Dec 122017
 
Linux news from the Internet.
File INTLX015.ZIP from The Programmer’s Corner in
Category UNIX Files
Linux news from the Internet.
File Name File Size Zip Size Zip Type
INTLX015.TXT 152457 53480 deflated

Download File INTLX015.ZIP Here

Contents of the INTLX015.TXT file


From: [email protected] (Timo Jaakko Sillanp{{)

Subject: HELP! Problems with rootimage.
Message-ID:
Date: 7 Apr 92 12:48:47 GMT
Sender: [email protected] (Usenet pseudouser id)
Distribution: comp
Organization: Helsinki University of Technology, Finland
Lines: 31
Nntp-Posting-Host: blob.cs.hut.fi

I reported this problem two weeks ago but got no help.
So, here it is again.

I got rawrite.exe, bootimage and rootimage (0.95a) from nic.funet.fi.

I booted with bootimage diskette in drive A and Linux told
me 'Loading....'.Then I selected one of the listed video modes.
Linux told me to insert rootimage diskette in drive A and press
ENTER. I inserted it and pressed enter. Diskette drive light
and motor went on for a second and that was all. Nothing else
happened.

I got a new copy from nic.funet.fi and tried again - still
didn't work. I have tried it on three different machines
but it doesn't work! Unfortunately all of those machines
are 486/33 with AMI BIOS dated 05/05/91 or 06/06/91, so
I don't know whether it is a BIOS problem. My machine has
OPTI chipset, the other two have some other chipset.

Am I doing something totally wrong?
I get rootimage.Z from nic.funet.fi (yes, binary mode is on),
uncompress it and write it to a floppy disk (1.44Mb) using
rawrite.exe.
What should I do?

Please at least tell me to shut up if I'm wrong. This uncertainty
is killing me.

TS.

[email protected]


[next article]
From: [email protected] (Clifford A Adams)
Newsgroups: alt.os.linux,comp.os.linux

Subject: Re: Beginner's Guide: Installation (Possible errors?)
Summary: Possible confusion of partitions...
Keywords: linux, installation, partitions, correction
Message-ID: <[email protected]>
Date: 7 Apr 92 20:31:27 GMT
References: <[email protected]>
Distribution: alt
Organization: University of New Mexico, Albuquerque NM
Lines: 91


First of all, I would like to congratulate "I Reid" on an
excellent beginners guide. I beleive there are some errors in the
hard disk installation part, however. (It looks like a problem of not
updating some of the manual when other parts change.) I could be
wrong, but the four parts below look really suspicious...

[here are appropriate headers and my comments on the guide]

From: [email protected] (I Reid)
Subject: Beginner's Guide: Installation
Message-ID: <[email protected]>
Date: 2 Apr 92 22:18:33 GMT

******************************
Section II: Installation to Hard Disk
[...]

3.
[...]
then you use the "unnameable" command to partition your hard disk.
The format of this command is

n sys_id first_cluster last_cluster sys_name

where n ............ is the partition number
sys_id ....... is a magic number (ID in example above)
and sys_name ..... is optional and should not be used ('cos it
makes things fail (strange but true)

[...]
For the example disk I would type the following

pfdisk> 2 6 0 292
pfdisk> 1 128 293 521
[**********1: should these lines read:
pfdisk> 1 6 0 292
pfdisk> 2 128 293 521
(swap the partition #s)]
Explanation time..... partition 1 is now a Dos partition with 293
clusters (0 to 292 inclusive). Partition 2 is a Linux partition with
228 clusters.

Last bit... last but not least is the 'a' command ('a' for active)
which marks a partition as the one to boot from. I chose partiton 1
(DOS) 'cos Linux cannot boot entirely from hard disk without some
assistance (more later (eventually)). So ....

pfdisk> a 2
[**********2: should this read: "pfdisk> a 1" ?]

4. That wasn't too bad was it? Now all you need to do is make a Linux
filesystem, [...]

# fdisk

Disk 0:
/dev/hda1: 61516 blocks active 16-bit DOS (>=32M)
/dev/hda2: 48090 blocks unknown partition type 0x80
[...]

Next you use the mkfs command (make filesystem) with the following
format

mkfs -c device blocks

the -c option specifies that it should perform some checking as it
works and the blocks figure is the one from fdisk (above). To make
a file system on this partition I would type

# mkfs -c /dev/hda1 48090
[**********3 should this line be: "# mkfs -c /dev/hda2 48090" ?]

Ah, finally.... you are ready to install to HD! Thanks to Jim
Winstead's excellent installation scripts this is easy as pie so
without further ado type the following

# mount /dev/hda1 /mnt
[**********4 should this line be: "# mount /dev/hda2 /mnt" ?]

******************************

Again, thanks for the effort in making installation much easier...
I'm going to try Linux from floppy today, and will probably HD install
it as soon as I get positive replies to this message...

--
Clifford A. Adams --- "Understanding is inasmuch as becoming."
[email protected] | 457 Ash St. NE/Albuquerque, NM 87106 | (505) 242-4519
Programmer in the USENET Moderation Project: tools for advanced filtering
(many-moderator groups, keyword/summary addition, prioritizing newsreaders...)


[next article]
From: [email protected] (I Reid)
Newsgroups: alt.os.linux,comp.os.linux

Subject: Re: Beginner's Guide: Installation (Possible errors?)
Keywords: linux, installation, partitions, correction
Message-ID: <[email protected]>
Date: 8 Apr 92 14:39:28 GMT
References: <[email protected]> <[email protected]>
Distribution: alt
Organization: Edinburgh University
Lines: 38

I'm not doing well this week with my beginner's guide to installation ๐Ÿ™‚

[email protected] (Clifford A Adams) writes:

>pfdisk> 2 6 0 292
>pfdisk> 1 128 293 521
>[**********1: should these lines read:
> pfdisk> 1 6 0 292
> pfdisk> 2 128 293 521
> (swap the partition #s)]

Quite correct! I changed things around on my drive because of a still
unresolved problem with bad blocks and I forgot to change some bits of the
guide. For the record, my partition 1 should be a Dos partition (type 6) and
partition 2 should be a Linux partition (type 128).

>pfdisk> a 2
>[**********2: should this read: "pfdisk> a 1" ?]

Yep.

># mkfs -c /dev/hda1 48090
>[**********3 should this line be: "# mkfs -c /dev/hda2 48090" ?]

># mount /dev/hda1 /mnt
>[**********4 should this line be: "# mount /dev/hda2 /mnt" ?]

Again both your changes are correct.

> Again, thanks for the effort in making installation much easier...
>I'm going to try Linux from floppy today, and will probably HD install
>it as soon as I get positive replies to this message...

I hope all goes well for you and I'm very sorry about the errata I posted. What
can I say, it was late, I was tired, I made a mistake. Sorry. I will repost a
(hopefully) correct guide tonight (and try to upload it to tsx or somewhere).

Iain


[next article]
From: [email protected] (Timo Jaakko Sillanp{{)

Subject: HELP! Problems with rootimage. (UPDATE)
Message-ID:
Date: 8 Apr 92 23:31:38 GMT
Sender: [email protected] (Usenet pseudouser id)
Distribution: comp
Organization: Helsinki University of Technology, Finland
Lines: 76
Nntp-Posting-Host: blob.cs.hut.fi

A couple of days ago I complained that rootimage does not work
on my computer. It still doesn't. But I have got a lot of
suggestions (many of which are quite similar to each other).
Thanks to all who tried to help me.
Here is a detailed list of my hardware and what I have already
tried.

WB 486/33, cache 64Kb, AMI BIOS 06/06/91, OPTI chipset
8Mb 80ns RAM (yeah I know they are slow), cache is 25ns
ISA bus
some cheap 16-bit Seagate IDE hard disk & two floppy disk
controller
drive 0: Quantum 40A, 40Mb
drive 1: Seagate ST-1144A, 130Mb
two floppy drives: A=1.44Mb, B=1.2Mb
Paradise compatible 16-bit CVGA 1024 card with 512Kb RAM
Genius 6000 serial mouse at COM1
Ventel 1200bps modem card at COM2
Morse VGAplus 800x600 SVGA color monitor
Sound Blaster 1.5, IRQ 7, port 220h
I don't know about the I/O card, but it has centronics, joystick
and two RS ports (joystick is switched off since Sound Blaster
has one)

I have two problems.

Problem one: rootimage diskette does not load.
After I put it in drive A and press enter, diskette drive
motor turns on for a couple of seconds but it sound like
nothing is loaded. Same problem on two other 486 machines
with AMI BIOS (but not with OPTI chipset).

Problem two: Linux complains about HD time out errors.
Solution: tell BIOS that there are no hard disks.
The other two machines does not have this problem.

What I have tried to solve problem one:
I have tried to make my system as slow as possible by switching
turbo off, setting memory wait states to max, turning internal
and external CPU caches off, disabling shadow RAM and dropping
bus speed to CLKI/6 (33Mhz/6 instead of normal CLKI/2.5).
Does not help.

I have tried to get many copies of the rootimage file, even from
two places: nic.funet.fi (rootimage) and sc.tamu.edu (miniroot
and boot_fd_ide). I have written those files to floppies using
rawrite.exe which I got from nic.funet.fi.

I have tried formatting my floppies with MS-DOS 5.0's format.com
and PC Tools 6.0's format (PCShell's format, not PCFormat) and
FDFormat.

Does my computer hang totally after the rootimage fails? No,
I can get task info by pressing Scroll Lock or Ctrl-Break. I can
write on the screen, but it has no effect. Task info:

0: pid=0, state=1, father=0, child=1, 2596/2932 chars free in
kstack, PC=0000B0DB
1: pid=1, state=2, father=0, child=-1, 2376/2932 chars free in
kstack, PC=0000B1AA

If I tell my BIOS that there are no hard disk, 2376/2932 changes
to 2600/2932.

I think that is all. To do:
I will try to verify with Norton's NU that rootimage file is
identical to the rootimage diskette.

Does anyone have more suggestions? Plug off all cards one by one
to see which causes the problem?
If it was only my machine, but I have problems with three different
machines.......????

TS.

[email protected]


[next article]
From: [email protected] (Jim Winstead Jr.)

Subject: Re: HELP! Problems with rootimage.
Message-ID: <[email protected]>
Date: 9 Apr 92 19:54:45 GMT
References: <[email protected]
.ac.uk>
Sender: [email protected] (The News System)
Distribution: comp
Organization: Harvey Mudd College, Claremont, CA 91711
Lines: 19

In article <[email protected]> [email protected] writes:
>In article , [email protected] (Timo J
aakko Sillanp{{) writes:
>> Am I doing something totally wrong?
>> I get rootimage.Z from nic.funet.fi (yes, binary mode is on),
>> uncompress it and write it to a floppy disk (1.44Mb) using
>> rawrite.exe.
>
>the rootimage is (I think !) for a 1.2Mb floppy. Try to find a rootimage for a

>1.44 disk - this could be the problem ...

The root image is 1.2 megs because that is the only size that will fit
on both a 1.2 meg 5.25" disk _and_ and 3.5" 1.44 meg disk. There are
not seperate 3.5" and 5.25" root images.

--
Jim Winstead Jr. (CSci '95) | "Catch a fish!"
Harvey Mudd College | -Geddy Lee,
[email protected] | San Diego Sports Arena
Disclaimer: Mine, not theirs! | January 20, 1992


[next article]
From: [email protected] (Farhan H Garib)

Subject: Re: HELP! Problems with rootimage.
Message-ID: <[email protected]>
Date: 9 Apr 92 19:49:22 GMT
References: <[email protected]
.ac.uk>
Sender: [email protected] (The Network News)
Reply-To: [email protected] (Farhan H Garib)
Distribution: comp
Organization: Columbia University
Lines: 30
Nntp-Posting-Host: cunixb.cc.columbia.edu

In article <[email protected]> [email protected] writes:
>In article , [email protected] (Timo J
aakko Sillanp{{) writes:
>> ...
>
>> Am I doing something totally wrong?
>> I get rootimage.Z from nic.funet.fi (yes, binary mode is on),
>> uncompress it and write it to a floppy disk (1.44Mb) using
>> rawrite.exe.
>
>the rootimage is (I think !) for a 1.2Mb floppy. Try to find a rootimage for a

>1.44 disk - this could be the problem ...
>
>Tony
>
> _
>Tony Kew, Oxford University Computing Services, (_)xford, England, U.K.
>====================================================|=========================
>Janet:- [email protected] --+-- Disclaimer :-
>Internet:- tony%[email protected] | Who is the more foolish?
>Earn/Bitnet:- tony%[email protected] | The fool, or the
>Phone:- +44 (0)865 273268 Fax:- +44 (0)865 273275 | man who follows him?
>==================================================\=|=/=======================
> \|/




I use 1.44MB floppies without a problem.

-Greg


[next article]
From: [email protected] (Bowen (Laughing Cheetah))

Subject: Re: HELP! Problems with rootimage.
Message-ID: <[email protected]>
Date: 9 Apr 92 20:49:12 GMT
References: <[email protected]
.ac.uk>
Sender: [email protected]
Distribution: comp
Organization: Purdue University Computing Center
Lines: 22

In article <[email protected]> [email protected] writes:
>In article , [email protected] (Timo J
aakko Sillanp{{) writes:
>> ...
>
>> Am I doing something totally wrong?
>> I get rootimage.Z from nic.funet.fi (yes, binary mode is on),
>> uncompress it and write it to a floppy disk (1.44Mb) using
>> rawrite.exe.
>
>the rootimage is (I think !) for a 1.2Mb floppy. Try to find a rootimage for a

>1.44 disk - this could be the problem ...

The images I have seem to work without any difficulties for either 1.2MB or
1.44MB disks. Since (to my understanding) they are raw images, it dosen't
really matter how big the disk is, just as long as the entire thing fits.

-bg
--
________________________________________________________________________________
Bowen Goletz
Purdue University Midi 'n Music
[email protected] [email protected]


[next article]
From: [email protected] (Drew Eckhardt)

Subject: Re: AS86
Message-ID: <[email protected]>
Date: 17 Apr 92 05:54:51 GMT
References: <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 44
Nntp-Posting-Host: kinglear.cs.colorado.edu

In article <[email protected]> [email protected] (Rob Coleman) write
s:
>I'm curious just what is necessary in an assembly source file to create
>an executable with either as/ld or as86/ld86. Maybe a better question
>is which is the preferred assembler (if there is one), and what is the
>basic format of the file, and method to assemble and link it? Ultimately,

as86 and ld86 are used for two files in the kernel source tree,
boot/setup.S, boot/bootsect.S. They create 80x86 REAL mode code.

You MUST use gas (as) and gld (ld) for 386 code. gas uses that AT&T
syntax, which differs radically from intel syntax.

Texinfo gas documentation is available from prep.ai.mit.edu under
the usual /pub/gnu directory, so I won't cover the Intel / AT&T syntax
differences other than saying you need the documentation.

Also, remember that as is designed to read compiler output, not human
generated code, and if you want ANY macro capabilities, use the
C preprocessor. Also, when given a .S file, gcc 2.x will run
cpp on it before passing it to as, and running gcc on a .s file
will 'do the right thing'

>I will probably just be hacking the assembly portion of the kernel which
>doesn't require that I know this, but I'd also like to be able to create
>standalone programs.
>Thanks..
>

For very small things, that have minimal use of the library functions
(or only syscalls), you can just make an assembler program, which is
entered at address zero, and has a call to __exit, assemble it,
and link it to libc.a. For more substantial things,
write a 'C' driver, or _main function in assembler

ie
.globl main
_main:

and link as a normal 'C' program.

Interfacing to 'C' is done in the usual way, with parameters pushed
in reverse order, on a 32 bit wide stack, 32 bit return address, caller
removing parameters. Naming convention is standard 'C', with all
global symbols prefixed with an _.


[next article]
From: [email protected] (Joseph Knapka)

Subject: virtual consoles/pty's
Message-ID: <[email protected]>
Date: 17 Apr 92 06:46:10 GMT
Organization: University of Georgia, Athens
Lines: 6

Here's a silly question: When I boot linux, it says "8 virtual
consoles, 4 ptys". I know what ptys are, I think -- the guys I get to
by alt-f[1-4]. So my question is, "What, exactly, is the difference
between a pty and a virtual console?"

Thanks,


[next article]
From: [email protected] (Linus Benedict Torvalds)
Newsgroups: comp.os.linux,alt.os.linux

Subject: Re: more on my problem(I still haven't got any help) : )
Message-ID: <[email protected]>
Date: 17 Apr 92 08:03:04 GMT
References: <[email protected]>
Organization: University of Helsinki
Lines: 21

In article <[email protected]> [email protected] writes:
>Ok, here goes:
>I had xcomm, it was running fine. suddenly, it started outputting to the
>screen 2 cr/lf for every one I hit on the keyboard. I went into kermit
>to see if it was xcomm, or the place I am calling. kermit was doing
>this:
>C-Kermit>set line /dev/ttys4
>
>C-kermit> set speed 2400

At least the double-spacing in kermit is due to login first enabling
ECHONL on the tty for the password recognition (it naturally wants no
echo on normal characters, but wants the newline to be echoed), and it
doesn't disable it afterwards. This doesn't result in any problems for
most things, but some programs (at least my port of kermit) seem not to
notice it: kermit disables echoing, but doesn't disable the ECHONL.

The cure for kermit is to write "stty -echonl" in your .profile or
similar - I don't know if this helps xcomm as well.

Linus


[next article]
From: [email protected] (Richard Alan Brown)

Subject: problems booting
Keywords: bootdisk
Message-ID: <[email protected]>
Date: 17 Apr 92 09:48:40 GMT
Organization: School of Physics, University of Melbourne
Lines: 15


A problem, not from [email protected] but it may interest him also.

I used rawrite to make a bootable disk and a rootimage disk. When I
boot my pc with the floppy in, I get the loading............
and the SVGA mode question.
when I press any other key, the system simply reboots, and cycles through
this same process.

Hmm.

Its probably obvious, but any ideas what might be going wrong?

alistair scott
[email protected]


[next article]
From: [email protected] (Craig Burley)

Subject: Re: GCC2.1 (Wonderful !)
Message-ID:
Date: 17 Apr 92 07:43:12 GMT
References: <[email protected]>
<[email protected]>
<[email protected]>
<[email protected]>
<[email protected]>
Sender: [email protected]
Organization: Free Software Foundation 545 Tech Square Cambridge, MA 02139
Lines: 195
In-reply-to: [email protected]'s message of 16 Apr 92 23:08:09 GMT

In article <[email protected]> [email protected] (Joh
n Plevyak) writes:

As I remember (from long ago) the Primos (the Prime operating system)
replaced all dynamic calls with calls to illegal addresses.

Specifically, I believe the linker did the replacing when building a program
image file (EPF or old-style) -- so the file had pointer-faulted references
to dynamic links (remove the "pointer-fault" bit in the pointer and you have
a pointer to a CHAR(*) VARYING string having the name of the routine needed.

A table in
the image mapped the index of the illegal address to a filename/symbol
pair. The filename was found or loaded and the address was patched. This
has the advantage that you could dynamically replace the library without
relinking (the linker was in the OS). This could be used on a 386 class
machine.

Well, you have to make sacrifices to do this on a 386 that you couldn't make
on Prime 50 series machines. Specifically, all 50 series machines had only
one general format for indirect pointers that imposed the pointer-fault, ring,
and extension bits on all such pointers.

The flat-address model for the 486 allows a task to address all 4 gigabytes
representable by a 32-bit pointer. In this form, dynamic linking is still
possible but adds an expense to each unsnapped pointer -- unless you reduce
the accessible space to 2GB by allocating one bit as the "fault" bit (so the
unfaulted version points to a string) or some such thing, you need more
memory for each faulted pointer.

The segmented pointers of the 486 (48-bit pointers) offer slightly more
opportunity for providing Prime-style unsnapped pointers without significantly
infringing on the overall accessible space in a program, I think. For example,
(and I'm just guessing here, after only reading the 486 arch manual once),
two segment numbers could be reserved to "unsnapped pointer to name in CS
seg space" and "unsnapped pointer to name in DS space" or some such thing.

But the lack of an explicit "fault" pointer means 486 OS architects do have
to make a tough choice when deciding on how to provide dynamic linking (the
"true" kind -- it's still quite easy to do the semi-dynamic kind that VMS
does, where all unsnapped pointers are snapped by the OS when an executable
is mapped to memory).

The disadvantages were:

1) your program could suddenly fail if the new library (installed by
someone else) had a bug

Yes, but it taught most of us Prime OS and library engineers to get our act
together, since even our pet programs would break if we screwed up a new
library install even before we'd relinked the programs! I.e. this might
really be an advantage overall. It teaches people to treat library changes
and testing more like they're doing OS work than doing single-program work.

2) you could not page directly off the executable

Either I misunderstand you or this is wrong. EPFs were directly paged into
memory as needed (that is, the "text" or code-space portion). PRIMOS generally
had pointers (especially unsnapped ones) in read/write segments (pointed to
by the Link Base, LB%) along with ECBs (procedure Entry Control Blocks) and
such. So, normally, changing these things to accommodate the latest placement
of a newly invoked EPF (program/library) didn't require changing pure code,
so that code could be paged in on demand, making startup potentially quite
fast. Some applications required pointers and ECBs to be placed in pure code
to improve shareability, but naturally this meant things they pointed to had
to be statically placed (typically in system-wide shared and unshared segments
rather than per-user shared/unshared segments), so they don't really count
in this discussion.

Sheesh, it's been a long time since I've thought about this stuff. If I
screw it up, Kevin Cummings will likely correct me, since he probably knows
far more about it than I do (and as I recall, knew more back when I was still
at Prime)!

3) your operating system was forced to intervene on the first of
every dynamic call... thus slowing startup time.

Mmm, maybe, I'd say the penalty wasn't so much slower startup time, but
slower execution of newly reached paths with the payoff often being _faster_
startup time. PRIMOS might not have to even read much of the program in
at all, just some LB%-relative stuff to adjust pointers and ECB link-segment
numbers for the user's particular segment placement. If the program was
already shared (mapped) via either another user or a recent invocation of
the same program by the same user, that'd be very little time. It might be
that a comparatively few link-snaps later, the user'd get a useful response
(like a prompt), faster than the old way of having to block-read a monolothic
program entirely into a static place in memory (and with more flexibility than
the latter).

Problem was, if the user issued a command, it might well involve hundreds or
thousands of new link-snaps (dynamic links not yet resolved), so _that_
command's first invocation would be slower than subsequent invocations.

The idea was that amortizing the cost of loading and linking across the
entire execution of the program would often cut down on the need to snap
many links at all, since (particularly for large monolothic apps, like SPSS)
users typically wouldn't invoke all the code in the program per invocation.
So it was actually expected to be practical for a user to frequently invoke
a large monolothic app just to do a quick command or two. Of course, that
was based on assumptions regarding user interaction that are increasingly
coming into question today for at least some popular application domains.

Still, thinking about designing an OS for the 486 causes me to weigh a lot
of interesting questions, like: should a given task be able to stack
multiple programs in arbitrary places (segments) in its memory space as
PRIMOS permitted (with EPFs but not with old-style static images), given an
OS that permits cheap tasks? I.e. is the cost of allowing arbitrary placement
of programs in per-task virtual memory worth the benefit, when an alternative
(cheap processes) is available? Can it be provided in a manner that costs
non-users of the feature very little? (My guess: no.)

Also, whether to provide access to segments in a 486 OS (instead of the
highly convenient flat-4GB-space model people expect coming from VAXland)
is a sticky question, because _lots_ of people want the flat memory model,
but the segmented model _does_ offer advantages. Problem: on the 486, the
pointer formats are different, so what does providing _both_ models cost in
terms of the OS interface -- i.e. do you need two sets of OS calls?

The advantages were:
1) no need to relink shared library programs
2) no need to link to libraries that this execution of the program
did not require... finer grain linking could be a win here.

I wonder how useful #2 has been in practice in the PRIMOS world. PRIMOS
provided not just dynamically linked libraries, but statically linked ones
as well (and the PRIMOS entry points essentially belonged in the latter
category). This meant that any library you expected to be occasionally
invoked for a "quickie" but that was inherently large, you could just make
static in your system, so the only cost was the link-snap cost (name-space
search, snap pointer, continue) but no library-loading cost.

In general I don't think that you will find an obvious clean and simple winne
r
in a war of dynamic linking strategies if you ask more that a few people.

That's for sure. A lot of the answer depends on the question -- if you want
big monolithic apps that use a variety of big libraries but not necessarily
all of them each time the apps are run, the PRIMOS approach is probably quite
good. I don't know if most UNIX apps fall into this area; either the apps
are short and sweet (tools, filters, etc) or monolithic with little use of
multiple shared libraries (EMACS comes to mind).

Note that one of the ideas of shared libraries under PRIMOS was that they would
be the primary means for providing services, such as electronic mail, remote
file transfer, queueing, etc. Under UNIX, these services are typically
provided by separate processes, accessed through IPC via tiny libraries for
apps to just get at the IPC stuff (X follows this model, for example).
Sharing tiny libraries isn't as important as sharing the X server code, which
if you'd talked to the PRIMOS architects back when dynamic linking and
libraries were being designed, would have felt should go into shared
libraries and used shared memory and semaphores to do the display management
and IPC between server tasks. (I have always felt the server model was better
than the shared-memory model, but couldn't convince the PRIMOS architects
otherwise back when the issue was being decided for most subsystems in
1978/1979. Nevertheless, the resulting dynamic-linking model seems to be
well-suited for the shared-library-service model, whether or not you like
that latter model.)

Point being that for what "we" want to do in UNIX, and what I am thinking of
doing in my OS research and design, PRIMOS' dynamic linking may provide good
lessons but not necessarily a good model. It _does_ stretch ones' thought
a bit, though.

The simplest solution is usually the best though, and (not intending to start
a war here) while Primos did dynamic linking which was nice, the lack of
child processes, memory protection of 'shelled' programs, the segmented
architecture, system calls with a mix PL1/G and FORTRAN interfaces and assort
ed
other goodies made me wish for Unix with or without dynamic linking
(perhaps it has changed, this was 1986).

I last "knew" PRIMOS at rev 19.4 in 1985, which is why I refer to it in the
past tense; I don't know how much it has changed. I agree with the
disadvantages you cite, except that I am unwilling to categorically state
that segmented architecture is bad (yes, it's hard, but it offers wonderful
optimization opportunities at the hardware/microcode level from what I can
tell, and also offers higher bit-efficiency for instructions in memory; I
don't think these advantages translate well to RISC, however). The lack of
memory protection between programs in the same task (stacked programs) was
sometimes a pain, but the segmented architecture _itself_ mitigated this
significantly (i.e. programs didn't tend to just run off one segment into
another, because the necessary pointer arithmetic was rarely implemented
except when an array was known to span segments, itself a rare condition,
and in any case getting _contiguous_ arrays wasn't always possible). So
though it may have happened occasionally, I don't remember having to chase
down any inter-program-stomping bugs like we all do under Mac or DOS or
similar (PC-like) OSes; similarly, the stack being in separate segments (and
separately managed by the OS) also allowed certain bugs to be caught much
earlier than in a typical flat-model OS (though a paging OS can help provide
almost as early detection of stack overflows as did PRIMOS' segmented
model).

Sigh, I _am_ rambling -- must be late, and I must not want to get any real
work done! ๐Ÿ™‚
--

James Craig Burley, Software Craftsperson [email protected]
Member of the League for Programming Freedom (LPF)


[next article]
From: [email protected] (Drew Eckhardt)

Subject: Re: virtual consoles/pty's
Message-ID: <[email protected]>
Date: 17 Apr 92 09:43:24 GMT
References: <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 20
Nntp-Posting-Host: romeo.cs.colorado.edu

In article <[email protected]> [email protected] (
Joseph Knapka) writes:
>Here's a silly question: When I boot linux, it says "8 virtual
>consoles, 4 ptys". I know what ptys are, I think -- the guys I get to
>by alt-f[1-4]. So my question is, "What, exactly, is the difference

Nope. ALT-Fn selects virtual console n, which is /dev/tty[n].

>between a pty and a virtual console?"

A pty is a "Pseudo TeletYpe". It lets you give a pipe to something that
expects to have a "real" tty attached to it, such as most editors,
an xterm, etc. There are master (/dev/ptyp[n]) and slave ends
(/dev/ttyp[n]) to it. One program can open a master end, and other programs
can read / write to the slave end to get / send to the master.

A virtual console is a console... that is virtual - so you can have
multiple terminal sessions on one physical console. Basically, it's
indistringuishable from any other tty device.




[next article]
From: [email protected].gwu.edu (Mike Me)

Subject: SCSI / Adapted support.... I dont get it!
Message-ID: <[email protected]>
Date: 17 Apr 92 07:31:32 GMT
Sender: [email protected]
Organization: George Washington University
Lines: 38
Originator: [email protected]

Hi people, I relatively new to this scene and in need of some
assistance.

I have an EISA system, and I have been able to get LINUX to work.
It took me a while... but hey! Anyway, I really want to get away from
this floppy method of running things, but CAN NOT seem to get the
thing to recognize my HD.

I got the bootimage 95+, which supposedly has (some) support for the
ADAPTEC 17XX controllers.

I'm not quite sure what the problem is, but I thought that since the
same controller is also in charge of the floppies I might have a
chance
of getting this thing to work.

I end up with the folloing errors when running the bootimage 95+

0 host adapters detected
detected 0 disks , 0 tapes
hard disk I/O error
dev 0301 block 1
kernel panic, unable to boot.

Am I doing somethiong wrong?

Also, the 95+ tar file comes with some c source? Am I to compile this
first? How can I on a floppy system?

Needless to say I'm confused, and could not find any help that
pertained to my problem in the FAQ pages.

Thanx, and I hope someone out there knows what I need to do to get
this to work.


Thanx,
MikeE


[next article]
From: [email protected] (Todd Radel)

Subject: CBREAK mode, etc.: Questions
Summary: Is cbreak broken?
Message-ID: <[email protected]>
Date: 17 Apr 92 13:03:01 GMT
Organization: University of Delaware
Lines: 24

Thanks to everyone who helped me get my printer set up as /dev/lp1. Now
for some more questions:

1) Is CBREAK mode broken in the .95c+ release? I'm trying to write some
stuff using the curses library, and I explicitly say cbreak();, but
I still have to press after every key I hit. I also tried
compiling XBBS version 7.91, which exhibited the same symptoms.

2) How do I set up a swap *file*? Not a separate partition, but a file.

3) Is Linux set up to be 8-bit clean? If I set up a BBS, will 8N1 callers
have a problem?

AdvTHANKSance to all who respond...

-- Todd
(getting better at this Linux stuff, but still a little confused...)


--
* Todd Radel Sysop, Dickinson Domain BBS *
| CIS/English FidoNet: 1:150/160; VirtualNET @3023 |
| University of Delaware REGISTERED Alpha Colony VI! |
* "You don't need to be crazy to be a sysop, but it helps!" *


[next article]
From: [email protected] (Brian Chojnowski)

Subject: Re: Graphics and IPC questions...
Keywords: Graphic standard and how to IPC?
Message-ID: <[email protected]>
Date: 17 Apr 92 15:08:59 GMT
References:
Sender: [email protected] (News)
Organization: University of Illinois at Urbana
Lines: 6


I plan to add a direct to screen library. Basically I envision
creating a terminal type that is a superset of vt100. Nothing as fancy or
system hog like X. Just a characterset based graphix system. So while we are
at it, if anyone has some info somewhere on how to write assembly stuff
under linux, I can start porting some of my dos-mode screen routines.


[next article]
From: [email protected] (Joseph Knapka)

Subject: Re: virtual consoles/pty's
Message-ID: <[email protected]>
Date: 17 Apr 92 16:07:08 GMT
References: <[email protected]> <[email protected]
olorado.edu>
Organization: University of Georgia, Athens
Lines: 15

In article <[email protected]> [email protected] (Dre
w Eckhardt) writes:
>In article <[email protected]> [email protected]
(Joseph Knapka) writes:
>>Here's a silly question: When I boot linux, it says "8 virtual
>>consoles, 4 ptys". I know what ptys are, I think -- the guys I get to
>>by alt-f[1-4]. So my question is, "What, exactly, is the difference
>
>Nope. ALT-Fn selects virtual console n, which is /dev/tty[n].
>

So that means I have eight virtual consoles available, but I can only
use four of them with Alt-Fkey sequences... how can I use vc's 5 thru 8?

Thanks^2,

Joseph


[next article]
From: [email protected] (Joseph Knapka)

Subject: Re: virtual consoles/pty's
Message-ID: <[email protected]>
Date: 17 Apr 92 16:16:24 GMT
References: <[email protected]> <[email protected]
olorado.edu> <[email protected]>
Organization: University of Georgia, Athens
Lines: 17


>>
>>Nope. ALT-Fn selects virtual console n, which is /dev/tty[n].
>>
>
>So that means I have eight virtual consoles available, but I can only
>use four of them with Alt-Fkey sequences... how can I use vc's 5 thru 8?
>
>Thanks^2,
>
>Joseph

Sorry to reply to myself, but a little expirimentation revealed that
the answer is, "You use the getty program to enable logins on the
other vc's, and then use Alt-F[5-8] to get to them..."

Joseph


[next article]
From: [email protected] (Bill Bogstad)

Subject: Re: expr (GNU shellutils)
Keywords: Softscroll
Message-ID: <[email protected]>
Date: 17 Apr 92 16:19:35 GMT
References: <[email protected]
alter.bellcore.com> <[email protected]>
Organization: Department of Cognitive Science, JHU
Lines: 29

In article <[email protected]> [email protected] (Niels Skov Ols
en) writes:
>expr is in the shellutils-1.6.tar.Z (or therabouts) on prep.ai.mit.edu.
>In this package are some other nice utils too.
>
>I compiled it with GCC2.1 with only small adjustments to the makefiles
>(I don't have echo) and a little finetuning in obvious places. I didn't
>make notes as I went along, so I don't remember the exact steps. It
>should't be a problem though. Mail me if it is and I will try to
>reconstruct what I did.

Well, I'm trying to compile it with GCC 1.40 and have found the
following problems so far.

1. no echo - put the following in your /bin directory

#!/bin/sh

echo $*

2. va_list redefined in stdio.h vs. stdarg.h - I modified stdio.h to
include stdarg.h rather then defining va_list itself.

3. No bison - get the binaries out of binaries/compilers on tsx-11.mit.edu
(I haven't actually gotten this far yet...)

If I get a chance tonight, I'll add more to the list above...

Bill Bogstad
[email protected]


[next article]
From: [email protected] (LCDR Michael E. Dobson)

Subject: Badblocks on hd
Message-ID: <[email protected]>
Date: 17 Apr 92 17:17:31 GMT
Organization: Naval Medical Research & Development Command
Lines: 12

If this is in one of the docs somewhere, please point me to it.

How does one deal with badblocks on your hard drive? I am running into
some on both my root partition and my /usr partition. Is there some way
under linux to identify and map them out?

Thanks!
--
LCDR Mike Dobson, Sys Admin for | Internet: [email protected]
nmrdc1.nmrdc.nnmc.navy.mil | UUCP: ...uunet!mimsy!nmrdc1!rdc30
AT&T 3B2/600G Sys V R 3.2.3 | BITNET: [email protected] or [email protected]
WIN/TCP for 3B2 R3.2 | MCI-Mail: 377-2719 or [email protected]


[next article]
From: [email protected] (Steve Ansell)

Subject: TEX for Linux at tsx
Message-ID: <[email protected]>
Date: 17 Apr 92 18:16:53 GMT
Sender: [email protected]
Organization: Computing Services, UC Davis
Lines: 34


In the binaries/usr.bin directory on tsx-11 there is a directory containing
what I assume to be the TEX package. It contains the following files:


dvips.tar.Z
dvilj2p.tar.Z
ptex.tar.Z
texfile1.tar.Z
texpk300.tar.Z
textfm.tar.Z

It has been a long time since I have done anything with TEX, but I thought
that this would be an ample time (besides, if I can create documents under
Linux my DOS stuff will be almost completely useless!). The problem is
that there is no explaination (or none that I saw) as to what these files
are. Before I download almost 2MB of stuff, I would like to know just
what I do and don't need. I know that dvips and dvilj2 convert the TEX
output files into the respective printer language and I think that the
texpk300 and textfm archives contain fonts (yes? no?). The main program seems
to lay in ptex and I am not sure, but I think texfile1 contains style sheets.
Could someone more knowlegable about TEX than myself confirm or correct this
and possibly tell me just how much of this is absolutely needed (I have very
limited HD space right now). Of course, an installation guide would be
nice, but I can usually hack my way through getting things installed if it
is possible. If this is covered elsewhere (i.e. in some README I missed)
please direct me to the correct location.


--
-Steven T. Ansell
Unix Consultant
Computing Services U.C.D.



[next article]
From: [email protected] (Nefaratu the Boutellian)

Subject: Suggestion for Graphics
Message-ID: <[email protected]>
Date: 17 Apr 92 19:36:27 GMT
Sender: [email protected] (USENET News Service)
Organization: University of Delaware, Newark
Lines: 13
Nntp-Posting-Host: pecan.cns.udel.edu

To those who want a graphics library for Linux, you might profit from
a study of the source code to the new 16/256-color graphics library
for DJGPP, the port of GCC to 386-mode DOS. This new library came
out with version 1.06 and from working with betas I know it to be
quite powerful, but not a humungous remote - display system like X.

-T

--
-- Tom Boutell
President, Technology House, U of Delaware. Got a similar resident group at your
school? Please get in touch!
INTERNET:[email protected] SNAIL: P.O.Box 295, Newark, DE 19715


[next article]
From: [email protected] (Bill Jones)

Subject: Software index for linux
Message-ID: <[email protected]>
Date: 17 Apr 92 19:44:21 GMT
Organization: 'op
Lines: 57

One thread that seems to be remaining at a constant level in both
alt.os.linux and here in comp.os.linux is what software is in what
package. I know that I have had that trouble often (what, for instance
is in tubes.tar.Z??)

There has been some mention of keeping track of these packages in a
"master index" somewhere and I beleive that Marc Corsini has mentioned
trying to do that along with the FAQ.

Now that I have reliable and stable access to news and e-mail, I would
be willing to undertake such a project if enough people felt that it
would be useful. What I am envisioning is

1. An index of what is available, what package it is in,
and what it does (so far, it seems as though most
everything is available at banjo.concert.net but the
descriptions might still help people determine what they
need).

2. A list of software that is currently being ported to linux
and the status of such a port (if this can be obtained).

3. A list of all of the patches that have been made available
along with the expected source code that each patch expects
to find.

4. Any other software-related items of interest.


If something like this were available, I would hope that others, especially
newcomers to linux, would be able to use it. In addition, it would help
cut down on the duplication of porting efforts and let people who are
working on the same project team up if they desire.

This would require the assistance of the authors of various programs and
ports so that it would remain up to date and, of course, would not be
mandatory that anyone participate. But, it strikes me that this is the
time to try and get something like this set up -- before the volume of
linux software, familiar and unfamilar, becomes so great that no one
would desire to keep track of it.

As I said, if enough people express an interest in this, I would be more
than willing to undertake it. Since I am not a software wizard (not even
close) this would be my contribution to the linux community. So, how
about it? Let me know.


Bill


------------------------------------------------------------------------
Bill Jones 1-614-785-0358

[email protected] Oh, to be in Montana now
[email protected] that spring is here!
[email protected]
------------------------------------------------------------------------

[next article]
From: [email protected] (Chris Waters)

Subject: make and .95a kernel rebuild
Message-ID:
Date: 17 Apr 92 17:50:49 GMT
Organization: Netcom - Online Communication Services (408 241-9760 guest)
Lines: 36


Ok, I gather that GNU make won't cut it for the kernel rebuild or
something, but I'm still curious about this bit from the Makefile:

all

Version
@./makever.sh
[etc.]

Shouldn't there be some targets for "all"? And maybe for "Version" as
well? Do I have a damaged set of kernel sources? (I got 'em from
tsx-11).

Other comments:

Is there a limit on the size of the DOS files that mtools will handle?
I've had problems with files around 3 meg or so.

Shouldn't "compress" and "uncompress" in the distribution root fs be
links? I admit that this wouldn't save much space, but it would save a
tad. Also, there's no link to "zcat".

Is there a reason for the "dummy line" in /etc/passwd?

In bash, MAIL seems to default to /var/spool/mail/username, rather
than to /usr/spool/mail/username?

Ok, that's all I've come up with in a couple of days of not-very-much-
spare-time-hacking. I'm sure I'll be back with more comments,
suggestions, etc. TTFN.

p.s. killer system--I am very impressed. Major kudos to Linus. ๐Ÿ™‚
--
Chris Waters | the insane don't | I figure the odds be fifty-fifty:
[email protected]| need disclaimers | I just might have something to say. --FZ


[next article]
From: [email protected] (John Plevyak)

Subject: Re: GCC2.1 (Wonderful !) (dynamic linking)
Message-ID: <[email protected]>
Date: 17 Apr 92 21:01:36 GMT
References: <[email protected]> urchy.gnu.ai.mit.edu>
Sender: [email protected] (News)
Organization: Nat'l Ctr for Supercomp App (NCSA) @ University of Illinois
Lines: 29
Originator: [email protected]


I agree with [email protected] (Craig Burley), but I would
like to clarify and sum up a bit.

Craig writes:

> Either I misunderstand you or this is wrong. ... PRIMOS generally
> had pointers (especially unsnapped ones) in read/write segments (pointed to
> by the Link Base, LB%) along with ECBs (procedure Entry Control Blocks) and
> such.

I stand corrected. In general, however, the problem is that if the functions ar
e
not statically bound to offsets at the call site, dynamic linking either require
s
(1) another level of indirection or (2) a fixup at the call site. Solution (1)
can be made more efficient if supported in hardware, and solution
(2) requires impure code. In general, both these solutions have real performanc
e
drawbacks. In case (2) pointers to functions have to be handled as well.

Segmentation is related issue. Segments can provide the extra level of
indirection for dynamic linking, handles for memory management etc, not
to mention other advantages. The segment sizes are the issue ('286 and '386).
Too few large segments are not good for small blocks or functions, too many
small ones and you can't allocate large arrays. The good and the bad... life.

The model for Unix has been flat memory, sharing by pages and basically
KISS. I'm all for experimentation though. My feeling is the proof is in the
pudding. Build it. If it works, everyone will use it.

john



[next article]
From: [email protected] (Kevin Cummings)
Newsgroups: comp.os.linux,alt.os.linux

Subject: Re: How to set the path ?
Message-ID: <[email protected]>
Date: 16 Apr 92 18:21:15 GMT
References: <[email protected]> <[email protected]
>
Sender: [email protected] (Kevin Cummings)
Organization: Prime Computer R&D
Lines: 45

In article <[email protected]>, [email protected] (Robert J. McNamara) writes:
> In article <[email protected]> [email protected]
unsw.edu.au (Matthew Jackson) writes:
>
> >Now I have a few questions (of course). When I was running the install script
s,
> >I checked they were flagged as executable, and tried to run them. No go. I th
en
> >tried /INSTALL/install and it worked ! OK, I thought Linux doesn't look in th
e
> >current directory for a file before searching the path (ala DOS) (is this
> >meant to work this way ? I am used to DOS and VMS, only just got started with
> >Unix). OK (he thinks again), why don't I put the current directory in the pat
h ?
> >I typed set and got a list of the environment variables and there was one cal
led
> >PATH, so I tried set PATH=....:. and still no go.
> >
> >So can someone tell me how to change the path under Linux ?
> >
> >Thanks in advance,
>
> Yeah, that kinda screwed me up to for a bit since I was used to csh and
> dos...
> Use that last method that you mentioned, but exclude the "set" command.
> Just type "PATH=.:/bin:/usr/bin:"

I found out that you get a "PATH" by default when you login.

put the line :

echo PATH=$PATH

at the top of your .profile file to see what your path is currently set to.
You can then append or prepend to it using the $PATH varibale, or you
can completely rewrite it using the method that Robert suggests.

Don't forget to EXPORT PATH before the end of your .profile file!

=================================================================
Kevin J. Cummings Prime Computer Inc.
20 Briarwood Road 500 Old Connecticut Path
Framingham, Mass. Framingham, Mass.

Work: [email protected]
Home: [email protected]

Std. Disclaimer: "Mr. McKittrick, after careful consideration,
I've come to the conclusion that your new
defense system SUCKS..." -- War Games
=================================================================


[next article]
From: [email protected] (David.L.)

Subject: ACU.PKG
Message-ID: <[email protected]>
Date: 16 Apr 92 17:51:06 GMT
Organization: Security Pacific Automation Co., LA, CA
Lines: 7

Someone posted a uuencoded ACU.PKG, a couple days back. Supposedly containing a
'smarter' getty (for in/out links). What archiver was used to make ACU.PKG ?

I've been unable to get at the getty ;^} ..
--
----> [email protected] OR [email protected]
David L.


[next article]
From: [email protected] (Jim Winstead Jr.)

Subject: Re: make and .95a kernel rebuild
Message-ID: <[email protected]>
Date: 17 Apr 92 23:52:20 GMT
References:
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, WIBSTR
Lines: 42

In article [email protected] (Chris Waters) writes:
>Ok, I gather that GNU make won't cut it for the kernel rebuild or
>something, but I'm still curious about this bit from the Makefile:

GNU Make works fine for me....

>all
>
>Version
> @./makever.sh
> [etc.]
>
>Shouldn't there be some targets for "all"? And maybe for "Version" as
>well? Do I have a damaged set of kernel sources? (I got 'em from
>tsx-11).

They may be damaged - mine say:

all: Version Image

Version:
@./makever.sh
...etc

> Shouldn't "compress" and "uncompress" in the distribution root fs be
>links? I admit that this wouldn't save much space, but it would save a
>tad. Also, there's no link to "zcat".

Uncompress should be a link to compress, but I apparently screwed up
when tarring up the root disk and rewriting it to a freshly formatted
floppy.

> Is there a reason for the "dummy line" in /etc/passwd?

I have had problems with adding entries to /etc/passwd, and that dummy
line fixed the problems, apparently. I don't know if it's really
necessary.
--
Jim Winstead Jr. (CSci '95) | "Catch a fish!"
Harvey Mudd College | -Geddy Lee,
[email protected] | San Diego Sports Arena
Disclaimer: Mine, not theirs! | January 20, 1992


[next article]
From: [email protected] (gary a moyer)
Newsgroups: alt.os.linux,comp.os.linux

Subject: trouble with curses
Summary: bad version?
Keywords: curses
Message-ID: <[email protected]>
Date: 17 Apr 92 23:13:05 GMT
Sender: [email protected]
Lines: 12

Has anyone had trouble using curses? I am _trying_ to compile the
visual version of gnuchess (that uses cureses), however, the farthest
I get is "curses.h: parse error before line 100" and then it slowly
chokes and dies. I can't remeber where exactly I got the curses
package from (some anon ftp site). By chance is there a newer
version floating around anywhere?

using: gcc1.4, 2M ram, 4M swapfile, with straight gnu source
gave up trying to use the included makefile, so am doing
command line compiles (only two files anyhow...)

Thx. [email protected]


[next article]
From: [email protected] (Herbert M. Zinn)
Newsgroups: comp.os.linux,alt.os.linux

Subject: GCC2.1 problems(still!)
Summary: still having prollems with ld and other stuff
Keywords: linux, GCC2.1, spoo, leftist crustaceans
Message-ID: <[email protected]>
Date: 17 Apr 92 15:49:52 GMT
Sender: [email protected] (Mr. News)
Organization: the "please help change some of the bulbs on the thousand points o
f light" cabal
Lines: 17


Still having problems with GCC2.1 and the 2.1sh-A package in particular.
Any time I try to compile anything,
I get a linker error that reports "---" is not a recognized option, and then pri
nts usage; in-between the first and
second -'s is a control character. I also cannot even begin to compile the 95c+
kernel; i get innumerable parse errors
in the 95c+ specific includes (/usr/src/linux/include/...)

I know I posted this before but I haven't been able to find any help eit
her in FAQ's or elsewhere (although I
did get email from a few people with the same problem, and no solution, thanks!)


I'm eaerly awaiting a chance to learn C++ in a unix like environment. I
just hope I dont lose all sanity and
install BSD out of impatience!

๐Ÿ˜Ž ๐Ÿ˜Ž ๐Ÿ˜Ž Thanks in advance, and once again thanks to Linus for a grea
t OS (at least up til it needed GCC2.1!8-) )

Herb

:


[next article]
From: [email protected] (David Begley)

Subject: Re: User's Guide project
Message-ID:
Date: 17 Apr 92 23:57:59 GMT
References: <[email protected]>
Sender: [email protected]
Organization: University of Western Sydney
Lines: 23

[email protected] (Joseph Knapka) writes:

>1) Do you think there is a need for such a document?

Yes.. absolutely. In fact, if you want to be *really* "complete", a
User's Guide, an Installation/Configuration Guide and a
Technical/Programmer's Guide (for the source code)..

>2) Does the organization outlined seem reasonable? Should it be
>expanded/changed? I know it's a bit vague at the moment, but it can't
>be made less so without feedback...

Fine so far.. once you get everything listed written/compiled, then you
can easily add extra chapters before release as necessary..


%% david %%

--
David J. N. Begley 58:2100/[email protected], 3:712/[email protected]
University of Western Sydney, Nepean [email protected]
Student Member of the Australian Computer Society, New South Wales Branch
"Are we dead yet?" "No." "But you promised!"


[next article]
From: [email protected] (David Prall)

Subject: Setting up system
Message-ID: <[email protected]>
Date: 17 Apr 92 22:15:46 GMT
Sender: [email protected]
Organization: George Washington University
Lines: 15
Originator: [email protected]


My roommate has recently been playing with linux on his laptop and I
was thinking about setting up a drive with linux on it myself. The system I
would be setting it up on is a Gateway 2000 486/33 ISA. My main hard drive is
an ESDI drive on an Ultrastor controller. I have a second Seagate SCSI ST-02
controller connected to a Syquest 44MB removable drive. I was thinking about
setting up a syquest cartridge and a boot floppy which would be my linux setup
on this computer. Would it be possible to setup the bootdisk so it didn't
bother with the ESDI drive and only looked at the SCSI drive? Would there
be any problem using the Syquest drive even though I wouldn't be changing
disks?

Thanks,
David Prall
[email protected]


[next article]
From: [email protected] (Greenup)
Newsgroups: comp.os.linux,comp.os.minix

Subject: HELP! SHOELACE IS STRANGLING ME!!!
Message-ID: <[email protected]>
Date: 18 Apr 92 01:25:17 GMT
Sender: [email protected] (USENET News System)
Organization: Montana State University
Lines: 13


I recently used shoelace to try & make linux boot. after no success,
i want to go back. HOW DO I GET RID OF THIS BOOT THING THAT KEEPS
ASKING WHICH PARTITIAN TO BOOT?!?!?!?!?

many thanks for any assistance-
john Greenup
[email protected]

--
john Greenup | snail mail:
Internet: [email protected] | 414 Langford Hall
Bitnet : not yet | Bozeman, MT 59771


[next article]
From: [email protected] (Scott Silverstein MD)

Subject: man pages
Keywords: linux
Message-ID: <[email protected]>
Date: 18 Apr 92 00:52:35 GMT
Sender: [email protected]
Distribution: comp.os.linux
Lines: 11
Nntp-Posting-Host: grip.cis.upenn.edu

Are man pages available for any of the gnu utilities?

Also, a good editor to import to linux might be 'elle', the
emacs-like editor supplied with MWC COHERENT. Might be suitable
for the distribution as it's small.

Also, there is a PD Basic-interpreter out there called Phil's Basic,
I believe. It's a clone of MS-BASIC. Would be interesting to get
running under LINUX for the retros among us.

-- Scott.


[next article]
From: [email protected] (Drew Eckhardt)

Subject: Re: SCSI / Adapted support.... I dont get it!
Message-ID: <[email protected]>
Date: 18 Apr 92 02:08:04 GMT
References: <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 63
Nntp-Posting-Host: caesar.cs.colorado.edu

In article <[email protected]> [email protected] (Mike Me) wr
ites:
>Hi people, I relatively new to this scene and in need of some
>assistance.
>
>I have an EISA system, and I have been able to get LINUX to work.
>It took me a while... but hey! Anyway, I really want to get away from
>this floppy method of running things, but CAN NOT seem to get the
>thing to recognize my HD.
>
>I got the bootimage 95+, which supposedly has (some) support for the
>ADAPTEC 17XX controllers.
>
>I'm not quite sure what the problem is, but I thought that since the
>same controller is also in charge of the floppies I might have a
>chance
>of getting this thing to work.
>
>I end up with the folloing errors when running the bootimage 95+
>

1. Is this the .95c+ Image? Early versions are missing Adaptec
Support.

>0 host adapters detected

^^^^

The SCSI kernel is only known to work with 1542 adapters (And Ultrastor
14F's, and Seagates), but I don't know about the 17xx series. Your
SCSI host wasn't even detected - but if the 17xx series uses the
same chip as the 1540 series, it would just be a signature problem
on detection. To fix it, you need a rebuilt kernel that will detect
the 17xx board, assuming it is Adaptec 1542 compatable, and
"Do the right thing."


>detected 0 disks , 0 tapes

>hard disk I/O error
>dev 0301 block 1
>kernel panic, unable to boot.

This is covered in the FAQ - and although not your only problem (see
above) can be rectified by changing the bytes at offset 508 and
509 to 0.

>Am I doing somethiong wrong?
>
>Also, the 95+ tar file comes with some c source? Am I to compile this
>first? How can I on a floppy system?
>

If it will work on your system out of the box, the SCSI binary
(vmunix.Z on headrest.woz.colorado.edu, something else
on other sites), will work. Source is provided to the entire
SCSI package, like the rest of the Linux kernel.


>Needless to say I'm confused, and could not find any help that
>pertained to my problem in the FAQ pages.

There is a separate FAQ for SCSI, I should probably get everything
organized and get it in the "official" FAQ.


[next article]
From: [email protected] (Drew Eckhardt)

Subject: Re: Setting up system
Message-ID: <[email protected]>
Date: 18 Apr 92 02:13:35 GMT
References: <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 19
Nntp-Posting-Host: caesar.cs.colorado.edu

In article <[email protected]> [email protected] (David Prall)
writes:
>
> My roommate has recently been playing with linux on his laptop and I
>was thinking about setting up a drive with linux on it myself. The system I
>would be setting it up on is a Gateway 2000 486/33 ISA. My main hard drive is
>an ESDI drive on an Ultrastor controller. I have a second Seagate SCSI ST-02
>controller connected to a Syquest 44MB removable drive. I was thinking about
>setting up a syquest cartridge and a boot floppy which would be my linux setup
>on this computer. Would it be possible to setup the bootdisk so it didn't
>bother with the ESDI drive and only looked at the SCSI drive? Would there

Get the SCSI boot image, and change the bytes at offset 0508 and 0509 to
00 00. Linux will completely ignore the ESDI disk, and other than
it showing up when you run fdisk, nothing will happen.

>be any problem using the Syquest drive even though I wouldn't be changing
>disks?

Nope. Don't change the cartridge after booting though.


[next article]
From: [email protected] (M. Saggaf)

Subject: telecomm in Linux, users
Keywords: kermit, zmodem, passwd, user
Message-ID: <[email protected]>
Date: 18 Apr 92 02:25:39 GMT
Sender: [email protected] (News system)
Distribution: comp.os.linux
Organization: Massachvsetts Institvte of Technology
Lines: 14
Nntp-Posting-Host: erl.mit.edu

I have a couple of questions about Linux. First, kermit does not work.
It runs fine at first but it cannot even "connect". It times out after
issuing the command "connect" with the message "cannot get char:
EAGAIN", or something like that. Does that happen to everybody? I'm not
using kermit5A by the way (that one terminates with a segmentation
fault). xc runs fine. I have a hayes-compatible modem at com1 (ttys1).
Also, zmodem (rz) does not work. It also times out. Any suggestions
would be greatly apprecuated. I've been using mskermit and zmodem
under DOS for quite some time with no problems.

Second, how can I setup a user account other than root? -- i.e. how to
setup the /etc/passwd file, .. etc? Thank you.

-- [email protected]


[next article]
From: [email protected] (Rwo-Hsi Wang)

Subject: Mutant bison?
Message-ID:
Date: 18 Apr 92 05:24:22 GMT
Organization: CS Dept, University of Texas at Austin
Lines: 16
NNTP-Posting-Host: tokio.cs.utexas.edu

I used bison (from tsx-11) to compile the following input

%token N
%%
s : N
;

and got the following message:

bison : memory exhausted

on my 486-33 with 16 MB running 0.95c+. Am I the only one having this
mutant bison? It has no problem compiling other big .y files, though.

--
Rwo-Hsi


[next article]
From: [email protected] (Timo Jaakko Sillanp{{)

Subject: HELP! Problem with rootimage. (UPDATE 2)
Message-ID:
Date: 16 Apr 92 14:52:27 GMT
Sender: [email protected] (Usenet pseudouser id)
Distribution: comp
Organization: Helsinki University of Technology, Finland
Lines: 17
Nntp-Posting-Host: judge-dredd.cs.hut.fi

I still haven't got Linux to load rootimage diskette if I'm
using 1.44Mb diskettes. (I have tried on four different
machines: 3*486/33 + 1*386SX/20.)

However, yesterday I tried Linux on two machines with 1.2Mb
5.25" floppy drives and everything worked fine. Both were
386SX machines with 2Mb RAM and Phoenix BIOS.

Maby rawrite doens't work correctly with 1.44Mb floppies
on my machine? I will try to switch floppy drive cables
and boot from 5.25" drive. If it works, fine. But it is
only temporary solution since OS/2 2.0 wants to install
from 3.5" floppy drive. (Well, maby 5.25" diskettes for
OS/2 exist, but I don't want to use them.)

TS.
[email protected]


[next article]
From: [email protected] (Craig Burley)

Subject: Re: GCC2.1 (Wonderful !) (dynamic linking)
Message-ID:
Date: 18 Apr 92 07:33:58 GMT
References: <[email protected]>

<[email protected]>
Sender: [email protected]
Organization: Free Software Foundation 545 Tech Square Cambridge, MA 02139
Lines: 51
In-reply-to: [email protected]'s message of 17 Apr 92 21:01:36 GMT

In article <[email protected]> [email protected] (John
Plevyak) writes:

I agree with [email protected] (Craig Burley), but I would
like to clarify and sum up a bit.

[Obviously a person of fairly high intelligence... :-]

Craig writes:

> Either I misunderstand you or this is wrong. ... PRIMOS generally
> had pointers (especially unsnapped ones) in read/write segments (pointed to
> by the Link Base, LB%) along with ECBs (procedure Entry Control Blocks) and
> such.

I stand corrected. In general, however, the problem is that if the functions
are
not statically bound to offsets at the call site, dynamic linking either requ
ires
(1) another level of indirection or (2) a fixup at the call site. Solution (1
)
can be made more efficient if supported in hardware, and solution
(2) requires impure code. In general, both these solutions have real perform
ance
drawbacks. In case (2) pointers to functions have to be handled as well.

Yes indeed. PRIMOS (and many other OSes) used (2). But I don't know for
sure how serious the performance drawbacks are; again, weighing shareability
and revisability against run-time and hardware-design issues is difficult,
because these issues are very deep and dense (they go all the way down to
issues like "how should the TLB work" and "how should cache work" and all the
way up to "how should programmers work" :-).

Segmentation is related issue. Segments can provide the extra level of
indirection for dynamic linking, handles for memory management etc, not
to mention other advantages. The segment sizes are the issue ('286 and '386)
.
Too few large segments are not good for small blocks or functions, too many
small ones and you can't allocate large arrays. The good and the bad... life
.

Yes, I don't think segments are really useful in today's world unless they are
both numerous and large, as appears to be the case with the 486 (and
apparently not with the 386, certainly not with the 286).

The model for Unix has been flat memory, sharing by pages and basically
KISS. I'm all for experimentation though. My feeling is the proof is in the
pudding. Build it. If it works, everyone will use it.

Absolutely. We have much work to do in the flat-model world just to get
software up and running; for now, I think we should reserve the issues of
segmentations to "what-if" just to keep our minds open to what might happen
in the future and how we might improve performance down the road. Also to
keep us from making decisions that'll seem foolish sometime in the future.
--

James Craig Burley, Software Craftsperson [email protected]
Member of the League for Programming Freedom (LPF)


[next article]
From: [email protected] (Thomas Wuensche)

Subject: cslip/ka9q help needed
Message-ID:
Date: 18 Apr 92 09:15:16 GMT
Sender: [email protected] (Mr. News)
Reply-To: [email protected]
Organization: Leibniz-Rechenzentrum, Muenchen (Germany)
Lines: 12

It does not belong there, but using linux took me to the question.
So probably somebody can help.

I want to connect my laptop running linux to my DECStation at work.
I'd like to use slip for that purpose. KA9Q does it on the linux
side, but what do I need on the DECStation side. I tried the
cslip packet, but with no success. The best thing I ended up
was a "SYN sent" message from KA9Q.

Any help appreciated

Thomas


[next article]
From: [email protected] (Robert J. McNamara)

Subject: pfdisk
Message-ID: <[email protected]>
Date: 18 Apr 92 10:20:31 GMT
Reply-To: [email protected] (Robert J. McNamara)
Organization: Lawrence Berkeley Laboratory
Lines: 36
NNTP-Posting-Host: 128.3.252.177

Maybe someone who knows "pfdisk" can help me out here...
I'm trying to create a partition for linux in the location
of the second logical partition on the extended partition. (Linux
recognizes this as /dev/hda6). Problem is, pfdisk can only
set partitions 1 to 4 -- not up to 6. Perhaps I'm using it wrong,
in which case I'd appreciate if someone would explain
how pfdisk is supposed to function, and what, if any, differences
it has from the msdos fdisk program.

The REASON for all of this is: I have a linux setup on the above
partition (that I created with dos's fdisk) and I've got my
filesystem installed on it. I've been TRYING to set up
Shoelace, (following the summarized directions that were posted
here yesterday) and the winiboot program comes up fine and offers
my selections to me (1 and 2 are the only valid selections that it
offers me since 3 & 4 are zeroed out). If I select '1', DOS boots (as
it's supposed to). If I select '2', the system locks with a message
"NO ROM BASIC
SYSTEM HALTED"
So...after trying several things, I came up with the thought that
perhaps the partition that I was attempting to
"laceup /dev/hda6 wini" needed to be a Linux (Minix 1.5+?)
compatible partition. If this ISN'T the problem, could someone
tell me what I COULD be doing wrong? Note that I followed the
directions posted here on the 17th exactly. The "boot" line
of the config file is set to "/lboot" (my patched boot image
that starts Linux just fine from floppy and goes on to the file
system on the hard drive).

Thanks...

+-=-=-=-=-=-=-=-=-=-=-=-+-=-=-=-=-=-=-=-+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
| Rob McNamara | Happy! Happy! | Lankhmar BBS |
| [email protected] | Joy! Joy! | 1-916-757-6391 |
| Lawrence Berkeley Lab | | 1200/2400/9600/14.4k HST/v.32bis |
+-=-=-=-=-=-=-=-=-=-=-=-+-=-=-=-=-=-=-=-+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+


[next article]
From: [email protected] (Tor Lillqvist)

Subject: Linux vs. 386BSD vs. Mach+HURD
Message-ID:
Date: 18 Apr 92 19:21:37 GMT
Sender: [email protected]
Distribution: comp
Organization: Technical Research Centre of Finland, Laboratory for Information
Processing (VTT/TIK)
Lines: 15

A question to the Linux users and developers: Do you intend to keep on
using Linux, and developing more and more features for it, even when
386BSD or the GNU OS (Mach+HURD) evetually are as stable as Linux is
now? Is the intention to keep Linux "small and simple," and not
implement many non-POSIX features? Presumably Mach+HURD will demand
much more RAM to run comfortably, for instance?

I guess at least Berkeley sockets and TCP/IP are already in the
process of being ported to Linux, and somebody is probably porting the
BSD file system, etc. What will then be the difference between Linux
and 386BSD? Linux' innermost kernel was written from scratch directly
for the 386, does this mean Linux will always be noticeably faster?
--
Tor Lillqvist,
working, but not speaking, for the Technical Research Centre of Finland


[next article]
From: [email protected] (Tor Lillqvist)

Subject: Re: Graphics and IPC questions...
Message-ID:
Date: 18 Apr 92 19:34:41 GMT
References: <[email protected]
du>
Sender: [email protected]
Organization: Technical Research Centre of Finland, Laboratory for Information
Processing (VTT/TIK)
Lines: 17
In-reply-to: [email protected]'s message of 17 Apr 92 15:08:59 GMT

In article <[email protected]> [email protected] (Brian Choj
nowski) writes:
I plan to add a direct to screen library. Basically I envision
creating a terminal type that is a superset of vt100. Nothing as fancy or
system hog like X. Just a characterset based graphix system. So while we are
at it, if anyone has some info somewhere on how to write assembly stuff
under linux, I can start porting some of my dos-mode screen routines.

Ugh. Isn't that a step backward? Why create a completely new,
nonstandard, *local-only*, graphics library? I don't think there is
much chance that people will write programs for Linux that use your
direct-to-screen library. To put it bluntly, if you want to fiddle
with direct hardware access assembler routines, stick to DOS. Even if
you don't need networked graphics, I'm certain a significant fraction
of the Linux community wants it.
--
Tor Lillqvist,
working, but not speaking, for the Technical Research Centre of Finland


[next article]
From: [email protected] (Alex R.N. Wetmore)

Subject: Re: HELP! Problem with rootimage. (UPDATE 2)
Message-ID:
Date: 18 Apr 92 12:21:46 GMT
Distribution: comp
Organization: Freshman, Math/Computer Science, Carnegie Mellon, Pittsburgh, PA
Lines: 16
In-Reply-To:

Excerpts from netnews.comp.os.linux: 16-Apr-92 HELP! Problem with
rootimag.. by Timo J. Sillanp{{@niksul
> Maby rawrite doens't work correctly with 1.44Mb floppies
> on my machine? I will try to switch floppy drive cables
> and boot from 5.25" drive. If it works, fine. But it is
> only temporary solution since OS/2 2.0 wants to install
> from 3.5" floppy drive. (Well, maby 5.25" diskettes for
> OS/2 exist, but I don't want to use them.)

Try getting a copy of boot_b.exe. It will write a new boot sector to a
floppy disk that will boot your B drive, so you could have a 3.5 in A
that has been boot_b'ed, and then the linux boot disks in the 5.25
drive. It think that it can be found in either the sysutl or dskutl
directories on the msdos tree of simtel (or wuarchive).

alex


[next article]
From: [email protected] (Jyrki Kuoppala)

Subject: A note on beginner instructions for Linux
Message-ID: <[email protected]>
Date: 16 Apr 92 16:50:02 GMT
Sender: [email protected] (Usenet pseudouser id)
Reply-To: [email protected] (Jyrki Kuoppala)
Organization: Helsinki University of Technology, Finland
Lines: 16
Nntp-Posting-Host: lusmu.cs.hut.fi

I'm packing Linux on a Unix box to diskettes to be transported to a
DOS machine on which I'd like to test Linux on. I'm using the stuff
from /pub/OS/Linux on nic.funet.fi, and it seems to be easy enough to
pack the things to the diskettes.

However, I didn't find anywhere documented what binaries and files are
on the root image diskette. I think it would be nice to list what
stuff is on the root image so people would know what other stuff they
need to install at the beginning. It would also be nice to have a
list of programs which are needed for which things (eg. "for
rebuilding the kernel you'll need gcc*.tar.Z, kernel***.tar.Z, etc.).

Sorry if I've just overlooked something, I'm in a bit of a hurry to
copy the disks.

//Jyrki


[next article]
From: [email protected] (Michael Pereckas)

Subject: Patch to add Dvorak keyboard
Keywords: keyboard console kernel patch
Message-ID:
Date: 18 Apr 92 20:26:45 GMT
Sender: [email protected] (News)
Organization: University of Illinois at Urbana
Lines: 112


This patch for linux/kernel/chr_drv/keyboard.S does two things: it
causes the ./del key on the keypad to produce a period, instead of a
comma, and it adds a Dvorak keyboard.

The first change will probably appeal to US users. Others may prefer
the comma. If there is a lot of difference of opinion on this, maybe
num_table should be moved into the national keyboard definitions. The
second change is great if you, like me, like the Dvorak keyboard
layout. Unfortunatly, the only way to change keyboards is to reboot
with a different kernel, so the Dvorak keyboard is a problem is more
than one person use the machine and they don't all know Dvorak. (this
only effects the console, serial port connections are uneffected.)

I post this on the off chance that someone is interested. If you
choose to use this, remember that although it seems to work fine for
me, this is an example of "programming by meta-w", that is, I copied
the US keyboard definition (using the emacs command meta-w) and

modified it, without really understanding it.


This patch is for linux/kernel/chr_drv/keyboard.S
It works for all the 0.95* versions, I think (!)
********** CUT HERE **********
*** keyboard.S.ori Wed Apr 8 16:57:58 1992
--- keyboard.S Wed Apr 8 17:03:10 1992
***************
*** 18,23 ****
--- 18,24 ----
* KBD_FR for Frech keyboard
* KBD_UK for British extended keyboard
* KBD_DK for Danish keyboard
+ * KBD_DVORAK for Dvorak (US) keyboard
*/

.text
***************
*** 251,257 ****
.ascii "789-456+1230."
#else
num_table:
! .ascii "789-456+1230,"
#endif
cur_table:
.ascii "HA5-DGC+YB623"
--- 252,258 ----
.ascii "789-456+1230."
#else
num_table:
! .ascii "789-456+1230."
#endif
cur_table:
.ascii "HA5-DGC+YB623"
***************
*** 611,616 ****
--- 612,667 ----
.byte 0,0,0,0,0 /* 4A-4E */
.byte 0,0,0,0,0,0,0 /* 4F-55 */
.ascii "\\"
+ .fill 10,1,0
+

+ #elif defined(KBD_DVORAK)
+
+ key_map:
+ .byte 0,27
+ .ascii "1234567890\\="
+ .byte 127,9
+ .ascii "',.pyfgcrl/]"
+ .byte 13,0
+ .ascii "aoeuidhtns-"
+ .byte '`,0
+ .ascii "[;qjkxbmwvz"
+ .byte 0,'*,0,32 /* 36-39 */
+ .fill 16,1,0 /* 3A-49 */
+ .byte '-,0,0,0,'+ /* 4A-4E */
+ .byte 0,0,0,0,0,0,0 /* 4F-55 */
+ .byte '<
+ .fill 10,1,0
+
+ shift_map:
+ .byte 0,27
+ .ascii "[email protected]#$%^&*()|+"
+ .byte 127,9
+ .ascii "\"<>PYFGCRL?}"
+ .byte 13,0
+ .ascii "AOEUIDHTNS_"
+ .byte '~,0
+ .ascii "{:QJKXBMWVZ"
+ .byte 0,'*,0,32 /* 36-39 */
+ .fill 16,1,0 /* 3A-49 */
+ .byte '-,0,0,0,'+ /* 4A-4E */
+ .byte 0,0,0,0,0,0,0 /* 4F-55 */
+ .byte '>
+ .fill 10,1,0
+
+ alt_map:
+ .byte 0,0
+ .ascii "\[email protected]\0$\0\0{[]}\\\0"
+ .byte 0,0
+ .byte 0,0,0,0,0,0,0,0,0,0,0
+ .byte '~,13,0
+ .byte 0,0,0,0,0,0,0,0,0,0,0
+ .byte 0,0
+ .byte 0,0,0,0,0,0,0,0,0,0,0
+ .byte 0,0,0,0 /* 36-39 */
+ .fill 16,1,0 /* 3A-49 */
+ .byte 0,0,0,0,0 /* 4A-4E */
+ .byte 0,0,0,0,0,0,0 /* 4F-55 */
+ .byte '|
.fill 10,1,0

#else


[next article]
From: UPP2[email protected] (Guido Kueppers)

Subject: Re: IDE-drive performance with linux
Message-ID: <[email protected]>
Date: 18 Apr 92 16:31:28 GMT
Sender: [email protected]
Organization: RHRZ-Uni-Bonn
Lines: 60

Meanwhile I have tried to come up with some figures for the IDE drive
performance problem. What I did was copying /dev/kmem (4MB) to disk and
in a second step to copy the resulting file (kmem.dump) from one
partition to another. Running this test on the MFM drive as well as on
the IDE drive showed that the former was definitely faster.
On both setups I used the same kernel, compiled from the unmodified
0.95c+ source found on tsx-11.mit.edu. There were no other
processes running except for /etc/update, which I had no way to kill on
the setup with the MFM drive. This is what I've got:

This is a test with the Seagate ST251 MFM drive and WD1006V controller:
=======
bash#df

inodes inodes inodes blocks blocks blocks mount
devices total used free total used free point
-------------------------------------------------------------------
/dev/hda1 6525 125 6400 199575 1373 18202 /
dev/hda2 7395 3 7392 22185 239 21946 mnt/
bash# sync
bash# time cp /dev/kmem /kmem.dump <--- write 4MB from memory to disk
0.17u 55.08s 1:00.25 8pf
^^^^^^^
bash# ls -l /kmem.dump
-rw------- 1 root root 4194304 Apr 17 21:06 /kmem.dump
bash# sync
bash# time cp /kmem.dump /mnt/ <--- copy a 4MB file from hda1 to hda2
0.18u 10.12s 2:35.36 8pf
^^^^^^^

bash#

This is the same test performed on a Rodime 3259A IDE drive:
=======

/ # df

inodes inodes inodes blocks blocks blocks mount
device total used free total used free point
------------------------------------------------------------------
/dev/hda2 18504 163 18341 55512 14094 41418 /
/dev/hda3 18504 706 17798 55512 19165 36347 /usr
/dev/hda4 18576 1007 17569 55728 18105 37623 /usr/local
/ # sync
/ # time cp /dev/kmem /kmem.dump
0.19u 62.66s 1:28.08 8pf <--- About 50% more than with the MFM drive

/ # ls -l /kmem.dump
-rw------ 1 root root 4194304 Apr 17 21:30 /kmem.dump
/ # sync
/ # time cp /kmem.dump /usr/local/bin/
0.23u 14.18s 2:58.79 8pf <--- about 30 secs more than with the MFM
drive

I don't know what this sort of testing is worth after all, but it seems to
substantiate my doubts about IDE drive performance. Perhaps the problem
lies with the particular drive I'm using. Perhaps it has to do with the IDE
drive's internal buffering scheme. Someone willing to comment on this?

Guido


[next article]
From: [email protected] (Sotiris C. Vassilopoulos)

Subject: Looking for 'expr' and 'tr' for 0.95c+
Message-ID: <[email protected]>
Date: 18 Apr 92 13:34:36 GMT
Sender: [email protected]
Organization: University of Virginia
Lines: 10

Hi all,

Can somebody out there with a working 'expr' and 'tr' e-mail me a
copy? I am using 0.95c+ and g++ complains that it cannot find 'expr'.
Some other programs, I tried to install, are missing 'tr'.

Just the executables (using either static or shared libraries) would be fine.

Tnanks in advance,
- Sotiris


[next article]
From: [email protected] (Brian Chojnowski)

Subject: Re: Graphics and IPC questions...
Message-ID: <[email protected]>
Date: 18 Apr 92 21:33:13 GMT
References: <[email protected]
du>
Sender: [email protected] (News)
Organization: University of Illinois at Urbana
Lines: 15


>Ugh. Isn't that a step backward? Why create a completely new,
>nonstandard, *local-only*, graphics library? I don't think there is
>much chance that people will write programs for Linux that use your
>direct-to-screen library. To put it bluntly, if you want to fiddle
>with direct hardware access assembler routines, stick to DOS. Even if
>you don't need networked graphics, I'm certain a significant fraction
>of the Linux community wants it.

I dont intend on making a local only system. What I do want to avoid is the
system requirements of X. I dont think there are any X telnet programs for
the 286. I already have an application, and Unix would be a good multi-user
platform for it, but it requires characterset based graphics.

Brian Chojnowski


[next article]
From: [email protected] (Josh Yelon)

Subject: Re: Patch to add Dvorak keyboard
Keywords: keyboard console kernel patch
Message-ID: <[email protected]>
Date: 18 Apr 92 22:01:25 GMT
References:
Sender: [email protected] (News Database (admin-Mike Schwager))
Organization: University of Illinois, Dept. of Comp. Sci., Urbana, IL
Lines: 7

Just a suggestion... perhaps it might make sense, with all the
keyboard kernel patches going around, to make a system call that
dynamically modifies the keyboard xlation table on the fly. That
way, somebody could write a nice user-friendly utility that edits
and loads up keyboard tables...

- Josh


[next article]
From: [email protected] (Scott Silverstein MD)

Subject: Re: IDE-drive performance with linux
Message-ID: <[email protected]>
Date: 18 Apr 92 23:01:26 GMT
References: <[email protected]>
Sender: [email protected]
Lines: 17
Nntp-Posting-Host: grip.cis.upenn.edu

>
>I don't know what this sort of testing is worth after all, but it seems to
>substantiate my doubts about IDE drive performance. Perhaps the problem
>lies with the particular drive I'm using. Perhaps it has to do with the IDE
>drive's internal buffering scheme. Someone willing to comment on this?
>
>Guido


Perhaps the slowness involves the drive's own internal 'translation mode'
operations? I run MWC COHERENT with a Seagate IDE drive and it works
very speedily. I will get LINUX 0.95c+ up and running as soon as I find
the time and energy and run similar tests.

Are you using the drive in its native or translation mode?

-- Scott.


[next article]
From: [email protected] (Charles Hedrick)

Subject: timezone support, date, CMOS clock support: archive sites please note
Message-ID:
Date: 18 Apr 92 23:22:13 GMT
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 151

I've just put some stuff in /pub/linux on cs.rutgers.edu. I'd appreciate
it if the archive sites would install in an appropriate place:

timebin.tar.Z - date, clock, /usr/lib/zoneinfo
timesrc.tar.Z - source for clock, zic, timezone files
time.doc - documentation for the above (this is also included in timebin)

This supercedes the date stuff I put out a month or two ago for gcc
1.4. Probably it should be integrated into the gcc 2.1 distribution
as part of the libraries.

These files are intended to be used with the gcc 2.1 libc. That libc
has the necessary library routines to handle time zones, but without
the files included here, the code won't do anything. I believe
everything except clock is from Gnu. I wrote clock.

I have included a new copy of date, built from the Gnu shellutils 1.6.
(I didn't include it in timesrc, since it's the straight Gnu code.)
There's nothing special about this date, but you need a date that's
built with the 2.1 libc, and this is.

I have included a new program, clock. This will set or read the
hardware (CMOS) clock. Currently there is no way to set the CMOS
clock. I had patched the kernel so that setting the Unix time also
set the CMOS clock. However Linus didn't want to take the patch.
Indeed he wants to get rid of the code currently in the kernel that
initializes the Unix clock from the CMOS clock. He feels (correctly,
I think) that the kernel shouldn't need code that duplicates the time
zone stuff in libc. The idea is that you put a call to clock in
/etc/rc to initialize the Unix clock from the CMOS clock. You can
also call it with another argument to set the CMOS clock from the Unix
clock.

Note that clock uses /dev/port to play with the CMOS clock. Since
/dev/port may not be writeable by users, clock should probably be made
setuid. It contains its own check to make sure that only root can
change the time (though it doesn't protect the CMOS clock -- this is
consistent with DOS behavior, under which there is no protection. If
you want only root to be able to change the CMOS clock, don't install
this setuid.) I considered putting the CMOS stuff into date. However
this seems a violation of the Unix design philosophy. It would have
added several extra options to date, and I couldn't figure out any way
of making it very clean.

Here's time.doc, which explains stuff in more detail:

-----------------------------------------------------
Setting up time and time zones

There are several things involved in getting time right under Linux:

- /usr/lib/zoneinfo contains files that define what time zone you
are in. If they are missing, no time zone calculations
are done, i.e. your internal clock is assumed to be on
local time rather than the Unix standard of GMT. The only
file that you absolutely need is /usr/lib/zoneinfo/localtime, but I
recommend also having /usr/lib/zoneinfo/posixrules. Posixrules
is typically a copy of or link to localtime. Localtime defines
your default zone. Posixrules is needed to interpret the TZ
variable, which is used if you want to specify a zone other than
the default.

- the "date" command can be used to set or display the date/time.
Note however that it does not set the hardware clock, so
next time you reboot, you'll be back to the old time.
I recommend that after changing the time with "date", you
use "clock -w" or "clock -u -w" to update the hardware clock
as well. (See below.)

- the "clock" command can be used to set or display the date/time
in the hardware (CMOS) clock. Typically your /etc/rc script
will contain
clock -s
which will cause the Unix date/time to be initialized from the
CMOS clock when you boot. If your CMOS clock is set to GMT
(which is what I recommend) the correct command is
clock -u -s

The binary time distribution should be untarred under /usr. It
contains lib/zoneinfo, bin/date, bin/clock, and doc/time.doc (this
file). Once you've installed these files, you'll want to do four
things:

1) set /usr/lib/zoneinfo/localtime and /usr/lib/zoneinfo/posixrules.
You should copy the file for your time zone. E.g. if you are in the
U.S. Eastern time zone, do

cd /usr/lib/zoneinfo
cp US/Eastern localtime
ln localtime posixrules

Localtime defines the local time zone. Posixrules defines the zone to
be used to interpret the TZ environment variable. Since it's far more
convenient simply to use the right time zone file, nothing more will
be said here about how the TZ variable is used. Unless you intend to
use TZ, you can ignore the next paragraph.

If you want exact POSIX behavior, posixrules should be a copy of or
link to one of the U.S. time zone files. (For non-U.S. daylight
rules, the TZ variable defines the daylight transition rules.)
However it may make more sense practically for it to be the same as
localtime, as shown in the instructions above.

2) Once you've set up localtime and posixrules, you can remove the
rest of the files in /usr/lib/zoneinfo, if you're sure you'll never
want to operate in any other time zone. Or you can keep just the few
time zones that you might need.

3) Put the correct "clock" command into /etc/rc. Which command to use
depends upon whether you want your hardware clock to keep local time
or GMT. I recommend using GMT, since that will allow daylight savings
transitions to be completely automatic. However the same clock is
used by DOS, and some people don't like the time in DOS being GMT. I
use Unix-compatible software under DOS. It uses the TZ environment
variable to do time zone conversion. Thus I prefer the clock being
GMT even under DOS. But some people may not like that. Anyway, if
your hardware clock is set to the local time, put the line

clock -s

in /etc/rc. This will set the Unix time from your hardware clock,
doing the necessary time conversion. If your hardware clock is set
to GMT, then you'll need the -u option:

clock -u -s

4) Now make sure that your hardware clock is set correctly. Try
"clock" with no arguments. It will print the current setting of the
hardware clock. Make sure it is right, and that it is either local or
GMT, as you decided. (If the hardware clock is supposed to be GMT,
you can use "clock -u". This will convert from GMT to local and
display it.) To set the clock, first use the "date" command to get
the date right in Unix. Then use "clock -w" to set the hardware
clock. Note that "clock -w" will set the hardware clock to the local
time, and "clock -u -w" will set it to GMT. Verify with "clock" that
the hardware clock is as you want it.

From now on, the time should be right. If your hardware clock loses
or gains time, you can update it at a future date by the same
procedure just described: first get the Unix time right using "date"
and then use "clock -w" or "clock -u -w" to set the hardware clock.

If your hardware clock is set using local time, make sure to reset it
when daylight time changes. If you're running Unix when daylight time
changes, the Unix time will adjust automatically. In that case, all
you need is "clock -w" to update the hardware clock. If you aren't
running Unix during the transition, then your time will be an hour off
the next time you boot. In that case, set the correct Unix time using
"date", and then use "clock -w" to update the hardware clock. If your
hardware clock is set using GMT time, none of this is necessary --
daylight time transitions will happen automatically.


[next article]
From: [email protected] (Risto Kankkunen)
Newsgroups: comp.os.linux,alt.os.linux

Subject: Re: all kinds of questions (autoparking)
Message-ID: <[email protected]>
Date: 18 Apr 92 23:29:36 GMT
References: <[email protected]> <[email protected]
rado.edu> <[email protected]>
Organization: University of Helsinki, Department of Computer Science
Lines: 12

In [email protected] (Kevin Cummings) writes:
> So unless you are shutting down your system to reboot DOS (or something
> else), your disk heads will not be parked by LINUX. Maybe this is the
> cause of some of the trashed file systems some people are experiencing?

But isn't head parking suggested only before moving the computer or the
disk itself? Although there is a greater possibility for disk damage
when the heads stay above the disk on power-down, it isn't a problem in
practise on everyday desk-top use.

--
no sig today


[next article]
From: [email protected] (Charles Hedrick)

Subject: Re: IDE-drive performance with linux
Message-ID:
Date: 18 Apr 92 23:57:36 GMT
References: <[email protected]>
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 26

For what it's worth, I just tried the same test you did,

cp /dev/kmem /mnt/kmem.dump

It took me about 40 sec (note that it takes about 5 sec beyond when
the cp returns, because of buffering). I ended up with a slightly
larger file than you, 4.5M. Copying it to another partition took
1 min, 20 sec. I also tried

cp /mnt/kmem.dump /dev/null

It took 12 sec. That's certainly not up to a Sparcstation 2, but for
a 386sx on an AT bus, I think it's fine. This is with the new Connor
510MB IDE drive, which I believe has very fast seek time. I would
expect that copying partitions would be primarily an issue of seek,
and how well the OS optimizes things, not transfer rate. Indeed
probably only the read time (cp into /dev/null) is dominated by
transfer rate.

Unfortunately I don't have an MFM drive to compare with. My machine
can't handle more than one drive at a time. I also can't tell you
much about the hardware. I depend upon a local PC dealer to do
sensible things with controllers, etc. I only claim to know about
software. (I realize that a 500MB IDE disk is sort of wierd. I
intentionally avoided SCSI because I wanted to be able to play with
OS's, and generally IDE is more likely to be supported.)


[next article]
From: [email protected] (david nugent)

Subject: Re: Need help with swapping
Keywords: linux, swap
Message-ID:
Date: 18 Apr 92 11:10:44 GMT
References: <[email protected]>
Reply-To: [email protected]
Organization: Unique Computing Pty Ltd, Melbourne, Australia
Lines: 19

[email protected] (Hongjiu Lu) writes:

> |> into a library using "ar rc libfoo.a foo1.o foo2.o ..." I was not
> |> able to link these libraries to my other sources. ld complained
> |> about undeclared symbols in the text-segment but nm showed that all
>
> Use ranlib before linking.
>
> Should that be in FAQ? I thought everybody knows it.

In all fairness, the majority of Sys V/386's have ranlib built into ar, so
ranlib doesn't exist on those systems.


..............................................................................
david nugent Public Access Usenet "Only Nixon can go to China"
[email protected] +61-3-792-3507 - ancient Vulcan proverb
3:632/[email protected], 58:4100/[email protected], 199:4242/[email protected], 33:300/[email protected]
PO Box 260, Endeavour Hills, Victoria, Australia, 3802.


[next article]
From: [email protected] (David A. Sinclair)
Newsgroups: comp.os.os2.misc,comp.os.linux

Subject: Any comments on Quantum 240A ??
Summary: Can I use this drive w/ Linux and OS/2 2.0??
Keywords: OS/2, Linux, 240A, Quantum
Message-ID: <[email protected]>
Date: 19 Apr 92 00:27:24 GMT
Sender: [email protected] (USENET News System)
Organization: Brandeis University
Lines: 20

Hello, world.

I need to buy anew hard drive for my system so I can install Linux and
OS/2 2.0. I am considering the Quantum 240A, (240 meg IDE) and would
like to hear from anyone who is either using this drive with either
OS, or has tried and failed to do so. Anyone have any comments about
using this drive with OS/2 GA and / or Linux? Anything to watch out
for?

Thanks for any info you can provide!!

Cheers,



--
-----------------------------------------------------------------
David A. Sinclair - "Chester" | Bobby Fischer doesn't *want*
[email protected] | you to know where he is.
-----------------------------------------------------------------


[next article]
From: [email protected] (Geoffrey Furnish)
Newsgroups: comp.os.linux,alt.os.linux

Subject: zip dies.
Message-ID: <[email protected]>
Date: 18 Apr 92 08:30:40 GMT
Sender: [email protected]
Followup-To: comp.os.linux
Organization: The University of Texas at Austin, Austin TX
Lines: 9

Hi folks, zip is dieing on me. Generates a general protection fault
and lots of interesting uninteligible dump info. My line looked like:

zip -r zipfile dir/* -x dir/subdir1 dir/subdir2

Has anyone encountered this and/or any ideas what to do instead?
It works with the zip which is on a SUN I use.

Geoff Furnish


[next article]
From: [email protected] (Jim Burns)

Subject: Re: GNU m4 and chess work, awk, mawk and sort don't.
Message-ID: <[email protected]>
Date: 19 Apr 92 04:10:37 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 12

in article <[email protected]>, [email protected]
matik.rwth-aachen.de (Michael Haardt) says:

> awk -F @ '{ print $2 }'

> [email protected] gives [email protected] as output, not only bar as expected. Both GNU awk

Does it work w/o the space between -F and @ ?
--
BURNS,JIM (returned student)
Georgia Institute of Technology, 30178 Georgia Tech Station,
Atlanta Georgia, 30332 | Internet: [email protected]
uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt0178a


[next article]
From: [email protected] (Michael Pereckas)

Subject: ispell works
Summary: ispell compiles easily
Keywords: ispell emacs
Message-ID:
Date: 19 Apr 92 04:24:48 GMT
Sender: [email protected] (News)
Organization: University of Illinois at Urbana
Lines: 9

I have compiled ispell 2.0.02 with no problems using gcc 2.1.
Just look through the makefile to see that all the directories are
right, and set the compiler flag for sysv-type systems, as documented in
the makefile.

Ispell is a spelling checker that can be used with GNU Emacs. The
distribution contains elisp source for a friendly interface.
It can be found in the pub/gnu directory on prep.ai.mit.edu and on that
site's many mirrors.


[next article]
From: [email protected] (Carol Perkins)

Subject: MFM and IDE drives
Summary: IS anyone running an MFM and an IDE drive together?
Keywords: MFM, IDE
Message-ID: <[email protected]>
Date: 19 Apr 92 05:19:49 GMT
Organization: Northwestern University, Evanston, Illinois
Lines: 10

I would like to know if anyone has been successful in getting an MFM and an
IDE drive to coexist. I want to add an IDE to my system and continue to use
an ST251-1 as the boot drive. The IDE is an ST3144a with the ST08A
controller. The MFM controller is an NCC. I understand that it will probably
be necessary to change the controller or host adapter I/O port address. How
do you do that?

Any help will be appreciated.

[email protected]


[next article]
From: [email protected] (Kayvan Sylvan)

Subject: Re: ESDI drives
Message-ID:
Date: 19 Apr 92 07:50:30 GMT
References:
Sender: [email protected] (Kayvan Sylvan)
Organization: Sylvan Associates
Lines: 16
In-Reply-To: [email protected]'s message of 16 Apr 92 23:03:30 GMT

In article [email protected] (Demian A. J
ohnston) writes:

I think I might no your problem.. I had problems when I made a very
large partition.. You might want to try and smaller partition size. I
think Linux has a 64Mb limit... My partitions are on the order of
40,000 blocks...

Is this true (partition sizes limited to 64MB)??? If that's the case
it would be a shame. What about those of us who have larger drives?

---Kayvan
--
| Kayvan Sylvan, Sylvan Associates, [email protected], (408) 978-1407 |
| Consulting, Training, Development, SysAdmin, {BSD,SVr3,SVr4} Unix Systems |
| "Think Globally, Act Locally." "Dubito ergo sum - I doubt therefore I am." |
| Proud Dad of Katherine Yelena (2.5 years) and Robin Gregory (born 2/28/92) |


[next article]
From: [email protected] (Hongjiu Lu -- Graduate Student)

Subject: Re: ispell works
Keywords: ispell emacs
Message-ID: <[email protected]>
Date: 19 Apr 92 06:59:11 GMT
References:
Sender: [email protected] (USENET News System)
Organization: Washington State University
Lines: 7

In article [email protected] (Michael Pereckas)
writes:
>I have compiled ispell 2.0.02 with no problems using gcc 2.1.
>

Can we have the latest one? 3.xx. I think it was released in Jan. 1992.

H.J.


[next article]
From: [email protected] (Steve Frampton)

Subject: Some Questions Before Taking The Plunge.
Keywords: linux install
Message-ID: <[email protected]>
Date: 19 Apr 92 01:20:40 GMT
Organization: Vicuna Systems, Kingston, ON (613) 547-5066
Lines: 40

Hi everyone.

Recently I have seen the light and decided to come over to Unix from
MS-DOS. I have been playing around with 0.95a of Linux to get my feet
wet and have been quite impressed.

Last night I switched hard drives around, now I have my 40 MB as MS-DOS
drive C:, and my 80 MB hard drive has been reformatted and it's DOS
partition removed (as far as MS-DOS is concerned it doesn't exist).

I have been following the messages on this newsgroup, I have read the
beginners guide, the installation instructions, and the FAQ's, and am now
ready to take the plunge.

I understand that I can use a hard disk partitioning tool that comes on
the 'root' disk (the name escapes me at the moment), then I would use
'mkfs' and then 'mount' the file system to get Linux installed on my 80
MB hard drive (it's a Seagate ST1102A IDE drive).

Is the above basically correct? Is there a guide somewhere that explains
how to add users, etc.? Has anybody met with success with regard to
running Waffle under Linux?

One thing concerns me, however...once I install Linux on my 80 MB hard
drive, MUST I continue to boot up Linux under floppy drive? Most
implementations of Unix that I have seen have, upon boot-up, displayed a
'boot' prompt -- the user could type DOS to enter MS-DOS, or (or
just wait a few seconds) to go into Unix...is there anything like this
available for Linux? I really hate the idea of have to use a floppy all
the time (and yet I still want to be able to jump into MS-DOS on those
rare occasions).

I am impressed with Linux! Good work!

Thanks in advance.

------------------------------------------------------------------------
Steve Frampton ([email protected]), Kingston, Ontario, CANADA
"It only took me 10 years of MS-DOS before I knew I should be using Unix
...am I quick or what?"


[next article]
From: [email protected] (Joseph Knapka)

Subject: rm badness
Summary: rm +r hangs
Message-ID: <[email protected]>
Date: 19 Apr 92 18:41:23 GMT
Organization: University of Georgia, Athens
Lines: 24

OK, this has happened twice in rapid succession, and I think it might
indicate a problem...

I used rm +r early this morning to try to remove my groff source tree.
Didn't work -- rm hung and didn't touch any of the stuff I told it to
delete. I was root, I was the only login, and there were no background
jobs running. And I wan't in the dir I was trying to delete. And the
partition had fsck'd cleanly at the last reboot.I tried to kill the
process (kill, kill -9) to no avail. Only a reboot would fix things.

So I fsck -a'd the partition after rebooting, and got two "zone marked
used, no file uses it" errors. OK. Tried to rm +r the directory again,
no problem. But then I tried to remove my old linux source tree, and
got the exact same behavior. Including the "zone marked used" error
from fsck. I'm not going to use rm +r anymore until I find out why
this happens.

Does anyone have a notion what might be going on here? Has anyone else
had similar problems? Is my filesystem FUBAR? If so, why won't fsck
tell me so?

Thanks for any advice,

Joseph


[next article]
From: [email protected] (Niels Skov Olsen)
Newsgroups: alt.os.linux,comp.os.linux

Subject: Keyboard lockups SOLVED (i.e. worked around)
Message-ID: <[email protected]>
Date: 19 Apr 92 18:10:29 GMT
Sender: [email protected]
Organization: Department of Computer Science, U of Copenhagen
Lines: 31

Ok, I isolated the problem to a line i kernel/chr_drv/keyboard.S.
In the routine that is supposed to set the LED's I uncomment the
second call to kb_wait. I sure as h*** don't what's going on,
but it seems to be some sort of quirk in some keyboards. This is
only a temporary work-around, and even though my CAPS LOCK, PAUSE,
SCROLL LOCK and NUM LOCK now don't cause lock up, the LED's are still
not working. The NUM LOCK LED is on and that's it.

Here is the chunk of code. The line I commented out is marked [NSO].

Excerpt of kernel/chr_drv/keyboard.S:

_set_leds:
movb _kleds,%al
cmpb old_leds,%al
je 1f
movb %al,old_leds
call kb_wait
movb $0xed,%al /* set leds command */
outb %al,$0x60
/* call kb_wait */ /* [NSO]: This line teases me... */
movb _kleds,%al
outb %al,$0x60
1: ret


I hope someone with keyboard hardware specific knowldege will look
into this.

Easter,

Niels


[next article]
From: [email protected] (Niels Skov Olsen)

Subject: Re: expr (GNU shellutils)
Keywords: Softscroll
Message-ID: <[email protected]>
Date: 19 Apr 92 18:38:59 GMT
References: <[email protected]
alter.bellcore.com> <[email protected]> <[email protected]
ogsci.cog.jhu.edu>
Sender: [email protected]
Organization: Department of Computer Science, U of Copenhagen
Lines: 109

[email protected] (Bill Bogstad) writes:

>In article <[email protected]> [email protected] (Niels Skov Ol
sen) writes:
>>expr is in the shellutils-1.6.tar.Z (or therabouts) on prep.ai.mit.edu.
>>In this package are some other nice utils too.
>>
>>I compiled it with GCC2.1 with only small adjustments to the makefiles
>>(I don't have echo) and a little finetuning in obvious places. I didn't
>>make notes as I went along, so I don't remember the exact steps. It
>>should't be a problem though. Mail me if it is and I will try to
>>reconstruct what I did.

> Well, I'm trying to compile it with GCC 1.40 and have found the
>following problems so far.

>1. no echo - put the following in your /bin directory

> #!/bin/sh

> echo $*

>2. va_list redefined in stdio.h vs. stdarg.h - I modified stdio.h to
> include stdarg.h rather then defining va_list itself.

>3. No bison - get the binaries out of binaries/compilers on tsx-11.mit.edu
> (I haven't actually gotten this far yet...)

>If I get a chance tonight, I'll add more to the list above...

> Bill Bogstad
> [email protected]

Ok, here's what I did (for GCC2.1) in cdiff form. This is perhaps overdoing
it, but it leaves no room for (mis)interpretation ๐Ÿ™‚

Untar this file in your shellutils directory and apply the patch. Then
apply the little patch in lib/ (They are both called linux.cdiff).
Then run configure and then make.

Niels

Here goes:

begin 644 sh.u.dif.tar.Z
M'YV0;-*XJ8/'[email protected]:MY$>2&F?8$`FB1HP;-&!PY'@C8XP8,F38^`CCHL^?0(,*C5AG#ITPM#9VA$YDZA4JUJM6K6+-JM:JB*[email protected]>6\H-,&SHLY:,JP85.'3AHVM19,P:\J8>5O&A<`$4\+0`:$DC!L0,7"`D*G#I0X9,1#GR"%#08O+(.[FWM[]_`@X/`29H8L0T=,$[/R"&9LH*NL&/+5O'::XT<[email protected]@0+8A<2>)D
M2A(M1;X0_P*%2I$F"9*8`9'G39VD<_(8+=,&!!F]`LO,`0'&J)PZ8P;[email protected]
MC9XR)T`(!)%][email protected]=+B[0..B-XC?P1TQ7'%:?"&$%$\L4803T4U'11)/#"[email protected]
M<5-H(=]XW^[email protected]"!&6]Z](9X;)ZCWAAQK])>1?T048<04(/0``A`[email protected]$
M?PJ$`,(0+0I1Q1$Q`G>&?T,8P4001\`H(PDH\%B$CT>D`%[email protected]@LO!"3&
ME$S.(<=!:
M'D&"Y666;XP1!AO^_0:%''^"8,:)()01QAAHR.>&48=VUN%H;YPA1QAML`""
M&R>VD6D>D99%[email protected]'&"2IZ*)"?>P$JHV68W99;9)==EBAPPE%X7'++-?=<
[email protected]]19AYUV^76787CCE4?'>>F!L-ZE[L%G*7W:W=?L?K$"2,2$!!J(H((,[email protected]?"
[email protected]&26^&%WH&W88M"U1:B:667*[email protected])9ABDFDF"FBJJ6.;"2LY9YUWYCF%P/[1FH:M7X5%J*&(^@;"
MHHT^FI2DE%J*Z5H<@L"IIZ"*2JHM"#G,P`).BO'&YW\@3%&'&`@QFMZ):[email protected]'`[email protected]'#>8:;FM:W-:Z?ELQQIF"H'
MJD?)<489=,P1ZQ15"$%$$E+$2?'%()CJ!HX99:K#[email protected]?:V3!)N..)I[A!B
MI)2V';=`9X!``A"8DU"VYP>9KGJ8GC/91!`*2BE[C<05D:;I0+#>X96>W[NA
M?RY`@<033F2A`PB90J[S46M-KF/ED%X^7^:%'YYX"L&W74;HC>-%.N^IKPY"
MZR2\+D?L*,Q>N^GMX^Z$[K;W?O[O+G0_?$;%'Y_\\[email protected],!
M]9)BO<[email protected];W/;ZQ[HT"`Z\;FA=*M]I="XR%/>6L+PP&=1P4X)7"[email protected]"@.>UU[G/?HV#XUC"^#%Y/?2A,GP?;
M!T(1-D%^].,="E4HO#(0SX7_6Q>&-QB%#7D80V<,Y[@'
M1/"-[H+D.Z+YT(=$^+DOA':[email protected]&%_7OA\KKXQ3".\7%?$Z.DW%#&
[email protected]$X`_O-<$*[email protected]'>FXQ#LZ48_UFV+^5FC%C"2`:,"9CF;`TYGS
M_7$O9W`!INA0E"OZ#X:*-(SS&'6H7#*2M\'VW(^'\]BA*[email protected]:[email protected](-L)[email protected]!UL*TFV28H,OG8<0HTS.
ME,>D3UK6TI:WQ*4\/6N!-T_P`CN400[M^>,7S$.Z%X2`#/'1)PA.8)\M>`$&
M+M\$$*S>`&4%FQ5UC3&M>`TROG_49L9`O3]]B&-[[email protected]!LWYP8I.MAM-'E#
M"M^0\K?`#4Z-B0O2XKX4/D02D'G1`Z8CA0E)8D:PC4(4P1LQJ$Q-EDT$2:SC
M,_$832.4,)0<]",@L0C#[email protected])(3*P,?"3^S3M)[X%.K!=FJ03G`E8-*="83
MH3E"NTX3KWT<917-F47`[email protected]'-(UL)B];"5!(%B+\G893HVKIUL8AZE
M&<43YE6S?[QF(+,86AK:<+0--*TDV4C)>*XVF8U];#/79T?9UO6N4KRM-3D+
MPQHFB8M>I`,8?=G(P9;[email protected]+F+3NM8XNC5,RNU09)D[UT_2EIK2)25U!YG=
M[1[2>=P%+F'!6TRTMNVXF&QK!]$+6\EZM\H^ORUOGUWBISD-VE[0[-*[email protected][WM9"5ZV3I6EGHVC:S#<9F=[29&6YR
MQIMC>&4:8EGAMH!M>,ZSGO?,YS[[&2I:(M-`"@(FA?SYT!8!R48ZMD0D-8&(#F[SD!CM9#*)';96B]$TI4GF*5E)-ZE:[^[email protected]!L6C8HLIA%R?1L
MM:D32;-=4[=M7LX$,M6\`D*03!"[email protected];!]&4,RR.XP5V0!\H!(9YEP`,[email protected]!XM'Z.`0$4,;TC`&
M4!)8=3;[email protected]'+ZZ,+`997MW;@P>\
M-QRGXB8W3WWEX!E/YW3J9K>[SQ=OU,V[WO?.MYI,R6\4^[email protected]=1+XUYZ.MO>'3>WAB(S[Q:ED<#1C7.(M#CE_.0]X+I[!`'TZ0B>ZT6?>AJ0GG^DLYWS45UFWJI?AZLQCV]:[OBVPB[UG
M9;<4VM6>GPZU'01O1SLZYYXRN_>]#7GWF?O]?C5!B<769YDGD^TI%[IH[#YM
M<7UV8!`)0`5UX$U$\#[email protected](`,=D1HZ,`,[email protected]!K,[email protected]"LMX#,!Z`8#.`;[email protected]*B
M01J*$0.G`0/:[email protected]:4P,S``+F%G%[email protected]`("&"9_M`/.
M\P(J,#,@,@54(`7$<02B<@>HES/A]P;3076.4BJJY01!T`1%T`-6$`1,4`5%
M(`)>=QA.F(-,[email protected]"4P^`+^<78IDX%)9S'30CH>([email protected]&5GP(,9
ML0?^D1%S"(-7]P77UR$R(A#?`2ANB#<7)"HGT`,GP#U[B$YG4';JUQ[O\06#
[email protected](<@P2B1N!U)T8==409PH(D6"`(LR`(N.'@MX!\R^$\TB`8V&(H:J(-N
M0(H9X8-`.!A"2(1.8(36DH25LH1-:#MF!$?E89DN(;[email protected]<(APN"8HD6.7,[email protected]:(>:J(MZ(@DP8F2^$_TD2V6>(_OH8D908^>>#ZP2(9`(F9`*N9`,V9`.
M^9`0&9$2.9$469$6>9$8F9$:N9$^[email protected]&9(B.9(D69(F>9(HF9(JN9(L
9,XF9,ZN9,\V9-4`9(B
`
end


[next article]
From: [email protected] (Chris Waters)

Subject: 850 gig extended*2 partition on 300 meg drive? ๐Ÿ˜‰
Message-ID:
Date: 19 Apr 92 19:46:29 GMT
Organization: Netcom - Online Communication Services (408 241-9760 guest)
Lines: 28


Truly inconsequential, but I thought y'all might find this amusing. I
just tried the .95c+ bootimage out at tsx-11 (I had been using .95a),
and all of a sudden I get the following report from fdisk:

Disk 0:
/dev/hda1: 32728 blocks active 16-bit DOS (<32M)
/dev/hda2: 32760 blocks minix
/dev/hda3: 32760 blocks minix
/dev/hda4: 221760 blocks extended partition (don't use)
/dev/hda5: 32728 blocks 16-bit DOS (<32M)
/dev/hda6: 189000 blocks extended partition (don't use)
/dev/hda9: 850999312 blocks active? (0a) unknown partition type 0x6C
/dev/hda10: 272052916 blocks active? (73) unknown partition type 0x6E
/dev/hda11: 269488144 blocks active? (72) unknown partition type 0x79
/dev/hda12: 10668 blocks active? (53) unknown partition type 0x53

Boy, I think I'll start storing EVERYTHING on /dev/hda9!! ๐Ÿ™‚ ๐Ÿ™‚ ๐Ÿ™‚

I guess that, as stated in the installation guide, the extended
partition access is not fully debugged yet.....

Um, if no one else is working on this, maybe I'll take a look at the
code as soon as I get a chance. I've been trying to think of something
I can do to contribute to the project, maybe this is it?
--
Chris Waters | the insane don't | I figure the odds be fifty-fifty:
[email protected]| need disclaimers | I just might have something to say. --FZ


[next article]
From: [email protected] (Werner Almesberger)

Subject: Re: mtools and messed characters on hard boot help.
Message-ID: <[email protected]>
Date: 19 Apr 92 20:17:14 GMT
References: <[email protected]>
Sender: [email protected] (USENET News System)
Organization: Swiss Federal Institute of Technology (ETH), Zurich, CH
Lines: 25


In article <[email protected]> [email protected] (Brian Chojn
owski) writes:
> 1) I grabbed the binaries for mtools from tsx-11.mit.edu, and when they are
> run, I always get the message 'Device not found -- please specify device' or
> something like that. What have I not done to get these to work?

It's probably "Unknown device , please specify all parameters".

This means that the device name is not of any of the forms /dev/at*, /dev/ps*,
/dev/PS* or /dev/hd*. Mtools tries to guess disk parameters for those device
names, but it doesn't know how to handle different names. (This is supposed to
change someday.) You have to specify all parameters in the /etc/mtools file,
e.g.

A /dev/foobar 12 80 2 15 # 5.25" HD 1200k
X /dev/barfoo 12 40 2 9 # 5.25" 360k
B /dev/xyzzy 12 80 2 18 # 3.5" HD 1440k
Y /dev/yzzyx 12 80 2 9 # 3.5" 720k
C /dev/whatever 12 0 0 0 # any hard disk (12 bit FAT)

- Werner
--
_________________________________________________________________________
/ Werner Almesberger, ETH Zuerich, CH [email protected] /
/ IFW A44 Tel. +41 1 254 7213 [email protected] /
/_BITNET:[email protected]__HEPNET/CHADNET:_[20579::]57414::ALMESBERGER_/


[next article]
From: [email protected] (Jim Winstead Jr.)

Subject: Re: 850 gig extended*2 partition on 300 meg drive? ๐Ÿ˜‰
Message-ID: <[email protected]>
Date: 19 Apr 92 20:59:26 GMT
References:
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, WIBSTR
Lines: 21

In article [email protected] (Chris Waters) writes:
>
>Truly inconsequential, but I thought y'all might find this amusing. I
>just tried the .95c+ bootimage out at tsx-11 (I had been using .95a),
>and all of a sudden I get the following report from fdisk:
>
>[listing deleted]

The problem is not really with the Linux 0.95c+ kernel (or at least,
it shouldn't be!), but rather with the fdisk on the 0.95a root disk,
which is presumably the one you were using. That fdisk cannot handle
extended partitions properly at all, but apparently the 0.95c+ kernel
now handles them correctly.

The fdisk on the 0.96 root disk will handle extended partitions
correctly.... pfdisk still can't create 'em, though.
--
Jim Winstead Jr. (CSci '95) | "Catch a fish!"
Harvey Mudd College | -Geddy Lee,
[email protected] | San Diego Sports Arena
Disclaimer: Mine, not theirs! | January 20, 1992


[next article]
From: [email protected] (Guido Kueppers)

Subject: re: IDE drive performance with linux
Message-ID: <[email protected]>
Date: 19 Apr 92 20:07:41 GMT
Sender: [email protected]
Organization: RHRZ-Uni-Bonn
Lines: 8

Greetings,
thanks to all who have replied to my IDE drive problem. From the responses I
got so far I gather that the drive's poor performance is specific to my drive
and possibly due to inappropriate configuration (translation vs. native mode).
If there is any interest I will post my newly accquired wisdom on IDE drive
setup when the problem is solved (Or should I rather spare you the details?)

Guido


[next article]
From: [email protected]
Newsgroups: comp.os.linux,alt.os.linux

Subject: Re: compress/tar problems on SCSI installation
Message-ID: <[email protected]>
Date: 19 Apr 92 20:27:03 GMT
References: <[email protected]>
<[email protected]> <[email protected]>
Organization: Penn State University
Lines: 5

By golly that should work. Why didn't I think of it?

Actually I got around it by making a second root image disk and erasing
mkdir, rmdir, and cat from it to get enough space. All is working now, and
yes I am happy.


[next article]
From: [email protected] (Doug Evans)

Subject: Re: Linux SCSI support
Keywords: scsi, future domain, linux
Message-ID:
Date: 19 Apr 92 18:42:03 GMT
References: <[email protected]>
Organization: Edmonton Remote Systems, Edmonton, AB, Canada
Lines: 12

> Like I said, people are working on this, but if you have anything
> to contribute, feel free to join the SCSI mailing list at
>
> [email protected]

Ah hah! There are some mailing lists. Could someone tell me
what all of them are? Thanks.

Doug Evans |
..!{canada,uunet}!sspiff!dje | Vancouver 11, Calgary 0
[email protected] | There is no joy in Cowtown.
[email protected] | But how 'bout those Canucks, eh?


[next article]
From: [email protected] (Hongjiu Lu -- Graduate Student)

Subject: Re: expr (GNU shellutils), gcc 2,1 and libc.a
Keywords: gcc 2.1, libc.a, library
Message-ID: <[email protected]>
Date: 19 Apr 92 19:58:43 GMT
References: <[email protected]
alter.bellcore.com> <[email protected]> <[email protected]
ogsci.cog.jhu.edu> <[email protected]>
Sender: [email protected] (USENET News System)
Organization: Washington State University
Lines: 24

In article <[email protected]> [email protected] (Niels Skov Olse
n) writes:
>[email protected] (Bill Bogstad) writes:
>
>>In article <[email protected]> [email protected] (Niels Skov O
lsen) writes:
>>>expr is in the shellutils-1.6.tar.Z (or therabouts) on prep.ai.mit.edu.
>>>In this package are some other nice utils too.
>>>
>>>I compiled it with GCC2.1 with only small adjustments to the makefiles
>>>(I don't have echo) and a little finetuning in obvious places. I didn't

When you use gcc 2.1 to compile any program, be sure to check the
header files. LOTS of functions are in libc.a. You should use libc.a if
posssible. All the functons in libc.a are declared in the header files.
Look into unistd.h, stdlib.h, string.h, regex.h and time.h, etc.
Somebody is working on a complete list. I hope it will come out soon.
I think I can say almost all glibc.a and usual Unix C library functions
are in libc.a. If you find something very usful and not in libc.a,
please send me the source. I can take a look.

BTW, I tried to fix malloc(0) in stdlib.h. It will work with ANSI code.
Take a look at it. Be sure they work right. I strongly suggest to use
-Wall during compilation.

H.J.


[next article]
From: [email protected] (Ted Manos)
Newsgroups: comp.os.os2.misc,comp.os.linux

Subject: Re: Any comments on Quantum 240A ??
Message-ID: <[email protected]>
Date: 19 Apr 92 16:59:25 GMT
References: <[email protected]>
Organization: University of Illinois at Chicago
Lines: 30

In article <[email protected]>,
[email protected] (David A. Sinclair) says:
>
>Hello, world.
>
>I need to buy anew hard drive for my system so I can install Linux and
>OS/2 2.0. I am considering the Quantum 240A, (240 meg IDE) and would
>like to hear from anyone who is either using this drive with either
>OS, or has tried and failed to do so. Anyone have any comments about
>using this drive with OS/2 GA and / or Linux? Anything to watch out
>for?
>
>Thanks for any info you can provide!!
>
While I can't address using the LPS240AT with either OS/2 or Linux
directly (I run OS/2 on a Fujitsi M2624FA (520MB/12MS SCSI)), I HAVE
used a number of 240ATs in systems I have built for clients and
FRIENDS. It is probably THE NICEST, FASTEST IDE drive I have ever
come across (and I've used a lot of different drives!). I can't
think of any reason offhand why you should have any problems using
it with OS/2. I have no idea about Linux (I'm getting ready to get
386BSD 0.0 myself). Hope this helps.

-Ted
-------------------------------------------------------------------------
Ted Manos BITNET : [email protected]
Systems & Networking Consultant Internet: [email protected]
Alpha Omega Consulting Group, Ltd UUCP net: ...!uunet!aocgl!tmanos
400 Springhill Drive Bell net: (708) 980-7919
Roselle, IL 60172-2573 FAX net : (708) 980-4458


[next article]
From: [email protected] (Scott Silverstein MD)
Newsgroups: comp.os.os2.misc,comp.os.linux

Subject: Re: Any comments on Quantum 240A ??
Message-ID: <[email protected]>
Date: 19 Apr 92 23:29:22 GMT
References: <[email protected]> <[email protected]
vm.uic.edu>
Sender: [email protected]
Followup-To: comp.os.os2.misc
Lines: 9
Nntp-Posting-Host: grip.cis.upenn.edu

In article <[email protected]> [email protected] (Ted Manos) w
rites:

> I'm getting ready to get 386BSD 0.0 myself

I have little enough faith in 1.0 releases, let alone 0.0 releases...

I'll wait for 2.11, at least.

-- SS


[next article]
From: [email protected] (Alex C. Liu)

Subject: Partition tables suggestions...
Message-ID:
Date: 19 Apr 92 23:54:41 GMT
Sender: [email protected] (Alex Liu)
Organization: None to Speak of
Lines: 14
NNTP-Posting-Host: aludra.usc.edu

I was wondering, would it be possible to implement logical partitions
within a Linux physical partition? This would help us get around the
4 partition limit in MS-DOS. The idea would be simmilar to what
MS-DOS does with the Extended Partition can be partitioned into
Logical drives.

(BTW, is anybody implementing SysV style IPC? Is it gonna be in the
next release? ๐Ÿ™‚

--
_____________________________________________________________________________
Alejandro Liu |EMail: [email protected] |All mispellings are intentional
3131 Mc Clintock #373F |Voice: 213-745-2431 |Anything mentioned here is not
Los Angeles, CA 90007 | |necessarily true.


[next article]
From: [email protected] (Peter Williams 8169821)

Subject: shared versions of tex/mf/dvips available
Message-ID: <[email protected]>
Date: 19 Apr 92 23:53:31 GMT
Sender: [email protected]
Organization: Design Computing Unit, The University of Sydney
Lines: 13
Nntp-Posting-Host: gresley.arch.su.oz.au

*.a files for shared versions of tex/mf/dvi programs are available in
texmf.a.tar.Z and dvips.a.tar.Z in pub/linux at archsci.arch.su.oz.au
(129.78.66.1)

Makefiles are included (just change the target directory for the binaries)

Remember tex, latex and slitex are created by linking to virtex and mf is
created by linking to virmf.

--
Peter Williams |e-mail: [email protected]
Key Centre for Design Quality |phone: +61-2-692 2053 or +61-2-660 6156
University of Sydney |+++++++++++++++++++++++++++++++++++++++


[next article]
From: [email protected] (Adam Justin Thornton)

Subject: A new game port, and mtools problems...
Message-ID: <[email protected]>
Date: 19 Apr 92 23:26:04 GMT
Sender: [email protected] (News)
Organization: Milo's Meadow
Lines: 27

I'm using the mtools.a supplied with DLE.tar.Z in the gcc2.1 distribution,
and I have C defined as /dev/hda1 and D as /dev/hdb1, which is where they
are. However, when I try mtools it gives me mtools: command not
found error message. What am I missing? Is there some syntax I should be
aware of?

Second, I compiled dungeon (zork) last night; no problems, worked straight out
of the box. Once I get mtools straightened out or figure out how to get my
serial line to dial out, I'll upload it (the tar.Z file I'll upload has
all source and a statically linked binary, but you'll have to edit the makefile
and recompile if you want the Great Debugging Tool).

Third, I got /dev/lp as char. device, major 6 minor 1 (LPT1:), which is an
HPIIP. When I compiled the lp package, all worked OK, but catting stuff to the
printer causes it to be eaten (nothing ever appears). If someone has a working
printcap that fixes this, it'd be really cool. When I try to manually invoke
lpd it tells me the lock file exists, even though the one in /etc/printcap
manifestly doesn't. I'm using the .95c+ kernel.

Help would be appreciated.

Adam
--
"Adam Thornton plays the homosexual Horatio without succumbing to _too_ many
cliched stereotypes. His Horatio, Hamlet's bosom companion...has a penchant
for cream dresses and thigh-high black leather boots." | [email protected]
Rice and I don't share opinions, thankfully. | Retry Student A | 64,928 | ๐Ÿ˜‰


[next article]
From: [email protected] (David Engel)

Subject: Re: A new game port, and mtools problems...
Message-ID: <[email protected]>
Date: 20 Apr 92 01:49:33 GMT
References: <[email protected]>
Organization: Optical Data Systems, Inc.
Lines: 15

[email protected] (Adam Justin Thornton) writes:
: I'm using the mtools.a supplied with DLE.tar.Z in the gcc2.1 distribution,
: and I have C defined as /dev/hda1 and D as /dev/hdb1, which is where they
: are. However, when I try mtools it gives me mtools: command not
: found error message. What am I missing? Is there some syntax I should be
: aware of?

As described in README-DLE, you need to create symlinks to mtools for
each of the mtools commands.

David
--
David Engel Optical Data Systems, Inc.
[email protected] 1101 E. Arapaho Road
(214) 234-6400 Richardson, TX 75081


[next article]
From: [email protected] (Kirby Hughes)

Subject: linux vs. 386bsd (was Re: Any comments on Quantum 240A ??)
Message-ID: <[email protected]>
Date: 20 Apr 92 01:25:27 GMT
References: <[email protected]> <[email protected]
vm.uic.edu> <[email protected]>
Organization: U of Arizona CS Dept, Tucson
Lines: 16


In article <[email protected]> [email protected] (Scott Silverstein
MD) writes:
>In article <[email protected]> [email protected] (Ted Manos)
writes:
>
>> I'm getting ready to get 386BSD 0.0 myself
>
>I have little enough faith in 1.0 releases, let alone 0.0 releases...
>

Could someone please tell me which (free) unix I should get for my pc.
I know about linux from reading every note in here since this group started;
it sounds really good, but I'd like to know why I should use it instead
of, say, 386bsd (which I know nothing about).

Thanks,
Kirby Hughes


[next article]
From: [email protected] (Peter Williams 8169821)

Subject: Re: shared versions of tex/mf/dvips available
Message-ID: <[email protected]>
Date: 20 Apr 92 01:58:17 GMT
References: <[email protected]>
Sender: [email protected]
Organization: Design Computing Unit, The University of Sydney
Lines: 9
Nntp-Posting-Host: gresley.arch.su.oz.au

just a postscript to my previous message:

the shared binary files (when linked with the -s option) are much smaller than
the corresponding *.a files in most cases.

--
Peter Williams |e-mail: [email protected]
Key Centre for Design Quality |phone: +61-2-692 2053 or +61-2-660 6156
University of Sydney |+++++++++++++++++++++++++++++++++++++++


[next article]
From: [email protected] (Charles Hedrick)

Subject: Re: ESDI drives
Message-ID:
Date: 20 Apr 92 02:11:12 GMT
References: an.COM>
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 20

[email protected] (Kayvan Sylvan) writes:

>Is this true (partition sizes limited to 64MB)??? If that's the case
>it would be a shame. What about those of us who have larger drives?

The problem is that Linux is using the Minix file system, which is
limited to 64MB. There's work going on to do large file systems.
However now that extended partitions work, it shouldn't be a big
problem. Just create a big extended partition with a bunch of logical
partitions that are 64M or smaller. In 95c or later Linux should be
able to handle them fine. The public-domain partition programs don't
seem to be able to make extended partitions, but fdisk in DOS should
work fine.

The idea is that you make file systems on each of the logical
partitions, and then stick mount commands in /etc/rc. Once you've
done that, all your disk space will be accessible. The only serious
problem is that you have to worry about splitting up your usage
evenly across several 64MB pieces. But that should hold you until
the larger file systems are working.


[next article]
From: [email protected] (Charles Hedrick)

Subject: gdb still isn't working
Message-ID:
Date: 20 Apr 92 02:23:04 GMT
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 24

I've been trying to get gdb working. The problem is more subtle than
I had first thought. It worked fine up to 0.95b. However as of 0.95c
and 0.95c+, it stopped working. My main problem is that breakpoints
don't work. It's hard to be sure, but it looks like when the program
comes to a breakpoint, typically it gets segmentation fault, though
sometimes I also see the program terminate with signal 5 (breakpoint).
At first I assumed that this was because of changes in the header
files, such as a.out.h, but I've built 0.95c+ with the old a.out.h and
that doesn't fix it.

At this point I can't even tell whether the problem is in gdb or the
kernel. I'm hoping the folks who did gdb and ptrace in the first
place can look into this, since trying to learn the innards of gdb,
ptrace, and signal handling could take quite some time... I'd
be happy to test any fixes.

By the way, even in the working kernel, gdb doesn't trap errors. When
a program blows up, it's supposed to trap to the debugging, so you can
look around. It doesn't do that. I think die, in trap.c, needs to
check whether the program is being debugged or not, and do something
different.

Finally, gdb seems to lose control over the program being debugged
if my shell is tcsh. It works only with bash.


[next article]
From: [email protected] (Demian A. Johnston)

Subject: Re: telecomm in Linux, users
Message-ID:
Date: 20 Apr 92 02:36:56 GMT
Distribution: comp.os.linux
Organization: Sophomore, Physics, Carnegie Mellon, Pittsburgh, PA
Lines: 12
In-Reply-To: <[email protected]>

I had this problem before. I had found that by using the dial command I
was all set. This is a wild guess but I would imagine it is cause by
and noninitialized serial port. That is, stty speed 2400 >/dev/ttys?


Thats whats worked with me and Kermit...

Demian
---> [email protected]
---> [email protected]
---> [email protected]


[next article]
From: [email protected].edu (Curtis Yarvin)

Subject: Re: Yet another X question
Keywords: X Windows
Message-ID: <[email protected]>
Date: 20 Apr 92 03:44:55 GMT
References: <[email protected]>
<[email protected]>
Sender: [email protected]
Organization: Brown University Department of Computer Science
Lines: 20

In article <[email protected]> [email protected] (
Adam Thompson) writes:
>In <[email protected]> [email protected] (Hongjiu L
u -- Graduate Student) writes:
>
>>From my own experience, a real X11R4 can take as little as 10MB to run.
>>on a 386.
>
>Perhaps you're assuming he doesn't want to compile it ๐Ÿ™‚
>
>According to the docs in the R5 distribution, the source tree untars to
>roughly 150-200 Megs. To compile, expect to need at least another 100Meg.

Yes, but this is the fault of stupid makefiles that leave .o's lying
around. If you're clever and know what you're doing, you should be able to
build it utility by utility, and I doubt that all the object and source for
any one utility is more than a few megs.

The reason X is such a hog is that its designers assumed people have a few
hundred megs of disk lying around to waste.

c


[next article]
From: [email protected] (Drew Eckhardt)

Subject: Re: Some Questions Before Taking The Plunge.
Keywords: linux install
Message-ID: <[email protected]>
Date: 20 Apr 92 04:33:35 GMT
References: <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 51
Nntp-Posting-Host: ladymacb.cs.colorado.edu

In article <[email protected]> [email protected] (Stev
e Frampton) writes:
>Hi everyone.
>
>Last night I switched hard drives around, now I have my 40 MB as MS-DOS
>drive C:, and my 80 MB hard drive has been reformatted and it's DOS
>partition removed (as far as MS-DOS is concerned it doesn't exist).
>
>I have been following the messages on this newsgroup, I have read the
>beginners guide, the installation instructions, and the FAQ's, and am now
>ready to take the plunge.
>
>I understand that I can use a hard disk partitioning tool that comes on
>the 'root' disk (the name escapes me at the moment), then I would use
>'mkfs' and then 'mount' the file system to get Linux installed on my 80
>MB hard drive (it's a Seagate ST1102A IDE drive).
>

Makesure you have atleast 2 partitions on the 80M drive,
and that none of them are >= 64M, as that's the limit for the minix file
system.

>Is the above basically correct? Is there a guide somewhere that explains

Yes.

>how to add users, etc.? Has anybody met with success with regard to

It's under development.

>running Waffle under Linux?
>
>One thing concerns me, however...once I install Linux on my 80 MB hard
>drive, MUST I continue to boot up Linux under floppy drive? Most

You must have a small partition on your first physical hard disk
with a "miniroot" - ie /vmunix, /etc/config, /etc/disktab,
and /shoelace if you want to boot Linux off of the hard disk.

Either switch the disks so the 80M is the first physical disk,
and put all your linux stuff there, a small C: partition
for DOS, and use Winiboot to select what you boot, or put a
mini-linux partition on the 40M drive, and use Winiboot on it.

>implementations of Unix that I have seen have, upon boot-up, displayed a
>'boot' prompt -- the user could type DOS to enter MS-DOS, or (or
>just wait a few seconds) to go into Unix...is there anything like this
>available for Linux? I really hate the idea of have to use a floppy all
>the time (and yet I still want to be able to jump into MS-DOS on those
>rare occasions).

shoelace and winiboot.


[next article]
From: [email protected] (Jim Winstead Jr.)

Subject: Re: Some Questions Before Taking The Plunge.
Message-ID: <[email protected]>
Date: 20 Apr 92 05:35:37 GMT
References: <[email protected]> <[email protected]
edu>
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, WIBSTR
Lines: 30

In article <[email protected]> [email protected] (
Drew Eckhardt) writes:
>In article <[email protected]> [email protected] (Ste
ve Frampton) writes:
>>One thing concerns me, however...once I install Linux on my 80 MB hard
>>drive, MUST I continue to boot up Linux under floppy drive? Most
>
>You must have a small partition on your first physical hard disk
>with a "miniroot" - ie /vmunix, /etc/config, /etc/disktab,
>and /shoelace if you want to boot Linux off of the hard disk.
>
>Either switch the disks so the 80M is the first physical disk,
>and put all your linux stuff there, a small C: partition
>for DOS, and use Winiboot to select what you boot, or put a
>mini-linux partition on the 40M drive, and use Winiboot on it.

Another possibility is to get bootlin.zip, which was posted to
comp.os.linux, and use that to boot Linux from the DOS partition - I
just switched over to this today, and I have to say it's great. There
is the minor problem of it printing garbage instead of the 'Press
enter for SVGA' message, but that's relatively minor compared to how
much easier it makes life.

Otherwise, my root partition, which has everything except /tmp and
/usr, is only about 2 megs. I would recommend something like 4 megs,
just to make sure you have enough room. (Mine's actually 15 megs, but
only has 2 megs in it - time to repartition. ๐Ÿ™‚
--
Jim Winstead Jr. (CSci '95) | "Catch a fish!"
Harvey Mudd College | -Geddy Lee,
[email protected] | San Diego Sports Arena
Disclaimer: Mine, not theirs! | January 20, 1992


[next article]
From: [email protected] (Linus Benedict Torvalds)

Subject: Re: 850 gig extended*2 partition on 300 meg drive? ๐Ÿ˜‰
Message-ID: <[email protected]>
Date: 20 Apr 92 05:51:29 GMT
References:
Organization: University of Helsinki
Lines: 18

In article [email protected] (Chris Waters) writes:
>
>Truly inconsequential, but I thought y'all might find this amusing. I
>just tried the .95c+ bootimage out at tsx-11 (I had been using .95a),
>and all of a sudden I get the following report from fdisk:

[ "somewhat" erraneous output deleted ]

The old fdisk messes up extended partitions (as did the old kernel), and
shouldn't be used. Not that it does any harm, as the old fdisk only
reads from the disk, but as you saw, it doesn't exactly give good
results.

With 0.95c+, the available partitions are reported at bootup, so the
need for a fdisk is less, but I hope the next release will contain one
that works correctly.

Linus


[next article]
From: [email protected] (Linus Benedict Torvalds)

Subject: Re: gdb still isn't working
Message-ID: <[email protected]>
Date: 20 Apr 92 08:51:43 GMT
References:
Organization: University of Helsinki
Lines: 35

In article [email protected]
edu (Charles Hedrick) writes:
>I've been trying to get gdb working. The problem is more subtle than
>I had first thought. It worked fine up to 0.95b. However as of 0.95c
>and 0.95c+, it stopped working.

Yes, it's a bug in the kernel. It was there already in 0.95b and
earlier, but those had a (buggy) workaround that made it work for most
things anyway. The problem was that not all kernel mode -> user mode
changes checked the error conditions, so things like breakpoints didn't
work too well.

My personal version handles this correctly (as well as doing some other
things in a cleaner manner), but I'm not quite ready for a new release
yet. I could make YAAR (yet another alpha-release) or just mail
interested parties the fixes needed - mail me if you're interested, and
depending on the number of messages I get I'll make it a new release.

Here's a preview of 0.96 (* means already implemented):

* truncate/ftruncate/fchmod/fchown system calls

* io-bitmap allowing user processes controlled access to io-ports (thanks to
obz - needed for X)

* mmap for /dev/mem - (thanks to obz) allows X etc to use the frame buffers

* the signal-handling fixes needed for gdb

- multiple shared libraries (pmacdona)

- cleaned up special files: partly implemented already

and probably some other minor fixes.

Linus


[next article]
From: [email protected] (BATES, ROBERT PATRICK)

Subject: Some Q's...
Message-ID: <[email protected]>
Date: 20 Apr 92 11:04:00 GMT
Sender: [email protected]
Organization: Texas A&M University, Academic Computing Services
Lines: 24
News-Software: VAX/VMS VNEWS 1.41

OK - First things first. I just read a note that said that the Minix system
wouldn't handle any partitions larger than 64MB. Is this in reference to the
Linux partition, or any partitions? (Just curious - I'm running Linux on 30MB
and DOS under 100MB.)

Next question - When I get these patches from tsx-11.mit.edu, how do I go about
incorporatin' them into my current version (I just got 0.95a up Fri and got the
0.95c.patch from MIT tonight)?

Third and final - Is there a way I can get the .tar.Z files from an FTP site
and d/l to my DOS machine (since most of my InterNet work is done offa my VAX
account from my DOS apps), then somehow port them into Linux to uncompress? I
DO have a Sun account, but it's on the other side of campus, where I seldom
ever go... However, I will if I have no other alternative (say, a comm package
for Linux that I could use???)

Thanks,
Rob.

==============================================================================
Robert Bates Disclaimer:
[email protected] "I make no claims..."
Texas A & M University
==============================================================================


[next article]
From: [email protected] (Alex R.N. Wetmore)

Subject: Re: Graphics and IPC questions...
Message-ID:
Date: 20 Apr 92 14:31:20 GMT
Organization: Freshman, Math/Computer Science, Carnegie Mellon, Pittsburgh, PA
Lines: 24
In-Reply-To: <[email protected]>

Excerpts from netnews.comp.os.linux: 18-Apr-92 Re: Graphics and IPC
questi.. by Brian [email protected]
> >Ugh. Isn't that a step backward? Why create a completely new,
> >nonstandard, *local-only*, graphics library? I don't think there is
> >much chance that people will write programs for Linux that use your
> >direct-to-screen library. To put it bluntly, if you want to fiddle
> >with direct hardware access assembler routines, stick to DOS. Even if
> >you don't need networked graphics, I'm certain a significant fraction
> >of the Linux community wants it.
>
> I dont intend on making a local only system. What I do want to avoid is the
> system requirements of X. I dont think there are any X telnet programs for
> the 286. I already have an application, and Unix would be a good multi-user
> platform for it, but it requires characterset based graphics.

Why not implement these in vt100 mode, rather then with direct screen
writes then? Linux seems to have very fast terminal emulation (at least
on my 386/16 with a very old Paradise EGA), and you might as well take
advantage of those using termcap. That way people on terminals (and
telnet sessions, when those exist), can use the software too. It also
doesn't restrict the software to linux, but allows it to work on any
other unix platform (in theory at least).

alex


End of newsgroup comp.os.linux.



 December 12, 2017  Add comments

Leave a Reply