[485] in linux-scsi channel archive
Re: Generic Scsi Driver Bug Report and comments.
daemon@ATHENA.MIT.EDU (Eric Youngdale)
Wed Aug 9 13:44:51 1995
From: "Eric Youngdale" <eric@aib.com>
Date: Mon, 7 Aug 1995 10:51:00 -0400
In-Reply-To: woody@mail.csh.rit.edu (Craig "Woody"ward)
"Re: Generic Scsi Driver Bug Report and comments." (Aug 4, 1:18pm)
To: woody@mail.csh.rit.edu (Craig "Woody"ward), linux-scsi@vger.rutgers.edu
> I agree with the point for making this for all SCSI systems, but
>not with your solution. It's painfully obvious that we have to extend
dev_t
>to at least 32-bits, but thats about all anyone can agree on.
This much is fine, but the problem is that I can make a case that 32
is insufficient. I feel that it is probably a mistake from a performance
point of view to start using 64 bit quantities in the i386 kernel. Actually
Drew made the point that even 64 bits might be insufficient....
Ultimately my main objection to a static mapping is that it would
involve lots of assumptions about how many bits we need for various things.
For example, if we had a static mapping before wide scsi came out, we would
have assumed that there are only 8 different scsi IDs available, and 8
different scsi LUNs. With wide scsi both of these numbers are increased,
and
any static mapping would have been completely broken.
> I personally
>don't like the 'auto dev creation' method you suggest because I don't like
>the idea of something that low-level playing with a real FS. I also loath
>setup scripts and /dev subdirectories.
What do you mean by "that low-level"? The program uses normal
system
calls to create the devices - we are not bypassing the filesystem at all.
Isn't /etc/rc a setup script?
The entries do not have to live in /dev/scsi/ - this was just
something I created to keep it simpler for me to test and debug.
-Eric
--
"The woods are lovely, dark and deep. But I have promises to keep,
And lines to code before I sleep, And lines to code before I sleep."
