Re: Thread scheduler misbehaviour OR Oprofile bug?

2015-01-16 Thread Zoltan Kiss
On 16/01/15 19:37, William Cohen wrote: On 01/16/2015 01:29 PM, Zoltan Kiss wrote: On 16/01/15 15:38, William Cohen wrote: On 01/16/2015 09:01 AM, Zoltan Kiss wrote: Hi, I'm using OProfile to check some suspicious behaviour of dpdk-pktgen, and I can see something which troubles me. Either

Re: Thread scheduler misbehaviour OR Oprofile bug?

2015-01-16 Thread Zoltan Kiss
Hi, Another fun fact I forgot to mention: if I profile only this thread, and then run opreport with cpu filtering, it only shows results on the right cpu. So the wrong accounting only happens when I do system wide profiling. But not just with this thread and function, other functions appear a

Re: Thread scheduler misbehaviour OR Oprofile bug?

2015-01-16 Thread William Cohen
On 01/16/2015 01:29 PM, Zoltan Kiss wrote: > > > On 16/01/15 15:38, William Cohen wrote: >> On 01/16/2015 09:01 AM, Zoltan Kiss wrote: >>> Hi, >>> >>> I'm using OProfile to check some suspicious behaviour of dpdk-pktgen, >>> and I can see something which troubles me. Either the scheduler lies >>>

Re: Thread scheduler misbehaviour OR Oprofile bug?

2015-01-16 Thread Zoltan Kiss
On 16/01/15 15:38, William Cohen wrote: On 01/16/2015 09:01 AM, Zoltan Kiss wrote: Hi, I'm using OProfile to check some suspicious behaviour of dpdk-pktgen, and I can see something which troubles me. Either the scheduler lies about core affinity or Oprofile accounts some samples wrongly. This

Re: Thread scheduler misbehaviour OR Oprofile bug?

2015-01-16 Thread Zoltan Kiss
Hi, On 16/01/15 15:38, William Cohen wrote: On 01/16/2015 09:01 AM, Zoltan Kiss wrote: Hi, I'm using OProfile to check some suspicious behaviour of dpdk-pktgen, and I can see something which troubles me. Either the scheduler lies about core affinity or Oprofile accounts some samples wrongly. T

Re: Thread scheduler misbehaviour OR Oprofile bug?

2015-01-16 Thread William Cohen
On 01/16/2015 09:01 AM, Zoltan Kiss wrote: > Hi, > > I'm using OProfile to check some suspicious behaviour of dpdk-pktgen, > and I can see something which troubles me. Either the scheduler lies > about core affinity or Oprofile accounts some samples wrongly. > This userspace app runs in threads,

Re: Thread Scheduler

2014-03-24 Thread Andreas Mohr
Hi, > OK I read them and it seems that in order to change the scheduler, I have to > write a C code and call the sched_set* functions. Please correct that if I am > wrong. > > So in order to use my custom scheduler, I have to do the following steps. > > 1- Write my code using the APIs and sav

Re: Thread Scheduler

2014-03-23 Thread Mark Knecht
On Sun, Mar 23, 2014 at 7:27 AM, Mahmood Naderan wrote: > > > > > On Sunday, March 23, 2014 6:35 PM, Mark Knecht wrote: >>In terms of changing the scheduler I think that you possibly missed >>reading Documentation/block/switching-sched.txt in the kernel's >>documentation directory. >> >>HTH, >>Ma

Re: Thread Scheduler

2014-03-23 Thread Mahmood Naderan
On Sunday, March 23, 2014 6:35 PM, Mark Knecht wrote: >In terms of changing the scheduler I think that you possibly missed >reading Documentation/block/switching-sched.txt in the kernel's >documentation directory. > >HTH, >Mark That is IO scheduler. Can we use the same for threads/processes?

Re: Thread Scheduler

2014-03-23 Thread Mark Knecht
On Sun, Mar 23, 2014 at 4:01 AM, Mahmood Naderan wrote: > > > > > > >>On Thursday, March 20, 2014 6:53 AM, Mike Galbraith >> wrote: >> >>marge:~ # man -k setscheduler getscheduler >>sched_setscheduler (2) - set and get scheduling policy/parameters >>sched_setscheduler (3p) - set scheduling policy

Re: Thread Scheduler

2014-03-23 Thread Mahmood Naderan
>On Thursday, March 20, 2014 6:53 AM, Mike Galbraith >wrote: > >marge:~ # man -k setscheduler getscheduler >sched_setscheduler (2) - set and get scheduling policy/parameters >sched_setscheduler (3p) - set scheduling policy and parameters (REALTIME) >sched_getscheduler (2) - set and get sche

Re: Thread Scheduler

2014-03-19 Thread Mike Galbraith
On Wed, 2014-03-19 at 13:01 -0700, Mahmood Naderan wrote: > > One more question. Is it possible to dynamically change the scheduling > policy, e.g from fair to clock? marge:~ # man -k setscheduler getscheduler sched_setscheduler (2) - set and get scheduling policy/parameters sched_setscheduler (3

Re: Thread Scheduler

2014-03-19 Thread Mahmood Naderan
On Wednesday, March 19, 2014 4:36 PM, Jack Carrozzo wrote: >On Wed, Mar 19, 2014 at 7:34 AM, Mahmood Naderan wrote: > >> >> [...] >> Do you mean include/linux/sched.h ? >> Where is he implementation then? > >Your kernel source is incomplete. You can fetch a vanilla kernel >tarball from http:

Re: Thread Scheduler

2014-03-19 Thread Jack Carrozzo
On Wed, Mar 19, 2014 at 7:34 AM, Mahmood Naderan wrote: > > [...] > Do you mean include/linux/sched.h ? > Where is he implementation then? Your kernel source is incomplete. You can fetch a vanilla kernel tarball from http://kernel.org. jackc@kdev0 ~ $ ls /usr/src/linux-2.6.10/kernel/|grep sched

Re: Thread Scheduler

2014-03-19 Thread Mahmood Naderan
>On Wednesday, March 19, 2014 3:57 PM, Mike Galbraith > wrote: >On Wed, 2014-03-19 at 02:57 -0700, Mahmood Naderan wrote: >> >> Hi >> Maybe this is a noob question... >> Where in the linux kernel source (2.6), the >> >> thread scheduler has been implemented? Searching >> >> the web shows

Re: Thread Scheduler

2014-03-19 Thread Mike Galbraith
On Wed, 2014-03-19 at 02:57 -0700, Mahmood Naderan wrote: > > Hi > Maybe this is a noob question... > Where in the linux kernel source (2.6), the > > thread scheduler has been implemented? Searching > > the web shows that there should sched_fair.c but > > I can not find that in the source d