[Crash-utility] Add an option "-c" for commamd runq
Dave Anderson
anderson at redhat.com
Tue May 22 14:09:07 UTC 2012
----- Original Message -----
> Hi Dave,
> As you mentioned in last email, sometimes, we really need to compare
> timestamps of runqueue objects with those of current task's task_struct
> objects .
> The usecase where such feature is needed is the situation where curcent
> processes in crash dump is in the state with interrupt disabled. Then timer
> interrupts is also disabled, so there's clear difference between CPU's timestamp
> and current task's timestamp. The latter can be abtainned easily by using ps -l.
> But the field names of timestamp values in runqueue objects vary according to
> kernel versions, it's inconvenient and troublesome to get needed information
> about runqueue's timestamp for users. That's why adding an option to runq
> command.
> Any question please contact with me.
Right -- so this information is really only helpful if it is in collaboration
with the timestamp of the active task on each cpu.
But it seems counterintuitive to have to "runq -c" to show the timestamps, and
then have to go run "ps -l" to dig out out the active command that is running on
that cpu, and then compare them.
I'm thinking that it might make more sense to either:
(1) have the "runq -c" (or maybe "-t"?) option show each cpu's runqueue timestamp
followed by the timestamp of the active task on that cpu, or
(2) have the "ps -l" (or maybe a new "ps -L"?) option that first displays the
per-cpu timestamps, followed by the sorted tasks' timstamps
I think I prefer (1). What do you think?
Dave
>
> Regards,
> Wei Jiangang
>
> ----- Original Message -----
> > Hello Dave,
> > The patch is used to improve the existing command, which adds an
> option "-c"
> > to the the command "runq". The option "-c" displays the timestamp
> information about
> > the runqueue each of cpu. The timstamp information, which
> > accordings to
> the kernel
> > version you use, may be the value of runqueue's member. for
> > example,
> rq.clock or
> > rq.timestamp_last_tick, even runqueue.timestamp_last_tick.
> > Besides, I have already tested it on rhel4.8/rehl5.7/rhel6.2,
> without any error.
> > The result like this:
> > > crash> runq -c
> > > CPU 0: 66789719522523
> > > CPU 1: 66789719628339
> > > CPU 2: 66789719519625
> >
> > Regards,
> > Wei Jiangang
>
> Wei,
>
> What is the debugging scenario that you feel the need to know the
> timestamp of each cpu's runqueue? Perhaps you want to compare it to
> the "ps -l" output?
>
> Dave
>
> --
> Crash-utility mailing list
> Crash-utility at redhat.com
> https://www.redhat.com/mailman/listinfo/crash-utility
>
>
> --
> Crash-utility mailing list
> Crash-utility at redhat.com
> https://www.redhat.com/mailman/listinfo/crash-utility
>
More information about the Crash-utility
mailing list