On Tue, 01 Sep 1998 08:21:15 +0200  Marc Gutschner wrote:

> I have some trouble with my last sup of "current"... A "make build" 
> falls over flat on it's face when doing a "make cleandir" in the 
> directory "gnu/usr.bin/gdb", where it complains about a missing
> "_mips.mk", there is only a "_mipsel.mk"  :-( Now, are the DECstations
> "MIPS with the 'wrong' byte-order" ?, i.e. can I just take that one
> and create a link ? I've just checked the CVS update logs of the last 
> 10 days or so and can't find where it may have disappeared. Any pointer 
> is appreciated :-)

The whole gnu/usr.bin thing is in a state of flux at the moment with
work going on with ecgs and binutils.  However, the following lines from
the Makefile were recently removed:

	# Override this to build a cross-debugger.
	TARGET_ARCH?=$(MACHINE_ARCH:S/^mips$/mipsel/)

I built -current a week or two ago with the above lines in and it worked
ok.  Note however that at that time, strip, objdump, ranlib, objcopy,
size, ar and nm (the binutils things) didn't work.  I can't recall if
this has been fixed since then or not.

> Another thing: With the 1.3.2 kernel the machine with a TZ30 will just
> crash and re-boot when I do something like a "tar tv". When using a
> kernel built from last fridays 'sup' the tape works

Below is a message from Michael Hitch with a fix to your TZ30 problem.

Simon.
--
> From: "Michael L. Hitch" <mhitch@lightning.oscs.montana.edu>
> To: "emanuel stiebler" <emu@ecubics.com>, <port-pmax@NetBSD.ORG>
> Subject: Re: TZ30 on NetBSD 1.3.2
> Date: Fri, 17 Jul 1998 13:56:46 -0600
> Message-Id: <199807171956.NAA25973@lightning.oscs.montana.edu>
> In-Reply-To: "emanuel stiebler"'s message of Jul 17,  2:04pm

On Jul 17,  2:04pm, "emanuel stiebler" wrote:
> again problems with my tz30 tape. Upgraded all to 1.3.2 & trying to do an
> backup with:
> 
> tar cvf /dev/rmt1 *
> 
> i get :
> 
> buffer overrrun, rebooting !!
> 
> any ideas ?

  Is this on a 3100 or 5000?

  I think the 3100 SII SCSI driver can't deal with data requests where the
size doesn't exactly match the real data transferred, and the TZ30 mode
sense length is different from what the tz driver tries to use for a
generic tape.

  Here's a patch to let tz.c recognize the TZ30 and adjusts the mode sense
length accordingly:

--- /altroot/b/NetBSD-1.3.2/src/sys/arch/pmax/dev/tz.c	Fri May  8 16:47:11 1998
+++ ../../dev/tz.c	Sun Jul 12 14:01:34 1998
@@ -187,9 +187,12 @@
 	printf("tz%d at %s%d drive %d slave %d", sd->sd_unit,
 		sd->sd_cdriver->d_name, sd->sd_ctlr, sd->sd_drive,
 		sd->sd_slave);
-	if (i == 5 && inqbuf.version == 1 && inqbuf.qualifier == 0x50) {
+	if (i == 5 && inqbuf.version == 1 && (inqbuf.qualifier == 0x50 ||
+	    inqbuf.qualifier == 0x30)) {
 		printf(" TK50\n");
 		sc->sc_tapeid = MT_ISTK50;
+		if (inqbuf.qualifier == 0x30)
+			sc->sc_modelen = 14;
 	} else if (i >= 5 && inqbuf.version == 1 && inqbuf.qualifier == 0 &&
 	    inqbuf.length == 0) {
 		/* assume Emultex MT02 controller */

-- 
Michael L. Hitch			mhitch@montana.edu
Computer Consultant
Information Technology Center
Montana State University	Bozeman, MT	USA