Files
2024-02-19 00:23:35 -05:00

530 lines
20 KiB
Plaintext

From: Digestifier <Linux-Activists-Request@news-digests.mit.edu>
To: Linux-Activists@news-digests.mit.edu
Reply-To: Linux-Activists@news-digests.mit.edu
Date: Sat, 15 Feb 92 15:30:15 EST
Subject: Linux-Activists Digest #52
Linux-Activists Digest #52, Volume #1 Sat, 15 Feb 92 15:30:15 EST
Contents:
Re: Help! Can't boot! (Peter Cooper)
Re: I fixed the TAR problem; kermit; kmem?; DOS emulator (Peter Cooper)
Re: Help! Can't boot! (Court Demas)
Re: Help! Can't boot! (KiYun Roe)
Re: Help! Can't boot! (Sean Eckton)
More questions about Linux :-) (D.Bolla)
Re: who can't open utmp (Peter Orbaek)
init/getty/login announcement (Peter Orbaek)
["Mark W. Eichin": update sources (complete releases?)] (Theodore Ts'o)
[Theodore Ts'o: [Steven Wilson: rogue.tar.Z]] (Theodore Ts'o)
Re: Help! Can't boot! (Steven Wilson)
mtools source and binary available (Charles Hedrick)
Re: More questions about Linux :-) (Charles Hedrick)
----------------------------------------------------------------------------
From: comrade@uniwa.uwa.oz.au (Peter Cooper)
Subject: Re: Help! Can't boot!
Date: 15 Feb 92 01:17:08 GMT
jsaker@odin.unomaha.edu (Jamie Saker) writes:
: Court Demas <cd2a+@andrew.cmu.edu> writes:
:
:
: >I've been trying to get linux to run on my Zeos 486, but I can't seem to
: >get it to even boot up! When I boot up with the rawrited bootimage it
: >just hangs after reading the disk for a second or two. When I tried
: >rawriting from a PS/2 it would boot and then go into a loop displaying
: >what looked like:
: >I have a Zeos 486/33, 8mb, 210mb IDE ctrl, 1.44/1.2mb drives
:
: After posting about my problems w/ Linux boot not booting (disk spins several
: seconds, writes |- Loa -and hangs), I've noticed several other posts from
: owners of 486 systems with similar problems. Can anyone confirm that
: Linux presently hates the 486? Anyone been able to get it to run under
: a 486 and if so, what sort of configuration do you have?
Linux runs fine on the machine I use. It is a 486/33, 120 M IDE drive, 4
megs memory. Brand is something called 'comdex'. Just another clone I
think. SVGA trident card with 1 meg memory. Nice and stable.
------------------------------
From: comrade@uniwa.uwa.oz.au (Peter Cooper)
Subject: Re: I fixed the TAR problem; kermit; kmem?; DOS emulator
Date: 15 Feb 92 01:25:05 GMT
joel@wam.umd.edu (Joel M. Hoffman) writes:
: First of all, thanks to all those who responsed to the problem I was
: having with TAR. It took me an amazingly long time to realize that
: the 3.5" drive I use as a: is a 1.44M, and not 1.2M, drive.... :-)
:
: Has someone fixed the problem with kermit? I can get it to dial
: (sometimes), but when I try to connect, I have less than a second of
: communications before kermit decided I've had enough. Also, why
: doesn't:
:
: echo atx4dt 1234567 > /dev/tty64
:
: dial the phone?
Because the serial ports are by default in raw mode. i.e echo whatnot,
actaully sends the byte seqence "whatnot\n" to the modem, \n being a
newline, NOT the carrige return it wants.
try 'stty opost onlcr -a /dev/tty64'.
:
: Is /dev/kmem implemented/working? I'd like to run something like
: ``top.''
/dev/mm , /dev/kmem are not yet implemented. Linus? 0.13??
: As I understand, all a DOS emulator would need to do is remap memory
: and intercept DOS and BIOS interupts. At least that should handle
: well-behaved DOS programs. Many of the interupts could simply be
: converted into Unix system calls, I suspect. At any rate, is anyone
: working on this? Also, can the 386 have virtual 386 machines? For
: example, could such a DOS emulator run DJGPP's GCC for DOS, which is
: itself a 32 application? What about windows under Linux?
Dos emulators also need to intercept any hardware i/o, any screen
access. And (this is the sticky bit) the emulated DOS needs to reside at
virtual address 0. This is bad because currently linux is sitting there.
A 386 can't have virtualt 386 machine. Say goodbye to 386 stuff under a
DOS emulator. Windows under linux won't work in any sort of protected
mode. The virtual machine built into a 386 is an 8086. i.e. no protected
mode or anything like that.
:
: Again, many thanks Linus!
:
: -Joel
------------------------------
From: cd2a+@andrew.cmu.edu (Court Demas)
Subject: Re: Help! Can't boot!
Date: 15 Feb 92 05:25:37 GMT
Excerpts from netnews.alt.os.linux: 14-Feb-92 Re: Help! Can't boot!
tucker@yuppie.enet.dec.c (660)
> When I saw that and other problems similar to what other people have been
> describing in the installation problems postings, it was becase the file
> had been mangled before it got to the dos machine. I was using ftpmail
> to VMS, then kermit to dos. It took some experimenting before I got a
> bootimage that worked.
Before I transferred it to my machine I zipped it and everything else
with PKZ - so nothing happened there. Also, the executables (rawrite,
etc) all work on my machine, and they were all ftp'd at the same time
from the same site - in binary mode.
ACK!
I've tried both the 1.2 and 1.44 and nothing seems to work! Any other
suggestions???
-court
------------------------------
From: kroe@sbstaff2.cs.sunysb.edu (KiYun Roe)
Subject: Re: Help! Can't boot!
Date: 15 Feb 92 04:35:55 GMT
Regarding whether Linux runs on 486 machines: I have Linux running on
a Gateway 2000 486/33, 4MB RAM, 200MB Western Digital Piranha HD. It's
a Micronics motherboard with Phoenix BIOS.
--
KiYun Roe kroe@sbcs.sunysb.edu
Department of Computer Science
SUNY at Stony Brook
Stony Brook, NY 11794-4400 (516) 632-7675
------------------------------
From: ecktons@sirius.byu.edu (Sean Eckton)
Subject: Re: Help! Can't boot!
Date: 14 Feb 92 19:21:12 GMT
In article <jsaker.698088920@odin> jsaker@odin.unomaha.edu (Jamie Saker)
writes:
> Court Demas <cd2a+@andrew.cmu.edu> writes:
>
>
> >I've been trying to get linux to run on my Zeos 486, but I can't seem to
> >get it to even boot up! When I boot up with the rawrited bootimage it
> >just hangs after reading the disk for a second or two. When I tried
> >rawriting from a PS/2 it would boot and then go into a loop displaying
> >what looked like:
> >I have a Zeos 486/33, 8mb, 210mb IDE ctrl, 1.44/1.2mb drives
>
> After posting about my problems w/ Linux boot not booting (disk spins
several
> seconds, writes |- Loa -and hangs), I've noticed several other posts from
> owners of 486 systems with similar problems. Can anyone confirm that
> Linux presently hates the 486? Anyone been able to get it to run under
> a 486 and if so, what sort of configuration do you have?
>
> Curious...
>
I've got a friend that uses Linux all the time. This is his configuration
(to the best of my knowledge): 486-33, 4Meg, 120 M hard disk, 1.2M disk
drive (A:), 1.44M disk drive (B:), SVGA. He has had no problems at all. I
have run into some strange problems also. I tried to run it on a PS/2-80.
It started up and looked like it was going to work, but then after it asked
me to choose a video mode and I responded, it gave me an error:
copy_to_cooked: missing queues. At that point, it just died. I tried the
same thing on an old XT with a 386Inboard. I just kept getting those strange
errors like were listed:
ax 123
bx 141
cx 34235
dx 0914
Let me ask a question and see if this stimulates any ideas. I tried using a
PS/2 that would boot up on 1.44 and it didn't work. Could that be a
peculiarity with Linux? Maybe it doesn't like 1.44M drives on some machines.
I have seen other people post here and they use PS/2's and Linux is a bit
crippled, but it works.
--
Sean Eckton
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Internet: ecktons@sirius.byu.edu Brigham Young University, Provo, UT.
Packet Radio: kd6bik @ wb7esh.#orem.ut.usa.na
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
------------------------------
From: db1@ukc.ac.uk (D.Bolla)
Subject: More questions about Linux :-)
Date: 15 Feb 92 16:07:44 GMT
Hello !
Few more questions about Linux :-)
1) I have seen the draft 2 for the filesystem it seems fairly complete to
me and I think it is time for it to be published. When will this
happen ?
2) As it is now there is no /vmlinux file in the root directory. The
standard vay to boot is to boot all the kernel from floppy. I haven't
tried shoelace... so I can't know if that is the solution. What I would
like is to have a SMALL boot code in the floppy that loads /vmlinux and
execute it. Is this possible ? Will it be done in the next release ?
3) Related to the above question there is the problem of the ps command
that now uses an ioctl to do it's job. This is not very standard.
Usually what ps does is to use the symbols in /vmlinux to look in
/dev/kmem for the right stuff. Of course to have a ps like this we need
a /vmlinux and a /dev/kmem working. Will it be done ?
4) I just bought a book about UNIX system V system calls (Release 4 ) it is
quite useful since it describes the error codes and when they happens
( In additions to the availables syscalls in AT&T Unix )
and it is fairly complete. I heard thet SYS V r 4 is POSIX compatible.
Is this true ? . If it is compatible will be possible to try to follow
part of this book ?
BTW SunOs is going to be SysVr4 with Solaris .....
( If anybody is interested I will email the books details )
5) I am writting the format program. It will allow you to examine the
partitions of DOS Hd and change them. I started the coding.. but I would
like to know if somebody has already done it :-)
Damiano
------------------------------
From: poe@daimi.aau.dk (Peter Orbaek)
Subject: Re: who can't open utmp
Date: 15 Feb 92 14:24:14 GMT
zuazaga@ucunix.san.uc.edu (Humberto Ortiz-Zuazaga) writes:
>I've got the mylogin3 package up and running on my machine, and it
>works. Unfortunately, along with the problems with make and such already
>reported, I've run into a problem with the program who, included in the
>mylogin3 package.
>If I'm logged in as other than root, who fails, reporting
>_setutent: Can't open utmp: not owner
>Presumably meaning I'm not the owner. I'm not so hot at sysadmin, but
>does this have to do with setuid on the binaries? Isn't who supposed to
>be able to read utmp no matter whom called it? Whats the fix?
This is caused by an error on my part. The utmp2.c routines try to open
/etc/utmp as O_RDWR, even when not running as root. It's been fixed in
my new and improved :-) package: poeigl.tar.Z available from ftp.daimi.aau.dk
See separate announcement...
BTW. permissions on /etc/utmp and /etc/wtmp should be -rw-r--r--, we dont'
want to make the same mistake as Sun, making utmp world writable.
- Peter.
--
Peter Orbaek ----------------- poe@daimi.aau.dk | ///
Hasle Ringvej 122, DK-8200 Aarhus N, DENMARK | ///
| \\\///
"Strong typing is for people with weak memories" | \XX/
------------------------------
From: poe@daimi.aau.dk (Peter Orbaek)
Subject: init/getty/login announcement
Date: 15 Feb 92 14:28:13 GMT
I have just ftp'ed my package of init/getty/login for Linux to:
ftp.daimi.aau.dk file: /pub/Linux-source/poeigl.tar.Z
nic.funet.fi file: /pub/OS/Linux/incoming/poeigl.tar.Z
There are actually 2 inits, namely a hacked version of qpl-init, but it
doesn't work very well, and a more SysV like init that I wrote myself. It
works very well for me.
The getty program is a port of agetty posted to comp.sources.misc.
The login program is a port of BSD login first to HPUX then to Linux.
For init to work really well you should patch your kernel to run init as
process 1 instead of /etc/rc. But you can test it without that patch.
There's even a README file this time. :-)
The package still contains the complete UFC code, plus who, last and hostname
utils.
Try it, and let me know what you think.
I hope this will be in 0.13...
As an aside: I wrote about a bug in the kernel concering the check in
check_char_dev() in fs/open.c of tty->session == 0. As an experiment I removed
that check and built a new kernel, and the problems disappeared!
I doubt that this is a good way to solve it though.
- Peter (poe@daimi.aau.dk)
--
Peter Orbaek ----------------- poe@daimi.aau.dk | ///
Hasle Ringvej 122, DK-8200 Aarhus N, DENMARK | ///
| \\\///
"Strong typing is for people with weak memories" | \XX/
------------------------------
From: tytso@ATHENA.MIT.EDU (Theodore Ts'o)
Subject: ["Mark W. Eichin": update sources (complete releases?)]
Reply-To: tytso@athena.mit.edu
Date: Sat, 15 Feb 1992 17:52:52 GMT
This was forwarded to the wrong group because the
Linux-Activists@news-digests.mit.edu gateway was goofed up. My
apologies....
- Ted
======= Forwarded Message
To: Perl-Users@fuggles.acc.Virginia.EDU
From: eichin@ATHENA.MIT.EDU ("Mark W. Eichin")
Subject: update sources (complete releases?)
Reply-To: eichin@ATHENA.MIT.EDU ("Mark W. Eichin")
Date: Fri, 14 Feb 1992 23:54:43 GMT
I've sucked over just about everything on tsx-11.mit.edu:pub/linux and
one thing I noticed was that I couldn't find sources to update... I
understand not releasing all of the gnu sources that we could get
elsewhere, but I'd expect to find something like update as part of the
tree. Did I miss it? Anyone know where it is hiding? (Am I confused
for some other reason? :-)
In general, it would be nice to know that we had either source
or pointers to source for everything in the release (just to rebuild
it all, if nothing else.)
_Mark_ <eichin@athena.mit.edu>
MIT Student Information Processing Board
ps. Success story - I've got a 486/40mhz (AMI), 210M ide drive, 16M
ram, ET4000 SVGA/1Meg, and linux worked first try, I'm even doing
kernel builds now.
======= End Forwarded Message
------------------------------
From: tytso@ATHENA.MIT.EDU (Theodore Ts'o)
Subject: [Theodore Ts'o: [Steven Wilson: rogue.tar.Z]]
Reply-To: tytso@athena.mit.edu
Date: Sat, 15 Feb 1992 17:53:26 GMT
Also forwarded to the wrong news group due to a bug in the news/mail
gateway configuration....
- Ted
======= Forwarded Message
To: Perl-Users@fuggles.acc.Virginia.EDU
From: tytso@ATHENA.MIT.EDU (Theodore Ts'o)
Subject: [Steven Wilson: rogue.tar.Z]
Reply-To: tytso@athena.mit.edu
Date: Sat, 15 Feb 1992 01:37:51 GMT
In ~ftp/pub/linux/sources/usr.games/rogue.tar.Z
- Ted
======= Forwarded Message
Date: Fri, 14 Feb 92 15:55:14 PST
From: stevew@netcom.com (Steven Wilson)
To: tytso@MIT.EDU
Subject: rogue.tar.Z
I uploaded the file rogue.tar.Z to tsx-11.mit.edu. The name
should be explanatory. I ported the PD version of rogue to
LINUX over the last couple evenings...figured it was a good
way to learn a little about POSIX along the way...anyway,
if you could move this to the public directory I'd appreciate
it.
Steve Wilson stevew@netcom.com
======= End Forwarded Message
======= End Forwarded Message
------------------------------
From: stevew@netcom.COM (Steven Wilson)
Subject: Re: Help! Can't boot!
Date: 15 Feb 92 17:52:40 GMT
I've got Linux up and running and have been busy trying to
port the various gnu software packages, et al that equate to the basic
command set under unix. I haven't run across od yet... and
Linus mentioned using an od in a recent posting. Where
might the sources be located? As a further aside, has anyone
got mtools accessing their Hard disk. I've got to admit I'm a
bit chicken ;-) I've got two disks, one which is a full Dos partition
and the other with two parititions running with Linux. The DOS
partition is a 4.0 formated file system...is mtools up to this?
Steve Wilson
------------------------------
From: hedrick@dumas.rutgers.edu (Charles Hedrick)
Subject: mtools source and binary available
Date: 15 Feb 92 19:35:57 GMT
I've just put my copy of mtools on athos.rutgers.edu:/pub/linux. This
is reflected onto banjo.concert.net, which would probably be a better
place to get it. mtoolbin.tar.Z is the binaries and formatted man
pages, mtoolsrc.tar.Z is the sources. This is based on the most
recent mtools I could find, which supports large DOS disk partitions.
I've modified it slightly, to allow a Unix filename of "-" to indicate
standard in or standard out in mwrite and mread. This allows things
like my backup script, which does a tar piped into mwrite.
You may need to modify devices.c to match your system's floppy setup.
It's currently set so that a: is /dev/at0 and b: is /dev/PS1, I think.
(I confess I'm not entirely sure how that code works, as I only use
this software with the hard disk.)
I've tested the code carefully with hard disk files, and not at all
with floppy files.
------------------------------
From: hedrick@dumas.rutgers.edu (Charles Hedrick)
Subject: Re: More questions about Linux :-)
Date: 15 Feb 92 19:55:28 GMT
>3) Related to the above question there is the problem of the ps command
> that now uses an ioctl to do it's job. This is not very standard.
> Usually what ps does is to use the symbols in /vmlinux to look in
> /dev/kmem for the right stuff. Of course to have a ps like this we need
> a /vmlinux and a /dev/kmem working. Will it be done ?
I think many people now believe it's a mistake to have programs
reading kernel memory. This makes programs highly dependent upon the
structure of the kernel. E.g. under SunOS, whenever you get a new
version of the kernel, you have to change ps and the sharable
libraries on which it depends. In a new system, I'd strongly
recommend using some method other than /dev/kmem for programs like ps.
Preferably it should be a system call that you can use to ask for
specific data. This allows kernel data structures and variable names
to be changed without affecting programs, since the system call can
reformat the data. In my view, /dev/kmem is a remnant of the early
Unix philosophy that almost any hack is acceptable if it avoids the
need for adding something to the kernel. This makes sense on
PDP-11's, but on larger systems I'm willing to accept a bit of grwoth
in the kernel to avoid hacks like programs that have to understand the
kernel data structures. Please, let's stamp out /dev/kmem and any
other dependencies on kernel data structures in Linux.
I'm not sure I'd use an ioctl for this. I suggest adding a system
call specifically for getting information from the kernel. I'd
suggest looking at the TOPS-20 calls as a model. I'm a bit hazy on
details, but as I recall, you passed the system call a list of
descriptors saying what you wanted. Each descriptor was a table
number or name, an offset into the table, a location to put it, and
the amount of memory you had available at that offset. The table and
offset was because often you needed lists, e.g. lists of jobs, but
sometimes you might only be interested in one item. In the first case
you'd give an offset of zero (or -1, meaning the whole thing?) and
pass a large data area. In the second case you'd give the offset of
the job you wanted, and pass a small data area. By passing more than
one thing in a single system call, you allow the kernel to guarantee
that you get consistent information, something not possible by reading
/dev/kmem.
I think there have been other proposals in the Unix context, e.g.
/proc. If they also give independence from kernel data structures,
that's fine with me.
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: Linux-Activists-Request@NEWS-DIGESTS.MIT.EDU
You can send mail to the entire list (and alt.os.linux) via:
Internet: Linux-Activsts@NEWS-DIGESTS.MIT.EDU
Linux may be obtained via one of these FTP sites:
nic.funet.fi pub/OS/Linux
tsx-11.mit.edu pub/linux
tupac-amaru.informatik.rwth-aachen.de pub/msdos/replace
The current version of Linux is 0.12, released on Jan 14, 1992
End of Linux-Activsts Digest
******************************