_cstate=0), dom0 (xenpm set-max-cstate 1). An easy step to 
reproduce in our environment - run 2 VMs with DRBD-backed disk.


What "blktap_run_deferred" in this trace mean and why dom0 needs to call 
"blktap_device_restart"?

force_evtchn_callback+0xe/0x10
blktap_device_restart+0xdf/0xf0 [blktap2]
blktap_run_deferred+0xd1/0xe0 [blktap2]
blktap_ring_ioctl+0x3a5/0x4d0 [blktap2]

Thanks.

---
Best regards,
Eugene Istomin