Hi Alex,

On 1/12/20 8:46 AM, Alexander Lyakas wrote:

Hi Guoqing,

We have tried the proposed patch, but still hit the hung-task panic in
the same place[1]. How can we debug this further?

Can you print the information (atomic_read(&conf->active_stripes),
conf->max_nr_stripes and conf->cache_state)  just before the
wait_event_lock_irq? Then we can verify which condition is not meet
here. Also pls dump the value of sysfs nodes, something like.

linux:/sys/block/md5/md # find -mindepth 1 -maxdepth 1 -type 
f|sort|xargs -r egrep .


BTW, could you try latest kernel? And is it possible to reproduce the panic?

Thanks,
Guoqing