LKML Archive on
help / color / mirror / Atom feed
From: Davide Libenzi <>
To: Russell King <>
Cc: Linux Kernel List <>
Subject: Re: scheduler: IRQs disabled over context switches
Date: Sun, 23 May 2004 11:59:01 -0700 (PDT)	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <>

On Sun, 23 May 2004, Russell King wrote:

> The 2.6.6 scheduler disables IRQs across context switches, which is
> bad news for IRQ latency on ARM - to the point where 16550A FIFO
> UARTs to overrun.
> I'm considering defining prepare_arch_switch & co as follows on ARM,
> so that we release IRQs over the call to context_switch().
> #define prepare_arch_switch(rq,next)		\
> do {						\
> 	spin_lock(&(next)->switch_lock);	\
> 	spin_unlock_irq(&(rq)->lock);		\
> } while (0)
> #define finish_arch_switch(rq,prev)		\
> 	spin_unlock(&(prev)->switch_lock)
> #define task_running(rq,p)			\
> 	((rq)->curr == (p) || spin_is_locked(&(p)->switch_lock))
> The question is... why are we keeping IRQs disabled over context_switch()
> in the first case?  Looking at the code, the only thing which is touched
> outside of the two tasks is rq->prev_mm.  Since runqueues are CPU-
> specific and we're holding at least one spinlock, I think the above
> is preempt safe and SMP safe.

Other archs already do the above. The only reason I can think of is an 
optimization issue. The above code does a spin_lock/unlock more than the 
default code. For archs where the ctx switch is fast, this might matter.
Whereas archs with slow ctx switch might want to use the above code to 
reduce IRQ latency. Also, if the ctx switch code involves acquiring some 
"external" spinlock, care should be taken to verify that cross-lock 
happens (see ia64 code for example). IMO, if this is an optimization 
issue, and if the thing does not buy us much in terms of performance, I'd 
rather use the above code as default. Ingo?

- Davide

  reply	other threads:[~2004-05-23 18:59 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-05-23 16:43 Russell King
2004-05-23 18:59 ` Davide Libenzi [this message]
2004-05-23 19:38   ` Russell King
2004-05-23 23:04     ` Davide Libenzi
2004-05-23 23:33       ` Russell King
2004-05-24  0:27         ` Davide Libenzi
2004-05-24  8:37 ` Ingo Molnar
2004-05-24  6:41   ` Davide Libenzi
2004-05-24  9:05     ` Ingo Molnar
2004-05-24  7:10       ` Nick Piggin
2004-05-24  9:15         ` Ingo Molnar
2004-05-24 17:16       ` Davide Libenzi
2004-05-24 17:46         ` Davide Libenzi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \
    --subject='Re: scheduler: IRQs disabled over context switches' \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).