[5975] in linux-scsi channel archive
Re: WG: AW: cdrecord problems on recent Linux versions
daemon@ATHENA.MIT.EDU (Richard Gooch)
Tue Feb 23 00:19:42 1999
Date: Tue, 23 Feb 1999 15:10:26 +1100
From: Richard Gooch <rgooch@atnf.csiro.au>
To: Monty <xiphmont@MIT.EDU>
Cc: linux-scsi@vger.rutgers.edu
In-Reply-To: <199902220857.DAA02547@CUTTER-JOHN.MIT.EDU>
[Cc list snipped again]
xiphmont@mit.edu writes:
> >No problem. The file does not show up in the driectory. You can still
> >try to open it, if you know it's name. The kernel can autoload the
> >appropriate module. This differs from normal autoloading a little
> >bit. There you have the device files and the kernel perform a loading
> >request for block-major-XY ...
>
> And if you don't know the name? Then you have to keep trying every
> possible combination until you find it. One of my machines has
> nearly 30 SCSI CDROMs on it... for each CDROM to be opened, I have
> to try probing each target and LUN on all six chains? At what point
> do I give up and assume the device doesn't exist?
I hope I explained this in my previous message. To recap: you only
speculatively open the *directories* for a particular device type:
/dev/sr all SCSI CD-ROMs
/dev/ide/cd all IDE CD-ROMs
You see, there isn't a module for each SCSI CD-ROM. There's just one,
so when it gets loaded, *all* SCSI CD-ROMs on your system are
configured and each gets an entry in /dev/sr.
> A better way has to be found. The above guessing game is not
> acceptible.
You only need to "guess" a couple of directory names. And they are
"well known".
> >devfs has been out for a very loing time now and a lot of people use it
> >without problems. As it's fully backward compatible, using it does not hurt
> >anybody.
>
> So you say and so you think. I honestly don't mean to sound
> antagonstic (or condescending), but I simply can't believe what you
> say. Are you honestly claiming that devfs is 100%, totally bug-free?
> I thought not-- then it doesn't belong being introduced in the middle
> of a stable release. Stable means 'bugfix only'. Anything you might
> do that will knowingly introduce more bugs is not for a stable
> release.
While there may still be bugs in a devfs-enabled kernel
(i.e. CONFIG_DEVFS_FS=y), I don't think there should be any with
CONFIG_DEVFS_FS=n. That's what I mean by "doesn't break anything".
Also, any such bugs would be fixed, and would likely be found during
the pre-patches anyway. It's not as if every patch to a stable kernel
doesn't break anything.
Anyway, I doubt we'll agree on this point, so let's just shake hands
and agree to disagree, eh? :-)
Regards,
Richard....
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
