lkml.org 
[lkml]   [2004]   [Feb]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC][PATCH] O(1) Entitlement Based Scheduler
Hi!

> CPU usage rate caps
> -------------------
>
> A task's CPU usage rate cap imposes a soft (or hard) upper limit on the rate at
> which it can use CPU resources and can be set/read via the files
>
> /proc/<pid>/cpu_rate_cap
> /proc/<tgid>/task/<pid>/cpu_rate_cap
>
> Usage rate caps are expressed as rational numbers (e.g. "1 / 2") and hard caps
> are signified by a "!" suffix. The rational number indicates the proportion
> of a single CPU's capacity that the task may use. The value of the number must
> be in the range 0.0 to 1.0 inclusive for soft caps. For hard caps there is an
> additional restriction that a value of 0.0 is not permitted. Tasks with a
> soft cap of 0.0 become true background tasks and only get to run when no other
> tasks are active.

Why not use something like percent, parts per milion or whatever?

> When hard capped tasks exceed their cap they are removed from the run queues
> and placed in a "sinbin" for a short while until their usage rate decays to
> within limits.

How do you solve this one?

I want to kill your system.

I launch task A, "semaphore grabber", that does filesystem
operations. Those need semaphores. I run it as "true background".

I wait for A to grab some lock, then I run B, which is while(1);

A holds lock that can not be unlocked, and your system is dead.

This may happen randomly, even without me on your system.
Pavel
--
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:01    [W:0.132 / U:2.124 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site