----------------------------------------------
Original Message
From: "J"<jeebu19@yahoo.com>
Subject: Re: Q: deadlock
Date: Mon, 11 Mar 2002 12:09:46 +0530
>On Sun, Mar 10, 2002 at 05:51:33AM -0600, pwa wrote:
>>
>> I do not see why this is a problem since spinlocks are SMP only.
>> I thought the scheduler (called e.g implicetly by a timer interrupt)
would
>> eventually make the process holding the spinlock runable. Then this
process
>> would release the spinlock, and the busy waiting process wold enter the
>> critical region?
>
>Interrupt control path will not call scheduler, but while returning
>a check is made whether the interrupted control path was either
>user mode or VM86 mode. If neither the scheduler is bypassed.
>
>Regards,
>Sourav
Is a critical region protected by a spinlock considered an interrupt
control path?
The spin_lock function seems to be doing nothin more than a
while(locked)
;
Consider the following code:
spin_lock(&lock);
copy_to_user(buf, data, size);
spin_unlock(&lock);
AFAIK if the process (#1) holding the spinlock blocks while calling
copy_to_user, schedule is
called. The next process (#2) tries to enter the critical region, and spins
on the lock.
If/when the pages requested by the copy_to_user are retrived from swap, an
interrupt should occur.
Schedule should mark #1 as running, and #1 should release the lock.
Is the problem here the unlikely(?) situation that the pages are not
retrivable from swap?
-pwa.
_____________________________________________
Free email with personality! Over 200 domains!
http://www.MyOwnEmail.com
--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive: http://mail.nl.linux.org/kernelnewbies/
FAQ: http://kernelnewbies.org/faq/
