Re: [patch 1/6] hardirq: Make hardirq bits generic

[Thomas Gleixner]

On Tue, 17 Sep 2013, Thomas Gleixner wrote:

> On Tue, 17 Sep 2013, Geert Uytterhoeven wrote:
> 
> > On Tue, Sep 17, 2013 at 8:53 PM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
> > > --- linux-2.6.orig/arch/m68k/include/asm/hardirq.h
> > > +++ linux-2.6/arch/m68k/include/asm/hardirq.h
> > > @@ -5,17 +5,6 @@
> > >  #include <linux/cache.h>
> > >  #include <asm/irq.h>
> > >
> > > -#define HARDIRQ_BITS   8
> > 
> > > --- linux-2.6.orig/include/linux/preempt_mask.h
> > > +++ linux-2.6/include/linux/preempt_mask.h
> > > @@ -11,36 +11,22 @@
> > >   * - bits 0-7 are the preemption count (max preemption depth: 256)
> > >   * - bits 8-15 are the softirq count (max # of softirqs: 256)
> > >   *
> > > - * The hardirq count can in theory reach the same as NR_IRQS.
> > > - * In reality, the number of nested IRQS is limited to the stack
> > > - * size as well. For archs with over 1000 IRQS it is not practical
> > > - * to expect that they will all nest. We give a max of 10 bits for
> > > - * hardirq nesting. An arch may choose to give less than 10 bits.
> > > - * m68k expects it to be 8.
> > 
> > m68k needs some changes in arch/m68k/kernel/entry.S, cfr. this check
> > in arch/m68k/kernel/ints.c:
> > 
> >         /* assembly irq entry code relies on this... */
> >         if (HARDIRQ_MASK != 0x00ff0000) {
> >                 extern void hardirq_mask_is_broken(void);
> >                 hardirq_mask_is_broken();
> >         }
> > 
> > Haven't looked into the details yet...
> 
> Whee. Did not notice that one. Though I can't find anything
> interesting in the low level entry code... Looks like some more
> histerical left overs.

Duh. With brain awake I can see it.

The low level entry code is fiddling with preempt_count by adding
HARDIRQ_OFFSET to it to keep track of nested interrupts. If the count
goes to 0, it invokes do_softirq(). And you have another safety guard:

ret_from_last_interrupt:
        moveq   #(~ALLOWINT>>8)&0xff,%d0
	andb    %sp@(PT_OFF_SR),%d0
	jne     2b

That's due to the irq priority level stuff, which results in nested
interrupts depending on the level of the serviced interrupt, right?
And that's why you fiddle yourself with the HARDIRQ bits in the
preempt count to prevent the core code from calling do_softirq().

Though this scheme also prevents that other parts of irq_exit() are
working correctly, because they depend on the hardirq count being
zero, e.g. the nohz code.

Needs more thoughts how to fix that w/o wasting precious bits for the
HARDIRQ count.

Thanks,

	tglx










--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/