RE: [PATCH] kernel/hung_task.c: force ignore_loglevel before panic

[Liu, Chuansheng]

> -----Original Message-----
> From: Petr Mladek [mailto:pmladek@xxxxxxxx]
> Sent: Wednesday, December 12, 2018 6:18 PM
> To: Liu, Chuansheng <chuansheng.liu@xxxxxxxxx>
> Cc: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>; Sergey Senozhatsky
> <sergey.senozhatsky.work@xxxxxxxxx>; akpm@xxxxxxxxxxxxxxxxxxxx;
> sergey.senozhatsky@xxxxxxxxx; rostedt@xxxxxxxxxxx; dvyukov@xxxxxxxxxx;
> linux-kernel@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH] kernel/hung_task.c: force ignore_loglevel before panic
> 
> On Wed 2018-12-12 01:16:11, Liu, Chuansheng wrote:
> >
> >
> > > -----Original Message-----
> > > From: Tetsuo Handa [mailto:penguin-kernel@xxxxxxxxxxxxxxxxxxx]
> > > Sent: Tuesday, December 11, 2018 6:02 PM
> > > To: Liu, Chuansheng <chuansheng.liu@xxxxxxxxx>; Sergey Senozhatsky
> > > <sergey.senozhatsky.work@xxxxxxxxx>
> > > Cc: akpm@xxxxxxxxxxxxxxxxxxxx; pmladek@xxxxxxxx;
> > > sergey.senozhatsky@xxxxxxxxx; rostedt@xxxxxxxxxxx;
> dvyukov@xxxxxxxxxx;
> > > linux-kernel@xxxxxxxxxxxxxxx
> > > Subject: Re: [PATCH] kernel/hung_task.c: force ignore_loglevel before panic
> > >
> > > On 2018/12/11 10:16, Liu, Chuansheng wrote:
> > > > We may enhance it by:
> > > > -       if (sysctl_hung_task_warnings) {
> > > > +       if (sysctl_hung_task_panic || sysctl_hung_task_warnings) {
> > > >                 if (sysctl_hung_task_warnings > 0)
> > > >                         sysctl_hung_task_warnings--;
> > >
> > > Why ignore sysctl_hung_task_warnings? The administrator can already
> > > configure as sysctl_hung_task_warnings == -1 && sysctl_hung_task_panic
> == 1
> > > if he/she does not want to suppress neither sched_show_task() nor
> > > debug_show_all_locks()/trigger_all_cpu_backtrace(). Someone might want
> that
> > > sysctl_hung_task_warnings == 0 (which is a request to suppress only
> > > sched_show_task()) should not be ignored by sysctl_hung_task_panic == 1
> > > (which is a request to trigger panic).
> >
> >
> > My complete idea is in patch V1 which has been sent. Paste here:
> > If sysctl_hung_task_panic == 1, I will force sched_show_task(t) and set
> > hung_task_call_panic = true
> > hung_task_show_lock = true
> 
> Please, do not mix two changes into one patch.
Thanks your suggestion.

> 
> Add console_verbose() in one patch. It is simple and
> everyone has agreed with it so far.
Has sent it out, please help review.

> 
> Force sched_show_task() when hung_task_call_panic == 1 in
> another patch. It seems to be controversial and should be
> discussed/changed separately.
I found some other points also, and will send out patches later.