tlug: Sparc/Linux: 72-pin SIMMs and cutting edge kernels

["Stephen J. Turnbull" <turnbull@example.com>]

About SIMMs:

>>>>> Hajime == Hajime Saitou <hajime@example.com> writes:

    Hajime> I've tried using 72pin parity SIMM's in an "SLC", which is
    Hajime> like a ELC except bus speed and I think some other minor
    Hajime> stuff. It worked...
...
    Hajime> Well, I used to live in the UK. They are known for
    Hajime> understating things...

I've noticed, both facts :-).

Anyway, thanks to everybody who's replied on this topic.  The honest
vendor did not take back the 2 16MB SIMMs (Buffalo something or
others, 6000yen, details on request although Jim Tittsler said "just
steal from your neighbor" so anything should work) he sold me, and
they are now happily purring along in the Sparc, in tandem with a pair
of 4MB SIMMs from a company not to be named.  Thank 'em on my behalf,
John.

The "sagi" vendor walked off with the 2 SIMMs he sold me ("sensei wa
iranai node, o-adzukarishimasu"), and the accountant just told me that
no, we don't get reimbursed.  "Sagi" is therefore not slander/libel,
it is the literal truth.  Why two separate vendors?  Well, the AMI
machine has 4 SIMMs per bank, I originally planned to order 8MB SIMMs,
and just got 32MB = 2x16MB ("duh"), so I had to order two more.  Since
everybody keeps asking me why I do or don't do this or that, I'll
explain in advance: the Tsukuba-dai accounting office keeps extremely
strict control over choice of vendor.  They chose a new vendor for the
second round.  I think it's because both the faculty and the
accountants are extremely dishonest (by US standards, anyway; I know
that some Japanese think that Americans are congenital liars, despite
their offensive frankness), and the accountants, being Monbusho
jimu-kan, have the power to set things up so that they get the
kickbacks rather than the faculty.  My belief is that at most private
and former imperial universities, the faculty get the kickbacks.  I
know this is at least partly true from stories I have heard about
Christmas "bonuses in kind" from textbook sellers and computer dealers
from faculty at Kobe-dai and Todai.  Tsukuba-dai has the disdvantage
of being a Monbusho creature from the very beginning, so the jimu-kan
have the power.

About cutting edge kernels:

I'm having problems getting XEmacs betas to build right on
Sparc/Linux.  The executable gets linked, but when byte-compiling the
Lisp libraries, I always get core dumps.  It seems to be a memory
allocation problem (there are no stack traces in the core dumps; very
very bad).  (The extra memory seems to have helped, but we'll see.)

Since most other architectures seem to not have this problem, I
thought maybe upgrading the kernel would help.  Memory allocation is
always tricky since it needs to be really efficient.  Currently I'm at
2.0.27 (from the CD-ROM), I'm going to go to 2.0.31, maybe 2.0.32-pre2
tonight.  But what the hell, why not try the 2.1.x?  Anyway, to do
recent 2.1.x kernels (x >= 57), you need the new modutils.  So I built
them, but the kernel build failed.  The modutils claim backward
compatibility to 2.0.x, but they aren't.  Apparently there's a new
Sparc system call (#184) implemented in the 2.1.x kernels that isn't
there in 2.0.x, and insmod blows chunks all over the place.
Fortunately, either kerneld 2.1.55 doesn't use the system call or it
didn't get replaced for some reason, and accesses to the CD-ROM loaded
the sr_mod driver and everything was OK after reinstalling modutils
2.0.0 from RPM.

The reason that the 2.1.{57,...,63} builds fail is that the console
has been redesigned, but not reimplemented yet.  Unless I'm missing
something....

Anyway, what's the most recent 2.1.x kernel that anybody is running on 
a Sparc?
---------------------------------------------------------------
TLUG Thursday evening meeting: Nov. 13, 18:40 at Tokyo station
Starbuck's coffee.  Then move to dinner place at 19:00
info: demian@example.com | TLUG Saturday meeting is Dec. 13, 1997
---------------------------------------------------------------
a word from the sponsor:
TWICS - Japan's First Public-Access Internet System
www.twics.com  info@example.com  Tel:03-3351-5977  Fax:03-3353-6096