Re: [PATCH]: better usage of spinlocks in networking drivers

Jes Sorensen (Jes.Sorensen@cern.ch)
03 Jul 1999 20:21:40 +0200


>>>>> "Augusto" == Augusto Cesar Radtke <bishop@sekure.org> writes:

Augusto> On 3 Jul 1999, Andi Kleen wrote:
>> As far as I can see your patch adds local cli()s to the interrupt
>> handlers. This doesn't make too much sense, because they're
>> normally not reentrant (there will be no other interrupt for this
>> device as long as this one isn't acked). It only prevents
>> interrupts for other devices, which is bad for interrupt
>> latency. So your change is unnecessary.

Augusto> Sorry if it's my stupidy, but SMP machines can't cause
Augusto> interrupt reentrants?

A particular interrupt level/number currently being serviced will not
be regenerated before the first handler is finished. So you will not
get two interrupt #21's in parallel etc.

Jes

-
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/