From owner-freebsd-sparc Wed Nov 21 14:25: 1 2001 Delivered-To: freebsd-sparc@freebsd.org Received: from postfix2-1.free.fr (postfix2-1.free.fr [213.228.0.9]) by hub.freebsd.org (Postfix) with ESMTP id 8BEEB37B405; Wed, 21 Nov 2001 14:24:57 -0800 (PST) Received: from nas-cbv-7-23-62.dial.proxad.net (nas-cbv-7-23-62.dial.proxad.net [213.228.23.62]) by postfix2-1.free.fr (Postfix) with ESMTP id 0A1D7278; Wed, 21 Nov 2001 23:24:55 +0100 (CET) Date: Wed, 21 Nov 2001 20:39:17 +0100 (CET) From: =?ISO-8859-1?Q?G=E9rard_Roudier?= X-X-Sender: To: David O'Brien Cc: Garrett Wollman , , , Subject: Re: main-sym.gz In-Reply-To: <20011120205104.D21537@dragon.nuxi.com> Message-ID: <20011121201701.X2045-100000@gerard> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Sender: owner-freebsd-sparc@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org On Tue, 20 Nov 2001, David O'Brien wrote: > On Tue, Nov 20, 2001 at 11:42:26PM -0500, Garrett Wollman wrote: > > I noticed that tmm put a new Symbios kernel up on his Web area, so I > > tried booting that, and it doesn't seem to work quite right, but does > > at least get to the mountroot prompt: > > Tmm, Jake, and I worked on this for quite a while this afternoon. > > > > I tried to tell it to mount da0a as root, which made it very unhappy: > > > > Mounting root from ufs:/dev/da0a > > (da0:sym0:0:0:0): Retrying Command > > sym0:0: ERROR (a0:0) (0-a7-80) (10/9d) @ (scripta 400:f3100000). > > sym0: script cmd =3D 9f030000 > > sym0: regdump: da 10 80 9d 47 10 00 0f 80 00 80 a7 80 00 07 02 00 00 40= 00 08 ff ff ff. > > sym0: PCI STATUS =3D 0x2000 > > (noperiph:sym0:0:-1:-1): SCSI BUS reset detected. > > (da0:sym0:0:0:0): Retrying Command > > I got the same results. I've emailed the sym author to see if he can > give some insight. I just looked into the problem. Not have that much time this evening, btw. For now, I just see that the PCI device is getting a PCI BUS fault when trying to read from BUS the SCSI TARGET control block from memory on reselection. The physical memory address as seen from the PCI BUS that gets fault is the value of the DSA register (offset 0x10 of the register dump, little endian ordering) -> 0x00400000. The code seems to endianyze some address twice. The below minute patch should be tried. If it doesn't fix, I will try to find more time to-morrow evening for working on this problem. --- sym_hipd.c.ORIG=09Wed Nov 21 20:29:36 2001 +++ sym_hipd.c=09Wed Nov 21 20:30:05 2001 @@ -9371,7 +9371,7 @@ =09np->targtbl =3D (u32 *) sym_calloc_dma(256, "TARGTBL"); =09if (!np->targtbl) =09=09goto attach_failed; -=09np->targtbl_ba =3D cpu_to_scr(vtobus(np->targtbl)); +=09np->targtbl_ba =3D vtobus(np->targtbl); =09/* =09 * Allocate SCRIPTS areas. > > It's possible that there are hardware problems with this machine or > > the SCSI card; > > I don't believe so. You are very probably right. G=E9rard. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-sparc" in the body of the message