Newsgroups: comp.os.linux.announce
From: John Aycock <aycock@cpsc.ucalgary.ca>
Subject: REPOST: Adaptec 274x PRE-ALPHA interim driver release #3
Message-ID: <1994Aug4.123729.26920@cs.cornell.edu>
Date: Thu, 4 Aug 1994 12:37:29 GMT
Approved: linux-announce@tc.cornell.edu (Lars Wirzenius)

This release has some bugs fixed, support for synchronous SCSI as
well as scatter-gather, and should work on 1.1.x kernels as well
as 1.0.9 ones.  From the README:

@(#)README 1.8 94/07/29 jda

AHA274x INTERIM DRIVER

***  THIS SHOULD BE CONSIDERED PRE-ALPHA SOFTWARE.  USE AT YOUR OWN RISK  ***

BACKGROUND & LIMITATIONS

For various reasons, we ended up with one of these cards under the
impression that support was soon forthcoming.  In mid-May, I asked
Scott Ferris (the official person who's supposed to be writing this
driver) what documentation he used, _finally_ got it from Adaptec,
and started writing this driver.  It is now at what I would consider
a usable state.  I've already started running our news server using
this driver - if it is stable (and it appears to be so far), then my
development platform is going into service, meaning I will unfortunately
be able to provide only minimal support.

Note that Scott is still developing his driver, and when his is stable
it will supercede this one (hence the name "interim").  To help eliminate
confusion, this driver will officially remain at a "pre-alpha" level.  The
latest version can be obtained by anonymous ftp from ftp.cpsc.ucalgary.ca
in /pub/systems/linux/aha274x.

It supports EISA only, either single or twin-bus 274x cards (but not
the second SCSI bus of twin cards - see aha274x.c), and supports disconnection,
synchronous SCSI, and scatter-gather.  Unlike previous versions, abort() and
reset() are now implemented, and both hosts.c and aha274x.c should give a
clean compile.  It does not detect parity errors.

I wrote this using a 1.0.9 kernel.  I recently realized that several critical
kernel structures got changed somewhere in 1.1.x; this version has support
for both, and attempts to guess in aha274x.h as to which type of kernel it's
dealing with.  If you get compiler warnings building this, it probably means
that the auto-detect buggered up.  I am still running the driver under 1.0.9,
however - I've only tested under 1.1.32 enough to make sure that it booted.

Since I've gotten a lot of mail about this point, this driver will _not_
work with Adaptec 284x cards, just the 274x's.  The changes should be
trivial to make it work, but I don't have a 284x to test it with, sorry.
If someone wants to send me a patch that's great.  (This paragraph also
applies to the 294x PCI-bus cards.)

Under protest, this driver is subject to the GPL - see the file
COPYING for details.

Thanks to the following people for bug fixes/code improvements (also
thanks to the people who have sent me feedback):

        "David F. Carlson" <dave@ee.rochester.edu>
        Jimen Ching <jiching@wiliki.eng.hawaii.edu>

Special thanks to Drew Eckhardt <drew@kinglear.cs.Colorado.EDU> for
fielding my questions about synchronous negotiation.

INSTALLATION

0.  Make sure that the aha274x's BIOS is enabled or this WILL NOT WORK.
        This is because the EISA configuration information appears to be
        only accessible through a BIOS call - I've located the information
        on my machine via a hack, but I don't think it's portable.  I've
        made several requests for EISA technical information on the net
        without success.. so for the time being it's safer to have the
        aha274x's BIOS load all the information for me.  Also, it performs
        a lot of self tests of the card which I'm not keen on duplicating.

1.  Copy aha274x.{c,h,_seq} to /usr/src/linux/drivers/scsi.  The other
        files can be ignored - they're an assembler for the sequencer's
        assembly language and the sequencer program source.
2.  Make the appropriate changes to /usr/src/linux/Makefile and
        /usr/src/linux/drivers/scsi/hosts.c.
3.  Do the configuration, dependency building, and rebuilding.
4.  BACK UP YOUR KERNEL AND DATA.  I'm not kidding.
5.  Reboot.
6.  Test it and send your comments/patches/&c to "aycock@cpsc.ucalgary.ca";
        please consider sending me mail even if you have no problems, so
        I know if someone else is using this successfully.

Sorry the instructions aren't more complete, but if you can't fill
in the extra bits then you probably shouldn't be using this yet.
:ja

--
Send submissions for comp.os.linux.announce to: linux-announce@tc.cornell.edu
PLEASE remember Keywords: and a short description of the software.
