This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: Stap scripts to track task cpu changes
- From: William Cohen <wcohen at redhat dot com>
- To: Ankita Garg <ankita at in dot ibm dot com>, William Cohen <wcohen at redhat dot com>, prerna at linux dot vnet dot ibm dot com, systemtap at sources dot redhat dot com
- Date: Mon, 13 Jul 2009 16:07:12 -0400
- Subject: Re: Stap scripts to track task cpu changes
- References: <200907130937.32782.varunc@linux.vnet.ibm.com> <20090713041811.GA3755@in.ibm.com>
Ankita Garg wrote:
> Hello William,
>
>> Subject: Re: Stap scripts to track task cpu changes
>> Date: Thursday 09 Jul 2009
>> From: William Cohen <wcohen@redhat.com>
>> To: Ankita Garg <ankita@in.ibm.com>, systemtap@sources.redhat.com,
>> prerna@linux.vnet.ibm.com
>>
>> Ankita Garg wrote:
>>> Hello folks,
>>>
>>> Quite sometime back, I was faced with a situation where I needed to
>>> track instances when a particular task was being migrated away from a
>>> cpu. I used the following trivial script. The tid needs to be passed as
>>> parameter. Sharing it, hoping it might be useful for some folks.
>> I would like to add these example to the systemtap examples. I have looked
>> through them and made some tweaks. For both of them I formatted them not to
>> use
>> tabs so they can used in documentation.
>>
>
> Thanks for making the required formatting changes.
>
>> For chng_cpu.stp use the appropriate probe point, scheduler.cpu_on. Why the
>> check for "tid() != 0"?
>>
>> For migrate.stp The check "$1 != 0" seemed to be unneeded. Was there a reason
>> for using $p->pid and kernel_string($p->comm) rather tid() and execname()?
>> The
>> test is using "tid() == $1"
>>
>
> True, the checks for tid() !=0 seem unneccesary. In migrate.stp (ie in
> function __migrate_task), the task being migrated need not be the
> current task. Hence, using $p->pid and $p->comm, where $p is the task being
> migrated.
>
>> -Will
Hi Ankita,
A bit more tweaking on the examples. For the chng_cpu2.stp I moved the
assignment to threads associative array inside the if statement to avoid
overflowing the array. For migrate3.stp I made it take the executable name, so
that it works in a manner similar to chng_cpu2.stp. Was there multiple instances
of the same execname running one the machine or would using the execname be
sufficent for migrate.stp?
-Will
/* Filename: chng_cpu.stp
* Author: Ankita Garg <ankita@in.ibm.com>
* Description: Captures information on the number of times an executable
* switches cpu
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* Copyright (C) IBM Corp. 2009. All Rights Reserved.
*
*/
global threads
probe scheduler.cpu_on
{
if ((threads[tid()] != cpu() && (execname() == @1))) {
printf("\nthread %d (%s) context switched on cpu%d state: %d\n",
tid(), execname(), cpu(), task_state(task_current()));
print_stack(backtrace());
threads[tid()] = cpu();
}
}
/* Filename: migrate.stp
* Author: Ankita Garg <ankita@in.ibm.com>
* Description: Captures information on the migration of a thread
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* Copyright (C) IBM Corp. 2009. All Rights Reserved.
*
*/
probe kernel.function("__migrate_task")
{
comm = kernel_string($p->comm);
if (comm == @1) {
printf ("thread %d (%s) is migrating from %d to %d \n",
$p->pid, comm, $src_cpu, $dest_cpu);
}
}