From owner-freebsd-current Tue Mar 21 07:28:21 1995 Return-Path: current-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id HAA22652 for current-outgoing; Tue, 21 Mar 1995 07:28:21 -0800 Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.34]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id HAA22639 for ; Tue, 21 Mar 1995 07:28:10 -0800 Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id BAA19070; Wed, 22 Mar 1995 01:25:09 +1000 Date: Wed, 22 Mar 1995 01:25:09 +1000 From: Bruce Evans Message-Id: <199503211525.BAA19070@godzilla.zeta.org.au> To: bde@zeta.org.au, current@FreeBSD.org, se@MI.Uni-Koeln.DE Subject: Re: Sharing interrupts with PCI devices? Sender: current-owner@FreeBSD.org Precedence: bulk >Hmm, you are talking about the overhead >of returning a '0' or '1' ? >That's just one or two cycles ... One or two here, one or two there ... I'm looking at a cyclades intelligent serial board and driver that is 25% slower than an unintelligent 16450 board and 3 times slower than an unintelligent 16550 board. Perhaps it would be faster if cycles weren't wasted here and there. >ONLY devices that will be used with shared ints >have to return some value. In fact, this could And those handlers that are declared to return int :-). I'm more concerned about software complications and kludges involved in having only some handlers return int. >There will be a new function to register a >(possibly) shared interrupt. I think there should be separate interrupt registration and dispatch, etc. for different classes of interrupts. I can save a few usec in the ISA handlers if they spec'ed to not support level triggered interrupts. Bruce