Re: Today Linus redesigns the networking driver interface (was Re: tulip driver in ...)

Richard Gooch (rgooch@atnf.csiro.au)
Wed, 23 Sep 1998 10:15:15 +1000

Linus Torvalds writes:
>
>
> On Wed, 23 Sep 1998, Richard Gooch wrote:
> >
> > I don't understand why selectively marking/not marking a BH doesn't
> > give you what you want.
>
> The problem with not marking the BH is that there would be no timeout.
>
> In contrast, marking the BH but asking the interrupt return code not
> to do BH processing will still guarantee that the BH gets processed
> eventually anyway: at the latest during the next timer tick, but
> possibly long before that if we're idle.

Ah, OK, I see. I was working under the assumption that an interrupt
handler would hold off marking its BH until it got a "marker"
interrupt. The example Ted cited with PPP fit this: mark the BH when
you get the last character in the frame.
I guess not all handlers have the luxury of an end-of-frame marker,
though.

BTW: if you're idling, that doesn't mean that BH's are called before
the next timer tick. AFAIK do_bottom_half() is only called on return
from syscall/interrupt and when rescheduling.
Waiting for the next timer tick would be horrible for latency,
though.

Regards,

Richard....

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