From owner-freebsd-hardware Sat Aug 3 13:51:43 1996 Return-Path: owner-hardware Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id NAA14122 for hardware-outgoing; Sat, 3 Aug 1996 13:51:43 -0700 (PDT) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id NAA14110 for ; Sat, 3 Aug 1996 13:51:38 -0700 (PDT) Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.12/8.6.9) id GAA32529; Sun, 4 Aug 1996 06:47:47 +1000 Date: Sun, 4 Aug 1996 06:47:47 +1000 From: Bruce Evans Message-Id: <199608032047.GAA32529@godzilla.zeta.org.au> To: hardware@freebsd.org, randy@zyzzyva.com Subject: Re: Mapped geometry vs. Actual Sender: owner-hardware@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >I've long been an advocate of working out the true geometry >issues when setting up SCSI drives. >Could someone comment on whether this is still (ever) considered >to be worth the frustration? No, it's not unless the drive actually has a truly simple geometry (i.e., a fixed number of sectors per track) and very little buffering. >I had a recently frustrating experience with a 540M Quantum Fireball >that FreeBSD simply refused to accept my "true" drive geometry. >FreeBSD won and mapped to ???/32/64. The Fireball has a geometry of 2 heads and 88-177 sectors per track, and a 128K buffer. It doesn' have a truly simple geometry so you shouldn't try to tell FreeBSD its "true" geometry. ???/32sectors/64heads should be written as ???/64/32. 64 heads is impossible for bootable drives. Newfs uses the geometry ???/1/2048 unless it is invoked with special options. This strange geometry is used to limit newfs's "optimizations" to 1MB boundaries. The optimizations are usually pessimizations. The ???/64/32 geometry is probably only used by the BIOS and the bootstrap. Bruce