Subject: digest of port-sparc messages since 96/5/30 (part 2)
To: None <port-sparc@NetBSD.ORG>
From: Paul Kranenburg <pk@NetBSD.ORG>
List: port-sparc
Date: 06/12/1996 17:17:20
And here's part 2:
----snip----
From: greywolf@defender.VAS.viewlogic.com
id AA04984; Fri, 31 May 1996 17:18:33 +0500
id AA21296; Fri, 31 May 96 14:47:53 PDT
id AA01192; Fri, 31 May 96 14:47:52 PDT
Date: Fri, 31 May 96 14:47:52 PDT
Message-Id: <9605312147.AA01192@>
To: kgray@netcom.com
Cc: port-sparc@NetBSD.ORG
In-Reply-To: <Pine.SUN.3.91.960531075400.27239B-100000@netcom13> (message from
Kerry Gray on Fri, 31 May 1996 07:56:02 -0700 (PDT))
Subject: Re: getting sparc to boot off sd
Reply-To: greywolf@starwolf.com
Kerry Gray <kgray@netcom.com> sez:
The install doc says I must have a copy of a boot program from SunOS to
boot from the internal disk. I don't have SunOS. Is there a way to make
it boot from disk, or am I S.O.L., doomed to net-boot for all eternity?
No, if you're booting diskless, then you already have access to /usr/mdec.
Just boot diskless, mount the disk partition as, say, /mnt. Then
# cd /usr/mdec
# ./binstall ffs /mnt
NNN+P records in
NNN+P records out
NNNNN bytes transferred in T.TT seconds (N bytes/sec)
# reboot
and you should be all set.
Oh, yeah: Don't forget the rest of the root filesystem, and don't forget
to
# cd /mnt/dev; sh MAKEDEV std sd0 sd1 sd2 sd3 st0 cd0 pty0 pty1 ...
before you reboot. :-)
--*greywolf;
--
"I presume I need no introduction." -- Lestat
Date: Sat, 1 Jun 1996 00:40:58 +0200 (CED)
From: Krister Walfridsson <cato@df.lth.se>
To: Paul Boven <e.p.boven@student.utwente.nl>
cc: port-sparc@NetBSD.ORG
Subject: Re: Alas...
In-Reply-To: <199605310213.EAA04522@wit387304.student.utwente.nl>
Message-ID: <Pine.VUL.3.91.960601003639.19770A-100000@marvin.df.lth.se>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Hej!
On Fri, 31 May 1996, Paul Boven wrote:
> > > > sd0 at scsibus0 targ 3 lun 0: <QUANTUM, TRB850S, 0404>
> SCSI2 0/direct fixed
> > > > sd0: dma0: xfer (-65528) > req (8)
> > > > esp0: !TC [intr 10, stat 83, step 4] prevphase 1, resid 8
> > > > 810MB 3653 cyl, 4 head 113, sec, 512 bytes/sec
> >
> > Obviously, something has made it to the host, as the inquiry and mode sense
> > commands have at least produced some reasonable output.
>
> Well, on my ELC, even that doesn't happen, the drive is not recognised at
> all. It could be that Kirsten gets 44 bytes transfered, and I only 4, there
> is likely a bit more information in what he gets :)
I just remembered something which might (?) give us some more information...
I tried NetBSD 1.1 around 6 month ago and it didn't work, BUT it
wasn't the same error! I just tried it again, and NetBSD 1.1 works fine
as long as it doesn't try to write anything to the hard disk. When it does
it writes some permutations of
empty FIFO of 16
empty FIFO of 1
dma0: error
on the screen. Small files (a couple of bytes) seems to be OK (well..
I get the errors, but the information in the file seems to be right) but
I got core dumps when I tried to write longer files...
So at least one bug must have been introduced since 1.1?
/Krister
Date: Fri, 31 May 1996 18:28:01 -0400
From: der Mouse <mouse@Collatz.McRCIM.McGill.EDU>
Message-Id: <199605312228.SAA23057@Collatz.McRCIM.McGill.EDU>
To: kgray@netcom.com
Subject: Re: getting sparc to boot off sd
Cc: port-sparc@NetBSD.ORG
> The install doc says I must have a copy of a boot program from SunOS
> to boot from the internal disk.
The install doc is out of date, then. NetBSD/sparc now has fully
native boot code; if you can netboot, you can install the disk boot
programs (assuming, of course, that there isn't something bizarre with
your particular disk that makes netboot break...which is rather
unlikely at this point).
der Mouse
mouse@collatz.mcrcim.mcgill.edu
Message-Id: <1.5.4.32.19960531222515.006720c8@interpath.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Fri, 31 May 1996 18:25:15 -0400
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: "Kevin P. Neal" <kpneal@interpath.com>
Subject: Re: sparc NetBSD
Cc: "John Franklin Acree" <jfacree@eos.ncsu.edu>, port-sparc@NetBSD.ORG
At 08:38 AM 5/31/96 -0700, Jason Thorpe wrote:
>On Fri, 31 May 1996 07:35:05 -0400
> "John Franklin Acree" <jfacree@eos.ncsu.edu> wrote:
>
> > I'm trying to install NetBSD for a Sparc, onto a 424mb drive, where can
I get
> > the miniroot, already compiled, for this?
>
>You can get the NetBSD/sparc 1.1 distribution from:
>
> ftp://ftp.netbsd.org/pub/NetBSD/NetBSD-1.1/sparc/
>
>Ciao.
>
Hi guys, I'm trying to help John get his machine running, so that's why
I'm responding.
The machine is an IPX, it has the 424mb drive on it. The drive is partitioned
well enough to do an install (it did have SunOS on it, but accidents happen,
and now there is no /usr partition). SunOS sucks rocks without a /usr.
So his machine has a drive, no OS, no CD-ROM (we could scrounge one of mine,
but I'm not excited about taking apart my PC), no tape drive (we could
scrounge...anyway). We could network boot it, but then how do we install?
There doesn't seem to be a miniroot. Which means this machine is gonna sit
here and take up space in my graveyardcoughcoughcoughbedroom until doomsday
unless we can mount a real partition onto an NFS mount point. The alternative
is for a miniroot to appear, or SunOS media to appear. SunOS 4.1 doesn't
work, we already tried it.
Solutions?
--
XCOMM Kevin P. Neal, Sophomore, Comp. Sci. \ kpneal@interpath.com
XCOMM The Mirrorball Wizard of Smerp \ kpneal@eos.ncsu.edu
XCOMM Visit the House of RetroComputing at / Perm. Email:
XCOMM http://www4.ncsu.edu/~kpneal/www/ / kevinneal@bix.com
Message-Id: <199606020406.XAA00162@bullbox.jbl.com>
To: port-sparc@NetBSD.ORG
Subject: ELC memory
Date: Sat, 01 Jun 1996 23:05:07 -0500
From: Jon Buller <jonb@metronet.com>
I just aquired a Sparc ELC with no memory and no disk. I want to
run NetBSD on it, and use it as a console for my PC532. Can anyone
tell me, what kind of memory I need, and how much would make a
usable system for compiling kernels and as an X-Term? Do I just
need a 72 pin SIMM from a Mac or PC, or did Sun make some SIMM-like
memory so I can pay Sun prices for it? Also, what will I need to
do to get the system booting off of a SCSI drive? I don't have
any Ethernet for the ELC, but I could run PPP into the PC532. I
suspect that the ELCs ROM can't boot via PPP out the serial ports,
but I've never really dealt with Sun ROM configurations and
capabilities. Could I clone a disk off my little-endian PC532
somehow, or should I go borrow a tape drive?
Thanks for any help in advance,
Jon Buller
Message-Id: <199606021549.KAA01769@bullbox.jbl.com>
To: port-sparc@netbsd.org
cc: jonb@metronet.com
Subject: Sparc ELC questions
Date: Sun, 02 Jun 1996 10:49:28 -0500
From: Jon Buller <jonb@metronet.com>
[ I sent a message very similar to this one about 12 hours ago to
port-sparc@netbsd.org but haven't seen it yet. I am assuming it
got hosed somehow in the partition full problem that messed up
several of the other lists. If this appears twice, please accept
my apologies. ]
I just got a SPARC ELC with no memory or disk. I have some spare
SCSI disks to use, and I've given the install document a quick
once-over, but I have a few questions since I have never dealt much
with the SPARC boot ROMs.
1) Does the ELC take standard 72 pin memory SIMMs from Macs or PCs,
or will I need to call Sun Express and get special Sun memory at
a special Sun price? If they are standard, do they need parity?
2) I do not have any other Ethernet capable machines, can I do a diskless
boot via PPP over a serial port? If not, what is my best option for
getting an initial image on the disk? QIC150 tape? CD-ROM? A cloned
(and hacked) disk from my little-endian NetBSD/PC532?
Thanks for any help in advance,
Jon Buller
Message-Id: <199606031632.SAA04895@wit387304.student.utwente.nl>
Subject: My second kernel ;)
To: port-sparc@NetBSD.ORG
Date: Mon, 3 Jun 1996 18:32:10 +0200 (MET DST)
From: e.p.boven@student.utwente.nl (Paul Boven)
Reply-to: e.p.boven@student.utwente.nl
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Hello everyone,
A short followup on my previous posting "my first kernel":
I have installed all the -current binaries on my ELC, and now the
supped-current builds without a hitch, I was indeed using too old
a compiler.
Alas, this kernel ialso has the esp/dma problems still, as was
expected. Output is basicly the same as what I posted to the list
a few days ago, and therefore (and because the list seems rather
dead all of a sudden) I will not post it's output, unless someone
wants it really bad.
Happy hacking, Paul.
--
----------------------------------------------------------------------
Paul Boven, <e.p.boven@student.utwente.nl> PE1NUT QRV 145.575 JO32KF
"And she looked like she had sex, with a tyranosaurus rex"
Cannibal surf babe -- Afraid of sunlight -- Marillion
----------------------------------------------------------------------
id <m0uR7RM-000o5qC@pci.on.ca>; Tue, 4 Jun 96 21:31 EDT
Date: Tue, 4 Jun 1996 21:25:42 -0400 (EDT)
Message-Id: <199606050125.VAA00252@repeat.pci.on.ca>
From: David Gilbert <dgilbert@jaywon.pci.on.ca>
To: port-sparc@netbsd.org
Subject: si1 problems.
I have tried compiling a kernel with si1 (a second VME SCSI
board in my 4/260), and the whole thing seems flakey --- all manner of
filesystem errors... from dup inodes to bad directories... At the
point when I first found the flakeyness, I did not have the card
installed (and it didn't configure).
When I installed the card, it did indeed work (finding drives,
etc.) and the installed card has no effect on the workings of the
machine when it's not configured.
Anyways... it leaves me wondering (after loosing my news and
mail partitions to this adventure) if the si driver is happy with more
than one card?
Dave.
--
----------------------------------------------------------------------------
|David Gilbert, PCI, Richmond Hill, Ontario. | Two things can only be |
|Mail: dgilbert@jaywon.pci.on.ca | equal if and only if they |
|http://www.pci.on.ca/~dgilbert | are precisely opposite. |
---------------------------------------------------------GLO----------------
id MAA25965; Wed, 5 Jun 1996 12:07:27 +0900
id MAA11022; Wed, 5 Jun 1996 12:06:11 +0900 (JST)
id MAA17134; Wed, 5 Jun 1996 12:06:10 +0900
id MAA09400; Wed, 5 Jun 1996 12:06:08 +0900
id AA21159; Wed, 5 Jun 96 12:02:22 JST
Message-Id: <199606050305.MAA02108@ecology.bell.flab.fujitsu.co.jp>
To: port-sparc@NetBSD.ORG
Subject: PFU,S-4_20H supported?
Mime-Version: 1.0 (generated by tm-edit 7.66)
Content-Type: text/plain; charset=US-ASCII
Date: Wed, 05 Jun 1996 12:05:20 +0900
From: =?ISO-2022-JP?B?GyRCTCcxOhsoQiAbJEI/PxsoQg==?= / MINOURA Makoto <minoura@kawasaki.flab.fujitsu.co.jp>
CHANGES says NetBSD-current has supported sun4m machines.
I have in my office a PFU S-4_20H, which is a SPARCStation 20
compatible box with ROSS RT625 HyperSPARC processor, and want to
run NetBSD on it. Does anyone have tried on this model?
My concern is that according to the Solaris 2.5's /platform
directory the kernel for S-4_20H is different from the ordinary
sun4m (/platform/PFU,S-4_20H is a symbolic link to sun4mL directory),
and that in the current cpu.c I cannot find the string `RT625'.
Thanks!
--
MINOURA Makoto
minoura@flab.fujitsu.co.jp
Communication Network Systems Laboratory
Fujitsu Laboratories Ltd., Kawasaki, Japan
id AA12118; Wed, 5 Jun 96 01:10:23 PDT
id AA07803; Wed, 5 Jun 96 00:51:44 PDT
Message-Id: <9606050751.AA07803@isolar.Tujunga.CA.US>
To: current-users@NetBSD.ORG
Reply-To: current-users@NetBSD.ORG
Cc: port-sparc@NetBSD.ORG
Subject: Success building everything (sans kernel) into $DESTDIR!
Date: Wed, 05 Jun 1996 00:51:29 -0700
From: Greg Earle <earle@isolar.Tujunga.CA.US>
I've posted several messages recently concerning problems doing "make build"
with DESTDIR set. I'm happy to report that I was finally able to get it to
work completely earlier today.
Several people had asked me privately how I was doing it, so I thought I'd
mention what it took (hit "d" or "n" here if not interested ... ):
[This is on a SPARCstation 2 running pk's April 12th 1.1B snapshot (kernel and
userland).]
-------------------- >8 snip ... valuable coupon ... snip 8< -----------------
First, define $DESTDIR. (I used "setenv DESTDIR /usr/local/.build" as that
put it on a disk which wasn't my /usr/src disk, and it wasn't my disk
where I had re-pointed "/usr/obj".)
I then created ${DESTDIR}/usr/share/mk and went to /usr/src/share/mk and did a
"make install" with $DESTDIR set. This pre-populates ${DESTDIR}/usr/share/mk,
which is a necessary preliminary step.
I had to change 4 Makefiles a priori due to hiccups with previous attempts:
/usr/src/domestic/lib/libkadm/Makefile
/usr/src/domestic/lib/libkrb/Makefile
/usr/src/domestic/lib/libss/Makefile
/usr/src/usr.bin/vi/build/Makefile
In the 3 domestic libs, I had to change the "beforeinstall:" target to instead
be "includes:". See my previous post to current-users and replies as to why.
The usr.bin/vi/build/Makefile contained an "afterinstall:" target to install
the catalogs into "$(datadir)/vi/catalog". Unfortunately, "datadir" was
hard-coded to be "/usr/share", so the "make install" phase tried to install
them into /usr/share/vi/catalog, which didn't exist in my 1.1B system.
I changed "datadir= /usr/share" to "datadir= ${DESTDIR}/usr/share"
to work around this (John Kohl told me this has since been fixed).
With that, off we go! I copy /usr/src/Makefile to /usr/src/Makefile.build
and make a few changes:
(1) I use "make obj", so it's peppered with "&& ${MAKE} obj" target additions
(2) Wrap the "beforeinstall:" target inside ".ifnmake build"/".endif"
(3) Replicate the "make distrib-dirs" segment of the "beforeinstall:" target
at the beginning of the "build:" target:
build:
.ifndef DESTDIR
(cd ${.CURDIR}/etc && ${MAKE} obj && ${MAKE} DESTDIR=/ distrib-dirs)
.else
(cd ${.CURDIR}/etc && ${MAKE} obj && ${MAKE} distrib-dirs)
.endif
[...]
With these changes, we're ready to roll. One last thing to remember is that
"make" doesn't automatically pick up on $DESTDIR being set and thus
re-target where it looks for system .mk files (which is probably a Good Thing;
maybe you *don't* want it to look elsewhere). Thus, you have to remember to
use make's "-m" option to make it look in ${DESTDIR}/usr/share/mk for the
.mk files.
In other words, at this stage it's
# cd /usr/src
# make -f Makefile.build -m ${DESTDIR}/usr/share/mk build >& make.out &
and sit back and watch the fireworks.
The only errors (warnings, actually) I got from the whole thing were the same
ol' Kerberos-related "test" errors:
[...]
test -L krb_err.et -a ! -e krb_err.et && rm -f krb_err.et
*** Error code 1 (ignored)
test -L kadm_err.et -a ! -e kadm_err.et && rm -f kadm_err.et
*** Error code 1 (ignored)
test -L ss_err.et -a ! -e ss_err.et && rm -f ss_err.et
*** Error code 1 (ignored)
test -L std_rqs.ct -a ! -e std_rqs.ct && rm -f std_rqs.ct
*** Error code 1 (ignored)
test -L kadmin_cmds.ct -a ! -e kadmin_cmds.ct && rm -f kadmin_cmds.ct
*** Error code 1 (ignored)
These have been discussed previously on "current-users".
There were some warnings in /usr/src/lib/libutil (during the "make depend"
stage) and various other places about "util.h: No such file or directory",
but they weren't fatal. This is because my existing 1.1B system doesn't
have <util.h>, and "mkdep" didn't grok this.
Aside/diversion: I wasn't quite sure why this was, because <bsd.sys.mk> adds a
"-nostdinc -idirafter ${DESTDIR}/usr/include" to CFLAGS if $DESTDIR is set.
And bsd.dep.mk's ".depend" target looked like it's supposed to invoke "mkdep"
with $CFLAGS as an argument ... yet in lib/libutil, it got invoked as
"mkdep -a -DLIBC_SCCS /usr/src/lib/libutil/*.c", essentially. In the libutil
Makefile, we have "CFLAGS+=-DLIBC_SCCS". So, the CFLAGS+= addition due to
$DESTDIR doesn't get picked up. This is presumably because <bsd.dep.mk> is
included/evaluated *before* <bsd.sys.mk> in <bsd.prog.mk>. I don't know what
would break if <bsd.sys.mk> was swapped with <bsd.dep.mk> at the end of
<bsd.prog.mk>. (Chris? Christos? Anyone? Bueller?)
Anyway, that's all she wrote, everything else worked like a champ!
Much thanks to all of you who made it possible to build into ${DESTDIR} this
way! I can now do a "normal" build into the system directories and know
that it will work. VERY very cool.
- Greg
P.S. One pet peeve: I had to do this 3 times, because my SPARCstation crashed
twice (while I wasn't around, natch) during the first 2 attempts.
For some reason, the crash messages aren't surviving to get put into
/var/log/messages, and "dmesg" only shows the post-reboot output. pk?
(Yes, I've now enabled crash dumps *grin*)
id NAA03758; Thu, 6 Jun 1996 13:48:56 +0900 (JST)
id NAA15451; Thu, 6 Jun 1996 13:48:55 +0900
id NAA07045; Thu, 6 Jun 1996 13:48:54 +0900
id AA28305; Thu, 6 Jun 96 13:45:07 JST
Message-Id: <199606060447.NAA06497@ecology.bell.flab.fujitsu.co.jp>
To: port-sparc@netbsd.org
Subject: Re: PFU,S-4_20H supported?
In-Reply-To: Your message of "Wed, 05 Jun 96 12:05:20 JST."
<199606050305.MAA02108@ecology.bell.flab.fujitsu.co.jp>
Mime-Version: 1.0 (generated by tm-edit 7.66)
Content-Type: text/plain; charset=US-ASCII
Date: Thu, 06 Jun 1996 13:47:55 +0900
From: =?ISO-2022-JP?B?GyRCTCcxOhsoQiAbJEI/PxsoQg==?= / MINOURA Makoto <minoura@kawasaki.flab.fujitsu.co.jp>
I've sent some incorrect information.
First, the model name for my box was PFU,S-4_20A. The product
name is S-4/20H, so I was confused. /platform/PFU,S-4_20A is
a symbolic link to sun4m.
Second, I looked at the ROSS web page, and found out that
RT625 was a cache controler! But the Solaris startup message
says that the processor is RT625. What does this mean?
I'll try the -current on the machine. Thank you.
# I'm afraid that PFU machines are available only for Japanese market.
--
MINOURA Makoto
minoura@flab.fujitsu.co.jp
Communication Network Systems Laboratory
Fujitsu Laboratories Ltd., Kawasaki, Japan
id <m0uSayF-000o5oC@pci.on.ca>; Sat, 8 Jun 96 23:15 EDT
Date: Sat, 8 Jun 1996 22:20:56 -0400 (EDT)
Message-Id: <199606090220.WAA00395@repeat.pci.on.ca>
From: David Gilbert <dgilbert@jaywon.pci.on.ca>
To: port-sparc@netbsd.org
Subject: Problems with more recent kernels.
I have had problems with recent kernels that I have tried on
my sun4/260. With almost any of the kernels that I've compiled
lately, I have had numerous incidents of unallocated inodes whenever I
reboot and an fsck comes up.
Typically the files are either on my news spool, in the uucp
spool. I've only lost a little email from this... and loosing a few
news articles won't bother me... but it makes me a little bit afraid.
I lost one of the kernels that I compiled while I was running the new
kernels this way.
Most of these times, the kernel has panic'd with a filesystem
error of one sort or another. Bad inode is one that comes to mind.
To be specific, this is happening when I compile from the
tarballs that I ftp'd on 960604.
Dave.
--
----------------------------------------------------------------------------
|David Gilbert, PCI, Richmond Hill, Ontario. | Two things can only be |
|Mail: dgilbert@jaywon.pci.on.ca | equal if and only if they |
|http://www.pci.on.ca/~dgilbert | are precisely opposite. |
---------------------------------------------------------GLO----------------
Date: Mon, 10 Jun 1996 18:07:52 +0200 (CED)
From: Krister Walfridsson <cato@df.lth.se>
To: Paul Kranenburg <pk@cs.few.eur.nl>
cc: thorpej@nas.nasa.gov, port-sparc@NetBSD.ORG
Subject: Re: Alas...
In-Reply-To: <9605302214.AA00677@cs.few.eur.nl>
Message-ID: <Pine.VUL.3.91.960610180044.12872A-100000@marvin.df.lth.se>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Fri, 31 May 1996, Paul Kranenburg wrote:
> > > sd0 at scsibus0 targ 3 lun 0: <QUANTUM, TRB850S, 0404> SCSI2 0/direct fixed
> > > sd0: dma0: xfer (-65528) > req (8)
> > > esp0: !TC [intr 10, stat 83, step 4] prevphase 1, resid 8
> > > 810MB 3653 cyl, 4 head 113, sec, 512 bytes/sec
>
> Obviously, something has made it to the host, as the inquiry and mode sense
> commands have at least produced some reasonable output.
I have found the bug... and here comes the fix (to be applied to -current
sys/arch/sparc/dev/esp.c from May 22)
*** esp.c.old Mon Jun 10 13:12:50 1996
--- esp.c Mon Jun 10 14:02:31 1996
***************
*** 230,236 ****
printf(": ESP100");
sc->sc_rev = ESP100;
} else {
! sc->sc_cfg2 = ESPCFG2_SCSI2 | ESPCFG2_FE;
ESP_WRITE_REG(sc, ESP_CFG2, sc->sc_cfg2);
sc->sc_cfg3 = 0;
ESP_WRITE_REG(sc, ESP_CFG3, sc->sc_cfg3);
--- 230,236 ----
printf(": ESP100");
sc->sc_rev = ESP100;
} else {
! sc->sc_cfg2 = ESPCFG2_SCSI2;
ESP_WRITE_REG(sc, ESP_CFG2, sc->sc_cfg2);
sc->sc_cfg3 = 0;
ESP_WRITE_REG(sc, ESP_CFG3, sc->sc_cfg3);
/Krister
id <m0uTKnO-000o5oC@pci.on.ca>; Tue, 11 Jun 96 00:11 EDT
Date: Mon, 10 Jun 1996 23:57:02 -0400 (EDT)
Message-Id: <199606110357.XAA00672@repeat.pci.on.ca>
From: David Gilbert <dgilbert@jaywon.pci.on.ca>
To: port-sparc@netbsd.org, pk@netbsd.org, thorpej@netbsd.org
Subject: ie1 crash solved with rfc 1323!
With all this talk of RFC 1323 mucking things up, I decided to
see how it would affect my little network here by turning it off. I
have been having a problem for awhile that a spammed TCP connection
such as a telnet or ftp connection with large amounts of _outbound_
traffic from my sparc would cause it to hang with some memory errors
in areas of the kernel I just wasn't understanding.
When I was reading about the RFC 1323 stuff in the list, I
fired up my net connection and had a look at it online. It hit a
little too close to home... so I...
sysctl -w net.inet.tcp.rfc1323=0
on both my sparc and i386 box (both running NetBSD). Volia!
No more crashes. Before, I could generally reproduce the problem by
telneting between them and asking for an 'll' of a large directory. I
have been stressing the connection for awhile now, and everything's
running great!
Anyways, I suppose that I'll add my vote for turning rfc1323
off ... at least 'till it's working correctly.
Dave.
--
----------------------------------------------------------------------------
|David Gilbert, PCI, Richmond Hill, Ontario. | Two things can only be |
|Mail: dgilbert@jaywon.pci.on.ca | equal if and only if they |
|http://www.pci.on.ca/~dgilbert | are precisely opposite. |
---------------------------------------------------------GLO----------------
id <m0uTLJh-000o5oC@pci.on.ca>; Tue, 11 Jun 96 00:44 EDT
Date: Tue, 11 Jun 1996 00:37:07 -0400 (EDT)
Message-Id: <199606110437.AAA00506@repeat.pci.on.ca>
From: David Gilbert <dgilbert@jaywon.pci.on.ca>
To: port-sparc@netbsd.org, pk@netbsd.org, thorpej@netbsd.org
Subject: False alarm, but data still interesting.
Geez... my UUCP connection came up to flush the mail queue
even before the login prompt appeared. Anyways... a bit more data to
add about the RFC 1323 and my ie1 on my Sun4/260.
Previously, data on the order of 1K would cause the panic to
happen. In more intensive testing, with rfc1323 turned off, it takes
about 50K of data to panic the machine. Just to bring everyone who
hasn't been following this up to speed, trace from ddb gives me this
general series of functions each time it panics:
iestart()
ether_output()
ip_output()
tcp_output()
tcp_input()
ipintr()
soft01intr()
The fact that tcp_output() and tcp_input() are there is a real
common thread to the crashes. Now... when I examine the core file
that is generated without complaint, gdb says:
[2:4:304]dgilbert@repeat:/var/crash> gdb -k /netbsd netbsd.2.core
GDB is free software and you are welcome to distribute copies of it
under certain conditions; type "show copying" to see the conditions.
There is absolutely no warranty for GDB; type "show warranty" for details.
GDB 4.11 (sparc-netbsd), Copyright 1993 Free Software Foundation, Inc...
(no debugging symbols found)...
/u/crash/netbsd.2.core: Bad address.
(kgdb) quit
[2:5:305]dgilbert@repeat:/var/crash>
So... I assume that things arn't entirely happy.
Dave.
--
----------------------------------------------------------------------------
|David Gilbert, PCI, Richmond Hill, Ontario. | Two things can only be |
|Mail: dgilbert@jaywon.pci.on.ca | equal if and only if they |
|http://www.pci.on.ca/~dgilbert | are precisely opposite. |
---------------------------------------------------------GLO----------------
Message-Id: <199606122034.QAA22835@inner.net>
To: port-sparc@netbsd.org
Subject: Can't boot on a SS20
Date: Tue, 11 Jun 1996 17:39:03 -0400
From: Craig Metz <cmetz@inner.net>
I'm trying to boot NetBSD on a SparcStation 20. I have an external
disk that, when plugged into a SS1, boots and runs NetBSD happily, both with
the snapshot kernel and my own -current kernel. So far so good. (The disk
is on target 2 and my -current kernel maps tg2 to sd0 so the pack will
coexist happily both with our SunOS/Solaris drives and with our 4.4BSD drives)
When I take this over to the SS20 and do a:
ok boot disk2 netbsd
I get back the usual messages indicating it is about to boot, then,
when the UCB copyright would usually be displayed, I get:
Data Access Exception
Can anyone suggest a fix? I'd be a much happier camper doing my
kernel builds on a SS20 than a SS1!
-Craig
From: wilkes@ekrl.com
id AB12772; Tue, 11 Jun 96 16:36:13 PDT
Date: Tue, 11 Jun 96 16:36:13 PDT
Message-Id: <9606112336.AB12772@ekrl.com>
To: port-sparc@netbsd.org
Subject: Tadpole SPARCbook3 port??
Does NetBSD/sparc run on the Tadpole SPARCbook 3 laptop?
Any help/pointers greatly appreciated.
John Wilkes
inet: wilkes@Wilkes.COM | Tactics is knowing what to do when
usps: Box 2230; Aptos, California | there is something to do. Strategy
code: 95001-2230 | is knowing what to do when there is
icbm: 37 02 30 N / 121 48 45 W | nothing to do. --Savielly Tartakover
`[1;34;46mNet-Tamer V 1.03 - Test Drive