Re: [announce, patch, RFC] "big IRQ lock" removal, IRQ cleanups.

From: Oleg Nesterov (oleg@tv-sign.ru)
Date: Mon Jul 22 2002 - 09:45:02 EST


Hello.

> > [...] To do this from irq.c means that it must exit with interrupts off
> > and the the low level code needs to keep them off till the irtn. [...]

> yes, we are very careful to keep irqs disabled in do_IRQ(), both before
> and after calling the handler.

Note that smp_xxx_interrupt() functions must be carefull
with preemt_{disable,enable} brackets.

For example, smp_invalidate_interrupt() may be preempted
after put_cpu(). Probably not big deal (it is return path),
but it is better to use preempt_enable_no_resched() here -
let ret_from_intr: do its job.

smp_{error,spurious,thermal}_interrupt() - all of them
use printk() without bumping preemt_count and have problem
after spin_unlock_irqrestore(&logbuf_lock, flags).

If these problems worth fixing, then preempt_stop (cli)
can be killed in entry.S:ret_from_intr(), yes? If i understand
correctly none of the irq handlers should return to low level
code with irq enabled.

P.S.

May I suggest somebody with good english fix
Documentation/preempt-locking.txt?
It states, that disabled interrupts prevents preemption.
Yes, but only in a sense, that the delivery of reschedule
interrupt is suppressed.

Process with irqs disabled and current->preempt_count == 0 can
be preempted (with interrupts enabled) after spin_lock/unlock etc.
Even in UP case preemption can happen while calling wake_up_...().

Oleg.
-
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/



This archive was generated by hypermail 2b29 : Tue Jul 23 2002 - 22:00:39 EST