Date: Sat, 30 Aug 2003 21:20:43 +1000 (EST) From: Bruce Evans <bde@zeta.org.au> To: freebsd-current@freebsd.org Cc: Adam K Kirchhoff <adamk@voicenet.com> Subject: Re: IDE DVD playback on 5.1-CURRENT Message-ID: <20030830210326.I3799@gamplex.bde.org> In-Reply-To: <20030829182855.GC14251@dragon.nuxi.com> References: <Pine.LNX.4.33L2.0308270719400.10236-100000@thorn.ashke.com> <3F4DBD62.C8675452@mindspring.com> <20030828165031.GA91019@dragon.nuxi.com> <3F4F8916.8A3EB699@mindspring.com> <20030829182855.GC14251@dragon.nuxi.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 29 Aug 2003, David O'Brien wrote:
> On Fri, Aug 29, 2003 at 10:10:46AM -0700, Terry Lambert wrote:
> > David O'Brien wrote:
> > > On Thu, Aug 28, 2003 at 01:29:22AM -0700, Terry Lambert wrote:
> > > > 1) dd if=/dev/acd0 count=1 of=/dev/null
> > > > 2) dd if=/dev/acd0c count=1 of=/dev/null
> > > > 3) dd if=/dev/acd0a count=1 of=/dev/null
> > > ^^^^
> > > bs=2k
> >
> > Yes, sorry; I forgot that FreeBSD's dd does not stat its endpoints
> > to see if they are devices, and gets the st_blksize and insists
> > that it be used (at least internally) for S_IFBLK and S_IFCHR
> > values of st_mode.
> >
> > Perhaps it should be unbroken by someone with a commit bit and/or
> > the ability to have their send-pr's not rejected as relay SPAM.
>
> I doubt it, this wasn't necessary with SCSI cdrom's until GEOM -- PHK
> thought it a feature to have to require the user to remember the "bs=2k",
> so I don't think anyone will ever be able to get this "fixed" in the
> tree.
This is not only a feaure; it is Standard. From POSIX.1-2001-draft7.txt:
%%%
11599 NAME
11600 dd - convert and copy a file
...
11650 ibs=expr Specify the input block size, in bytes, by expr (default is 512).
11651 obs=expr Specify the output block size, in bytes, by expr (default is 512).
%%%
GEOM has many faults, but this is not one of them.
"dd if=/dev/[a]cd0c [bs=512]" hasn't worked since block devices were axed.
With block devices, you had to use the raw device, e.g., /dev/racd0c, to
attempt to read 512-blocks and fail on normal cdroms because their block
size isn't 512. /dev/acd0c was the block device so it could be read with
any size at some cost in efficiency.. E.g., requests to read it with a
block size of 512 caused the following enblocking and deblocking:
- dd block size of 512 converted to BLKDEV_IOSIZE = 2048
- i/o done with block size of 2048. This happens to be the same as
the physical block size, so it happens to work and doesn't involve
any more enblocking or deblocking
- BLKDEV_IOSIZE converted to dd block size.
BLKDEV_IOSIZE is now PAGE_SIZE and is only used for bogus things (since it
is still used but there are no block devices to use it on).
Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030830210326.I3799>