From owner-freebsd-bugs Sun May 3 12:35:32 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id MAA11912 for freebsd-bugs-outgoing; Sun, 3 May 1998 12:35:32 -0700 (PDT) (envelope-from owner-freebsd-bugs@FreeBSD.ORG) Received: from ns1.yes.no (ns1.yes.no [195.119.24.10]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id MAA11898 for ; Sun, 3 May 1998 12:35:25 -0700 (PDT) (envelope-from eivind@bitbox.follo.net) Received: from bitbox.follo.net (bitbox.follo.net [194.198.43.36]) by ns1.yes.no (8.8.7/8.8.7) with ESMTP id TAA01989; Sun, 3 May 1998 19:34:58 GMT Received: (from eivind@localhost) by bitbox.follo.net (8.8.8/8.8.6) id VAA12059; Sun, 3 May 1998 21:34:36 +0200 (MET DST) Message-ID: <19980503213435.36715@follo.net> Date: Sun, 3 May 1998 21:34:35 +0200 From: Eivind Eklund To: Nate Williams Cc: John Hay , freebsd-bugs@FreeBSD.ORG Subject: Re: i386/5398 References: <199805031749.TAA07478@zibbi.mikom.csir.co.za> <11869.894218238@critter.freebsd.dk> <199805031858.MAA20839@mt.sri.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.89.1i In-Reply-To: <199805031858.MAA20839@mt.sri.com>; from Nate Williams on Sun, May 03, 1998 at 12:58:27PM -0600 Sender: owner-freebsd-bugs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Sun, May 03, 1998 at 12:58:27PM -0600, Nate Williams wrote: >>> Just to get my understanding of the hardware a little better. Is this >>> blitting just doing something like a software memory copy, like what >>> the C function bcopy() does? And can that really block interrupts just >>> because it is done over the PCI bus? Or does blitting use something >>> different? >> >> I don't know how it is done exactly, check with XFree86. I just know >> that any PCI bus hogging will send your interrupt latency soaring :-( > > ps. It's not just PCI bus hogging, since I've got an ISA box that also > sees the same thing. As a matter of fact, I'm pretty sure it's > completely unrelated to any particular bus architecture. It isn't. I don't know if any PC-based hardware has a solution for the problem, but a bus design can avoid total busmastering and instead allocate fractions. It would give a slight slowdown when the capability was used (you'd have to abort a bus-run to let other devices have bus access), or it would require parallell bus tech, but it isn't undoable. Basically, it is the hardware parallel of multitasking. Eivind. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the message