Date: Sun, 15 Jul 2001 21:42:14 +1000 (EST) From: Bruce Evans <bde@zeta.org.au> To: Peter Wemm <peter@wemm.org> Cc: mjacob@feral.com, Jim Bryant <kc5vdj@yahoo.com>, scsi@FreeBSD.ORG Subject: Re: oddity with dump(8) or sa(4) in -current Message-ID: <Pine.BSF.4.21.0107152130000.50566-100000@besplex.bde.org> In-Reply-To: <20010714230005.14DC6380B@overcee.netplex.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 14 Jul 2001, Peter Wemm wrote: > Matthew Jacob wrote: > > > > On Fri, 13 Jul 2001, Bruce Evans wrote: > > > > > On Thu, 12 Jul 2001, Matthew Jacob wrote: > > > > > > > On Thu, 12 Jul 2001, Jim Bryant wrote: > > > > > On a side note, what happened to the "raw" device nodes [/dev/*rsa*]? I > 'm guessing that the lack of "raw" nodes doesn't make a diff, > > > > > as /dev/*sa* are all character-mode devices. > > > > > > > > No, all devices are block devices now. Linux has no raw, FreeBSD has no > > > > cooked. Feh. > > > > > > No. No devices are block devices now (or yesterday; block devices > > > went away in 4.0). Linux has no character disk devices; FreeBSD has > > > no block disk devices. Feh. Tapes are a little different from disks. > > > > Really? I'll have to tell the ANSI SCSI committee! I understated a little. Tapes are more than a little different from disks. > He means *buffered* devices (aka block, using bdevsw, and the buffer cache) > went away. Raw devices (aka character, using cdevsw, but going direct to I really did mean block devices... > the device) are what we kept. Raw devices are accessible depending on the > backing device.. eg: must be block sized transfers/seeks/etc for things > like tapes/disks/etc, but can be any for things like frame buffers. > Buffered (block, cached) can be written on any size/alignment and can be > seeked for disks. We have not had the gross B_TAPE flag for some time that > was used to stop the buffer cache getting too creative with the tape > strategy routines (ie: doing random IO). > > Unix stat(2) calls raw devices "character" and buffered devices "block", > even though raw devices can be both block or byte addressable and > buffered devices enabled character/byte IO. :-) Hence the tangled up > terminology. There are only character-special and block-special devices for stat(2) and in POSIX. I intentionally didn't "raw" or "cooked" because those terms are fuzzy. "block" is not fuzzy; it is just sort of backwards as it applied to disks (the block devices were byte addressable but the non-block devices are only block addressable). Bruce To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0107152130000.50566-100000>