From owner-freebsd-alpha Thu Jul 20 14:33: 3 2000 Delivered-To: freebsd-alpha@freebsd.org Received: from front6.grolier.fr (front6.grolier.fr [194.158.96.56]) by hub.freebsd.org (Postfix) with ESMTP id 0C73737B65D; Thu, 20 Jul 2000 14:32:43 -0700 (PDT) (envelope-from groudier@club-internet.fr) Received: from nas1-252.cgy.club-internet.fr (nas1-252.cgy.club-internet.fr [195.36.197.252]) by front6.grolier.fr (8.9.3/No_Relay+No_Spam_MGC990224) with ESMTP id XAA02551; Thu, 20 Jul 2000 23:32:22 +0200 (MET DST) Date: Thu, 20 Jul 2000 23:11:59 +0200 (CEST) From: =?ISO-8859-1?Q?G=E9rard_Roudier?= X-Sender: groudier@linux.local To: "David O'Brien" Cc: Andrew Gallatin , freebsd-alpha@FreeBSD.ORG Subject: Re: alpha snapshots? (was RE: Alphserver 2100 4/233) In-Reply-To: <20000720130323.B67647@dragon.nuxi.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Thu, 20 Jul 2000, David O'Brien wrote: > On Wed, Jul 19, 2000 at 11:37:40AM -0400, Andrew Gallatin wrote: > > Hmm.. I'd suggest axing the ncr driver (why is it still in GENERIC now= =20 > > that SYM supports the older devices??).=20 >=20 > I feel groudier would be a better judge of this than I am. > If he agrees, I'll axe it out. I can explain, but I am not the right judge for that at all. IMO, the following are to be considered for Alpha for latest SYM as in current and RELENG_4: - The SYM requires 32 bit DWORD (LWORD for Dec people) write (modify but not update) accesses to its internal data structures in main memory from the BUS (bridge) and from the CPU to happen atomically. (I think this is the same for NCR). I just hope gcc will not play with 64 bit access=20 when the driver states a volatile access to a 32 bit entity. - The SYM implements all barriers that appear to me needed (only one is =20 only useful:) ) from the CPU (mb) and from the PCI BUS (dummy reads,=20 but sometimes useful reads do the needed flushing of posted writes). - I also have checked that memory barriers used by NewBus are enough for=20 the expectations of the driver about IO/MMIO ordering. If I didn't miss something important, and modulo hardware erratas I donnot know about, I donnot expect the driver to cause more failures on Alpha than on IA32. But you must keep in mind that I donnot have Alpha Hardware and am relying on testers for the checking of my statement about SYM reliability on Alpha (compared to IA 32). Regards, G=E9rard. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message