9 Jun
2000
9 Jun
'00
8:54 p.m.
Jason T. Miller <jasomill(a)shaffstall.com> wrote:
Sleep I can understand :)
I really think (and sure hope!) that write enabling the label area
will fix the problem.
Having to do a "disklabel -W" on a disk before doing 'raw' I/O was
a change that came in when labels were implemented. Before labels
the tables were compiled into the driver and 'raw' I/O could scribble
all over the disk and the system would still know about the
partitioning. When I ported over disklabels from 4.3-Reno it seemed
like a "Good Thing" to be paranoid about preserving the label sector ;)
Well, of course it does. But it's also well worth keeping in mind that
while pcc is generally inferior to gcc in almost every other way, due
to its simplicity it *is* probably at least five times as fast. A lot
of the difference in speed we're talking about here, particularly
with regard to the kernel, is due to the use of a much slower compiler;
as much of the kernel as you *have* to build for a VAX (as opposed to
what you *can* build if you *want to*) hasn't really bloated a lot
between 4.3 and NetBSD. Runtime memory use is a somewhat different
matter, but we do still fit into Ragge's smaller VAXen pretty well.
Thanks to Michael for reminding me exactly what the situation with
the optimizer and kernel builds under 4.3 is. Though I think he
forgot to mention "inline" (ack! pffffft!)... :-)
--
Thor Lancelot Simon tls(a)rek.tjls.com
"And where do all these highways go, now that we are free?"
Received: (from major@localhost)
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) id FAA37445
for pups-liszt; Sat, 10 Jun 2000 05:21:16 +1000 (EST)
(envelope-from owner-pups(a)minnie.cs.adfa.edu.au)
(my loving father having discarded my
DECmate II as junk about ten years ago).
Then call your nearest DEC dealer, get a quote on the replacement price, and
sue your dad for the cost! Or report him to NKVD for vandalism of socialist
property.
--
Michael Sokolov Harhan Engineering Laboratory
Public Service Agent International Free Computing Task Force
International Engineering and Science Task Force
615 N GOOD LATIMER EXPY STE #4
DALLAS TX 75204-5852 USA
Phone: +1-214-824-7693 (Harhan Eng Lab office)
E-mail: msokolov(a)ivan.Harhan.ORG (ARPA TCP/SMTP) (UUCP coming soon)
Received: (from major@localhost)
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) id BAA34141
for pups-liszt; Sat, 10 Jun 2000 01:19:23 +1000 (EST)
(envelope-from owner-pups(a)minnie.cs.adfa.edu.au)
From Roger Ivie <rivie(a)teraglobal.com> Sat Jun 10
01:16:50 2000
Received: from ns1.teraglobal.com (ns1.teraglobal.com [63.210.171.3])
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) with ESMTP id BAA34137
for <pups(a)minnie.cs.adfa.edu.au>; Sat, 10 Jun 2000 01:19:19 +1000 (EST)
(envelope-from rivie(a)teraglobal.com)
Received: from [10.10.50.26] (208.186.13.23) by ns1.teraglobal.com with ESMTP
(Eudora Internet Mail Server 2.2.2); Fri, 9 Jun 2000 08:16:57 -0700
Mime-Version: 1.0
X-Sender: rivie(a)ns1.teraglobal.com
Message-Id: <v0421010ab566b93d974a(a)[10.10.50.26]>
In-Reply-To: <Pine.BSF.4.21.0006090342330.4154-100000(a)lizard.indiana.edu>
References: <Pine.BSF.4.21.0006090342330.4154-100000(a)lizard.indiana.edu>
Date: Fri, 9 Jun 2000 09:16:50 -0600
To: pups(a)minnie.cs.adfa.edu.au
From: Roger Ivie <rivie(a)teraglobal.com>
Subject: Re: RX50 read/write on FreeBSD
Content-Type: text/plain; charset="us-ascii" ; format="flowed"
Sender: owner-pups(a)minnie.cs.adfa.edu.au
Precedence: bulk
Jason Miller wrote:
(not only
does it only support stdin and stdout, but it uses both 'goto' and the
ternary operator; I tend to deeply offend the C style gods, late at
night when I think nobody's watching)
Could be worse. I deeply offend the C style gods right in the open where
everyone can see. Since I'm pretty much a hardware type, I do _everything_
in state machines. While that works great for everything from hardware to
Prolog, it does mean my code tends to assume the only available
control structure is "if( expr ) goto state;". My attitude is that the
state diagram is the program, the code is just an implementation detail.
I used to work for a company that did TURBOchannel devices. I did the
device drivers for all the platforms (VAX/VMS, Alpha/VMS, Ultrix, and
OSF/1) and I shipped source code (it wasn't a conscious decision on the
part of management; since I got to build the distribution kits, the source
code was included and management simply didn't argue with me). One day I
got a letter from someone who had just bought our TURBOchannel parallel
printer port offering to go through the code and remove all those evil
gotos for the low, low price of only $100 a page. I declined the offer.
--
Roger Ivie
rivie(a)teraglobal.com
Not speaking for TeraGlobal Communications Corporation
Received: (from major@localhost)
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) id DAA37120
for pups-liszt; Sat, 10 Jun 2000 03:55:47 +1000 (EST)
(envelope-from owner-pups(a)minnie.cs.adfa.edu.au)
From Johnny Billquist <bqt(a)Update.UU.SE> Sat Jun
10 03:53:05 2000
Received: from Zeke.Update.UU.SE (IDENT:2026@Zeke.Update.uu.se
[130.238.11.14])
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) with ESMTP id DAA37116
for <pups(a)minnie.cs.adfa.edu.au>; Sat, 10 Jun 2000 03:55:41 +1000 (EST)
(envelope-from bqt(a)Update.UU.SE)
Received: from localhost (bqt@localhost)
by Zeke.Update.UU.SE (8.8.8/8.8.8) with SMTP id TAA10654;
Fri, 9 Jun 2000 19:53:06 +0200
Date: Fri, 9 Jun 2000 19:53:05 +0200 (MET DST)
From: Johnny Billquist <bqt(a)Update.UU.SE>
To: Thor Lancelot Simon <tls(a)rek.tjls.com>
cc: Michael Sokolov <msokolov(a)ivan.Harhan.ORG>, pups(a)minnie.cs.adfa.edu.au
Subject: Re: Newer BSD thingies....nice but then again....
In-Reply-To: <20000608113634.A26968(a)rek.tjls.com>
Message-ID: <Pine.VUL.3.93.1000609195015.10628B-100000(a)Zeke.Update.UU.SE>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-pups(a)minnie.cs.adfa.edu.au
Precedence: bulk
On Thu, 8 Jun 2000, Thor Lancelot Simon wrote:
On Wed, Jun 07, 2000 at 05:46:55PM -0500, Michael
Sokolov wrote:
Um. Let me put it this way... Userland is a *lot* smaller in 4.3 than
NetBSD... How much time do you think that makes up? The same goes for the
kernel. It's not that 4.3 is faster per se, just that it has a lot less to
build.
jkunz(a)unixag-kl.fh-kl.de wrote:
My experience with compilers on the VAX leads me to believe that the
substantial "savings" seen over NetBSD or post-4.3 BSD distributions here
is almost entirely due to the compiler and options used. If Quasijarus
builds like CSRG 4.3 did, with pcc, it can't even use the optimizer *at all*
for the kernel build, due to severe bugs; either way, pcc runs a lot faster
than gcc though it generates code that runs a whole lot slower. Oh, yes. My VS4000m60 needs only 36 hours to go
through a "make build".
This is pure luxury.
And 4.3BSD-Quasijarus completes its make build on my CSRG dev mill, which is a
KA655 (3.8 VUPs, whereas your KA46 is 12 VUPs), in a little under 4 hours. The
GENERIC vmunix kernel is another 30 minutes. I'd be willing to bet that gcc -O0 would build
NetBSD at least ten times
as fast as gcc -O2; the VAX is (as we all know ;-)) a "rather complex"
processor, with "rather complex" instruction patterns, gcc is not the
swiftest of compilers in the first place, and it does a *lot* of work.
True.
Slow machines *are* good for demonstrating how good
your compiler is;
I recall that rebuilding "compress" with gcc on my 750, way back when,
pretty much doubled the amount of Usenet news I could handle in a day. :-)
:-)
Johnny
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)update.uu.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
Received: (from major@localhost)
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) id EAA37260
for pups-liszt; Sat, 10 Jun 2000 04:45:23 +1000 (EST)
(envelope-from owner-pups(a)minnie.cs.adfa.edu.au)
From "Steven M. Schultz"
<sms(a)moe.2bsd.com> Sat Jun 10 04:42:16 2000
Received: from moe.2bsd.com
(0(a)MOE.2BSD.COM [206.139.202.200])
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) with ESMTP id EAA37256
for <pups(a)minnie.cs.adfa.edu.au>; Sat, 10 Jun 2000 04:45:09 +1000 (EST)
(envelope-from sms(a)moe.2bsd.com)
Received: (from sms@localhost)
by moe.2bsd.com (8.9.3/8.9.3) id LAA18214;
Fri, 9 Jun 2000 11:42:16 -0700 (PDT)
Date: Fri, 9 Jun 2000 11:42:16 -0700 (PDT)
From: "Steven M. Schultz" <sms(a)moe.2bsd.com>
Message-Id: <200006091842.LAA18214(a)moe.2bsd.com>
To: jasomill(a)shaffstall.com, sms(a)moe.2bsd.com
Subject: Re: RX50 on RQDX3 on 2.11BSD
Cc: pups(a)minnie.cs.adfa.edu.au
Sender: owner-pups(a)minnie.cs.adfa.edu.au
Precedence: bulk
From: "Jason T. Miller"
<jasomill(a)shaffstall.com>
Oh - ok. I must have misread the initial posting that indicated the
complete copy went thru
dd if=testrx50.img of=/dev/ra12a
800+0 records in
800+0 records out
If the writing of the floppy bailed out after "2+0" then it is no
wonder the compare later fails - only the first sector was written.
write: Read-only file system
2+0 records in
2+0 records out
That's what I get. After doing
the "disklabel -W ra9" the "dd" works fine and the floppy
compares identical to the input file.
Still haven't tried it. Had to watch the Pacers game and get some needed
sleep. I've gone over ra.c several times -- that's
a fun piece of code. I've
written device drivers before, but really, was this a test of DEC
software engineers by DEC hardware engineers?
You know - I think it was a contest inside DEC to see who would go
crazy first. Reading the comments in the Ultrix drivers gave me
the impression that even within DEC getting clear and correct
documentation wasn't a given. Then there are Chris Torek's comments
in the 4.3-Reno and later MSCP drivers when he was in essence reverse
engineering (or outright guessing) the MSCP commands, options, etc.
Well, all my serial cables are three-wire (yes,
I'm lazy, but I get
1.8K/sec via SLIP at 19200, so I'm not too concerned), but the 'numerous
other goodies' I like.
Hmmm, that's got to be a DHQ or similar. I had real problems with a
DHV-11 and character loss when going over 9600. Also, if you want
to use "Kermit" you have to have RTS/CTS because that's a fairly
heavy weight protocol and the system can't keep up if the rate is
too high. With RTS/CTS in place I was able to use 38400 and not
loose a single character.
what I know and love. Give me 2.11BSD on a PDP over
Solaris on an
UltraSPARC any day (well, if anyone wants to _give me_ and UltraSPARC,
Slowaris? "Just say no" - I have to deal with that at work and
it was light night and day going from SunOS 4.1.x to Slowaris 2.x
on the same hardware. You *need* an UltraSparc just to restore the
system responsiveness.
I'll do the responsible thing and reevaluate my
claims -- and SunOS [4.1.x
that is] is a decent OS, but anyway, I digress). The only thing I want is
Bit long in the tooth and missing a lot of the improvements (and
fixes) in the IP/TCP stack that have been made over time. Still, it
was a much nicer system.
command history and filename completion in the Bourne
shell (having grown
used to Bash -- although it's a big memory pig and I admit I use it only
for the previously mentioned features, though I like the PS variable magic
characters, too -- I'm thinking about trying to hack the CH features of
tcsh (never been a C shell fan) into sh, maybe we should start a 2BSD
'ports' collection? Any suggestions for a name of this shell? Any
suggestions for freeing up my time to write it :)?
Might I suggest "pig"? <grin!>
I like and use 'csh' for everything except the basic scripts that go
into the system. Csh has filename completion that works fairly well,
only thing it doesn't have is arrowkey driven command editing.
But observe the bloat factor that comes with "niceties" such as
command history and command editing:
First there's the honest to Bourne shell:
text data bss dec hex
16576 2356 416 19348 4b94 /bin/sh
Then take a look at /bin/csh where there's history and a nicer
(to me scripting capability - doing arithmetic in csh is so much
easier than in sh):
55744 7104 3682 66530 103e2 total text: 69120
overlays: 7360,6016
Overlaid! Efficiently (the one overlay is called seldom) but overlaid
none the less.
And lastly 'tcsh' (and yes, there is a port of an older version of
tcsh for 2.11):
48960 14844 11986 75790 1280e total text: 140864
overlays: 15424,16000,14144,14016,16256,16064
Zounds! No hope of really being efficient - modules were packed where
they would fit. More than doubling the size of 'csh' seems to be
a VERY high price to pay for using the arrow keys if you ask me.
Oh, and 'tcsh' has another problem due to it's appetite for memory.
If it runs out of D space (more likely since it's so much larger)
you get logged out. Doing filename completion in 'tcsh' and being
in a directory with too many files is a sure way to be staring at
the login prompt shortly there after ;)
Steven Schultz
sms(a)moe.2bsd.com
Received: (from major@localhost)
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) id FAA37372
for pups-liszt; Sat, 10 Jun 2000 05:01:53 +1000 (EST)
(envelope-from owner-pups(a)minnie.cs.adfa.edu.au)
From Thor Lancelot Simon <tls(a)rek.tjls.com> Sat
Jun 10 04:59:38 2000
Received: from mail2.panix.com (mail2.panix.com
[166.84.0.213])
by minnie.cs.adfa.edu.au (8.9.3/8.9.3) with ESMTP id FAA37368
for <pups(a)minnie.cs.adfa.edu.au>; Sat, 10 Jun 2000 05:01:48 +1000 (EST)
(envelope-from tls(a)panix.com)
Received: from panix3.panix.com (panix3.panix.com [166.84.0.228])
by mail2.panix.com (Postfix) with ESMTP
id CACD615573; Fri, 9 Jun 2000 14:59:38 -0400 (EDT)
Received: (from tls@localhost) by panix3.panix.com (8.8.8/8.7.1/PanixN1.0) id OAA06564;
Fri, 9 Jun 2000 14:59:38 -0400 (EDT)
Date: Fri, 9 Jun 2000 14:59:38 -0400
From: Thor Lancelot Simon <tls(a)rek.tjls.com>
To: Johnny Billquist <bqt(a)Update.UU.SE>
Cc: pups(a)minnie.cs.adfa.edu.au
Subject: Re: Newer BSD thingies....nice but then again....
Message-ID: <20000609145938.A6135(a)rek.tjls.com>
Reply-To: tls(a)rek.tjls.com
References: <20000608113634.A26968(a)rek.tjls.com>
<Pine.VUL.3.93.1000609195015.10628B-100000(a)Zeke.Update.UU.SE>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2i
In-Reply-To: <Pine.VUL.3.93.1000609195015.10628B-100000(a)Zeke.Update.UU.SE>; from
bqt(a)Update.UU.SE on Fri, Jun 09, 2000 at 07:53:05PM +0200
Sender: owner-pups(a)minnie.cs.adfa.edu.au
Precedence: bulk
On Fri, Jun 09, 2000 at 07:53:05PM +0200, Johnny Billquist wrote:
My experience with compilers on the VAX leads me to believe that the
substantial "savings" seen over NetBSD or post-4.3 BSD distributions here
is almost entirely due to the compiler and options used. If Quasijarus
builds like CSRG 4.3 did, with pcc, it can't even use the optimizer *at all*
for the kernel build, due to severe bugs; either way, pcc runs a lot faster
than gcc though it generates code that runs a whole lot slower.
Um. Let me put it this way... Userland is a *lot* smaller in 4.3 than
NetBSD... How much time do you think that makes up? The same goes for the
kernel. It's not that 4.3 is faster per se, just that it has a lot less to
build.