Greg KH wrote:
> I really don't know if they got that patch or not, sorry.

What version of util-linux has that patch, do you know?

Well http://www.kernel.org/pub/linux/utils/util-linux/ only goes up to
q, so I'm guess that's it. I'll file a bug report, thanks.

> Because you should use O_DIRECT to access block devices directly, which
> properly uses the kernel io subsystems.  raw goes around all of that,
> and thinks it is smarter than the kernel.  In the end, it is wrong, and
> can be slower.

Fair enough. Wouldn't it make sense to make a simple raw module that
simply was equivalent to opening the block device with O_DIRECT if for
no other reason than, erm, "backwards" compatiblity with other unices -
for ease of the writers of mplayer/xine/ogle/etc.?

> Anyway, you can get around this by creating the device node by hand if
> you really want.  I suggest you file a debian bug to get them to fix the
> raw program to work properly with udev.

I would be interested to experiment, even though I'm sure you're
right... but I think I'm missing something... don't the major/minor
number need to be the same:

[phil@rider rc2.d]$ sudo mknod /dev/raw/raw1 c 22 0
[phil@rider rc2.d]$ sudo raw /dev/raw/raw1 /dev/hdc
Device '/dev/raw/raw1' is not a raw dev
[phil@rider rc2.d]$ ls -l /dev/raw/raw1 /dev/hdc
brw-rw----  1 root cdrom 22, 0 Apr 25 21:36 /dev/hdc
crw-r--r--  1 root root  22, 0 Apr 25 23:08 /dev/raw/raw1

--
Phil Dibowitz                             phil@ipom.com
Freeware and Technical Pages              Insanity Palace of Metallica
http://www.phildev.net/                   http://www.ipom.com/

"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety."
 - Benjamin Franklin, 1759