RE: FW: problem setting up affinity mask for irq intrupt

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Thanks a lot,

This(http://lxr.linux.no/linux+v2.6.36/Documentation/ia64/IRQ-redir.txt)
should be the next step after finding the solution for the first problem
(that is just using core0) !!! :D

I found something which is quite related
(http://www.alexonlinux.com/smp-affinity-and-proper-interrupt-handling-in-li
nux ) this article says that it may not be possible to change irq mask in
the systems with CONFIG_HOTPLUG_CPU enabled! 
Now I have 3 problems:

1) I cannot find out the relationship between CONFIG_HOTPLUG_CPU and irq
affinity.
2) I only have problem with eth0 and eth1 ( for example for the timer
interrupt I can change the mask and it is effective! Although it only uses
the first enabled core in the mask, which is related to IA64 feature)
3) I don`t know how to disable CONFIG_HOTPLUG_CPU !

If you can help me with solving any of these problems I would really
appreciate it :)

Regards,
RSH

-----Original Message-----
From: Mulyadi Santosa [mailto:mulyadi.santosa@xxxxxxxxx] 
Sent: Tuesday, November 30, 2010 10:56 AM
To: Raoufehsadat Hashemian Harandi
Cc: kernelnewbies@xxxxxxxxxxxx
Subject: Re: FW: problem setting up affinity mask for irq intrupt

On Wed, Dec 1, 2010 at 00:39, Raoufehsadat Hashemian Harandi
<rhashem@xxxxxxxxxxx> wrote:
> Yes, it is running, I also putted a heavy load with around 10000 interrupt
> per second to see if it changes the number but the output is still like
> this:

may we see the content of /etc/sysconfig/irqbalance? who knows, there
is a clue there...


> The interesting point is that the value of affinity mask changes
frequently,
> when I set it to FFFFFFFF it is FF for some seconds and then it changes to
> other values like 02 04 08 !

huh? dynamically changing affinity? weird...

BTW, please kindly check
http://lxr.linux.no/linux+v2.6.36/Documentation/ia64/IRQ-redir.txt,
perhaps it would give another clue for your case

PS: right now, IMHO, putting aside the "fact" that your affinitiy mask
could change automatically somehow, I am guessing that the irq is
handled that way due to some internal calculation. To make it short, I
read somewhere that by sticking interrupt to a CPU core, it actually
speed up response.

of course, this is not always true for every situation, but kernel
does exactly that at the first place. Also, by doing that, per cpu
cache hotness would likely be maintained. Interrupt handler always
running, L1/L2/L3 cache is effectively utilized....everybody happy :D

-- 
regards,

Mulyadi Santosa
Freelance Linux trainer and consultant

blog: the-hydra.blogspot.com
training: mulyaditraining.blogspot.com




--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ



[Index of Archives]     [Newbies FAQ]     [Linux Kernel Mentors]     [Linux Kernel Development]     [IETF Annouce]     [Git]     [Networking]     [Security]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux RAID]     [Linux SCSI]     [Linux ACPI]
  Powered by Linux