Dec 262017
 
Linux news from the Internet.
File INTLX023.ZIP from The Programmer’s Corner in
Category UNIX Files
Linux news from the Internet.
File Name File Size Zip Size Zip Type
INTLX023.TXT 254882 81500 deflated

Download File INTLX023.ZIP Here

Contents of the INTLX023.TXT file


From: [email protected]

Subject: 0.96 (boot) and mcc-interim, shell problem help.
Keywords: 0.96(boot), mmc-interim, shell level
Message-ID: <[email protected]>
Date: 13 May 92 21:20:18 GMT
References:
Sender: [email protected]
Followup-To: same as above
Distribution: na
Organization: Michigan State University
Lines: 22


I am having a strange problem with 0.96 linux-boot image
and the mcc-interim root and other binaries.

In the login shell the command
ls -l /dev | more
does nothing prompt comes back.

If a subshell is spawned using the command
sh
the command
ls -l /dev | more
works as expected.

BTW, the following commands
cat file | more and
pr file | more
work as promised. Any pointer to the problem will be
greatly appreciated.

With thanks,
- ishwar ([email protected])


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

Subject: Re: Security and "X" don't mix?
Keywords: X11, security, bugs
Message-ID:
Date: 13 May 92 21:38:47 GMT
References: <[email protected]> <[email protected]
>
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 31

>Phew, now that finals are done, I have some time. Here is the program
>that the author of sxterm (Nevo Hed) previously mentioned. My
>program, xkey, will watch all the key presses made on a given display.
>As long as you have an account on the same machine as that person and
>they xhost that machine, they are vulnerable (assuming that they
>aren't using any of the MIT magic cookie stuff).

I strongly urge you to use xauth (MIT magic cookie) if you're using X
in a multi-user environment. Under Linux it's probably irrelevant.
At Rutgers, users start X using a script, and that script
automatically sets up an .Xauthority file for them. To start
additional windows, they use another script that rsh's to a machine,
sets up an appropriate environment, and starts an X utility. Part of
what it does is transfer the appropriate information from .Xauthority
and make sure it is in .Xauthority at the other end. There are still
some holes, but exploiting them requires root access on the machine or
something equivalent. While individuals can protect themselves using
something like this, in practice setting it up is hard enough that I
think system managers really need to do it. Certainly none of our
users used xauth until we produced the scripts and made sure that all
of our instructions, default startup files, documentation, etc. used
them. We do not document xhost in introductory user documentation,
nor do we suggest an any way that use of xauth is optional. We find
that users generally will not take any trouble to protect themselves
from security problems, but once they get bitten they're upset.

While I generally believe in open discussions of security issues,
posting code to break system security is probably a bad idea. If
someone uses this program to cause damage, I think a lot of people
would consider both the author and poster of the program at least
partly liable.


[next article]
From: [email protected] (Jay Maynard)

Subject: Re: Security and "X" don't mix?
Keywords: X11, security, bugs
Message-ID: <[email protected]>
Date: 13 May 92 22:17:38 GMT
References: <[email protected]> <[email protected]
>
Sender: [email protected]
Organization: UT Health Science Center Houston
Lines: 16
Nntp-Posting-Host: oac.hsc.uth.tmc.edu

In article [email protected]
du (Charles Hedrick) writes:
>While I generally believe in open discussions of security issues,
>posting code to break system security is probably a bad idea. If
>someone uses this program to cause damage, I think a lot of people
>would consider both the author and poster of the program at least
>partly liable.

They just had this discussion in comp.sources.d about a Unix password cracking
program. The consensus is that any such program should be widely disseminated;
if the bad guys don't have it already, they can either develop it themselves
or get it from places that cater to bad guys.
--
Jay Maynard, EMT-P, K5ZC, PP-ASEL | Never ascribe to malice that which can
[email protected] | adequately be explained by a .sig virus.
"[...] have you noticed how many people have joined you on the back of
Rosinante to help subdue this particular windmill?" -- Dan Herrick


[next article]
From: [email protected] (S3679988)

Subject: Can't open display (X problem)
Message-ID: <[email protected]>
Date: 13 May 92 22:12:28 GMT
Sender: [email protected] (USENET News System)
Organization: University of Massachusetts at Amherst
Lines: 19
Nntp-Posting-Host: deimos.ucc.umass.edu
X-News-Reader: VMS NEWS 1.20

I have finally given up on my X windows problem, here is the scoop, if anyone
out there has any better ideas:

my Xconfig is set up for 640x480 mode, 25mhz clock. I am trying to
run on an AHEAD chipset SVGA board with 512K of RAM. According
to its manual, VGA mode (which i run it in since my monitor is VGA)
is compatible with the IBM standard.

when i try to start x (with startx) my screen blanks out, and i lose
ALL of my virtual consoles. With my terminal, when i type ps -axl,
there is no process running off of any of the virtual consoles, nor is
there an x -- or any other process of interest running.

What is interesting is that when i start X up from my terminal (with
the DISPLAY defined, and exported to my environment), i get the message
that it cannot open my DISPLAY up. I am running everything from root,
with teh sticky bit set on _ALL_ X stuff.

- craig hagan


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

Subject: Upgrading from 0.95a to 0.96
Message-ID: <[email protected]>
Date: 13 May 92 23:03:10 GMT
Sender: [email protected] (USENET News system)
Organization: University of Maryland, College Park
Lines: 10
Nntp-Posting-Host: next.wam.umd.edu

I'm about to update to 0.96, but have a few questions. I understand
the new kernel needs shared libs. Does that mean that I cannot just
replace the old kernel with the new? How much do I have to replace?
Do I have to upgrade from GCC 1.40 to 2.1? What about (non-system)
binaries? Do they have to be replaced/recompiled?

Thanks in advance!

-Joel
([email protected])


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

Subject: Re: Linux rebooting
Message-ID: <[email protected]>
Date: 13 May 92 23:04:30 GMT
References: <[email protected]> <[email protected]
rado.edu> <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 19
Nntp-Posting-Host: ophelia.cs.colorado.edu

In article <[email protected]> [email protected] (Hamish Macdonald
) writes:
>>>>>> On 13 May 92 17:53:09 GMT,
>>>>>> In message <[email protected]>,
>>>>>> [email protected] (Drew Eckhardt) wrote:
>
>Drew> It is perfectly logical to use an unitialized global when doing
>Drew> user code, because it will be in BSS which is zero filled.
>Drew> However, in the kernel, anything that depends on this will
>Drew> break.
>
>Why doesn't the kernel zero out the BSS as the first thing it does
>before doing any real work?

It's really sort of kludgy - the bootstrap loader loads X bytes
(512K?) off of the disk when it boots up, period, irregardless
of kernel size, and moves it to absolute address 0.

Since the bootstrap loader does not load a normal a.out header file,
like BSD's does, it doesn't know the size / location of BSS.


[next article]
From: [email protected] (A. V. Le Blanc)

Subject: Re: compiling help
Message-ID: <[email protected]>
Date: 13 May 92 06:51:54 GMT
References: <[email protected]>
Followup-To: [email protected]
Organization: Computing Centre, University of Manchester
Lines: 52

In article <[email protected]> [email protected] (Derrick C
. Cole) writes:
>I am obviously not doing something I should be doing. I have installed gcc2.1,
>rebuilt the kernel (0.95c+), booted off the hard disk, and am now merely trying
>to compile programs for system use, namely binutils-1.94, shellutils-1.6,
>fileutils-3.2 and textutils-1.3 (this while anticipating the 0.96 release,
>after which I'l tackle X!)

>Should I be flamed for thinking these packages would compile right out of the
>tar file? I have had 0% success doing this (read as not knowing what I'm doing
>;-)). Can someone point me in the right direction to proceed? Are there
>patches, standard things to change for linux, etc?

The patches which I used to compile the shellutils, fileutils, and textutils
for the MCC 'interim' release of 0.95c+, as well as the patches I used for
bash, compress, groff, and tar, are available by anonymous ftp from
ftp.mcc.ac.uk [130.88.200.7] in /pub/linux/mcc-interim/0.95c+/patches.
The names are a bit compressed because they come directly from my linux
system; e.g., fileut32.ud.Z. These files are mirrored at banjo.concert.net
(in /pub/Linux/mirrors/uk.mirror/mcc-interim/...). On my system, and not
because of the recent discussion, the sources for all of these things are
available as well, for example, via symbolic links from
/pub/linux/mcc-interim/0.95c+/links-to-sources/.

The patches are all unified diffs files, so they assume the availability
of the GNU version of Larry Walls's patch command. The first line in
each patch file is a cryptic instruction for applying the patch; for example

mv textutils-1.3 tu-linux;(cd tu-linux;./configure);patch -u -p0
(In the original the 'tu's' are variously expanded.) This assumes that
you have untarred the source, creating a directory named textutils-1.3,
that you are in the parent directory of tu..., and that running configure
on your system produces the same result that it did on mine. Mine is
precisely as installing the MCC 'interim' 0.95c+ version left it, with no
additions. All my patches assume gcc 2.1. I think all of them also
produce (when applied) a file named 'a' which, after doing a make, links
the compiled bits into a library independent '.a' file. Incidentally,
the following '.a' files are available by anonymous ftp from
.../mcc-interim/0.95c+/files.a: bash, fileutils 3.2, groff 1.05,
shellutils 1.6, textutils 1.3, and miscellaneous Linux-specific binaries.
All of these (except groff) were distributed linked with the shared
libraries in the MCC 'interim' release 0.95c+. Groff was not distributed
because I intended to put it on the 'man' floppy, and the version of
man (man 1.0) does not handle preformatted stuff properly (If you have
/usr/man/man1/xx.1 and /usr/man/cat1/xx.1.Z, it runs groff, but it doesn't
(and shouldn't) if either man1/xx.1 is missing or cat1/xx.1 is present.)
and I didn't have time to fix it.

I hope some of this stuff is useful to some of you.

-- Owen
[email protected]


[next article]
From: [email protected] (A. V. Le Blanc)

Subject: Re: How 'bout that interim releas?
Message-ID: <[email protected]>
Date: 13 May 92 08:02:11 GMT
References: <[email protected]>
Followup-To: [email protected]
Organization: Computing Centre, University of Manchester
Lines: 30

In article <[email protected]> [email protected] writes:
>Just wondering... any idea when the MCC interim release of Linux 0.96
> will be ready? I was really impressed with
> the ease with which the 0.95c+ installed.

There does seem to be quite a lot of interest! The situation is
complicated by my going away for two weeks on this Friday, and by my
wishing to include the new release of 2.1, with all its corrections
and new features. So, I am sorry to say, the next 'interim' release
will probably not be out until early June.

The ABC release should probably be out soon as well, and I hope it
will be at least as easy to install as mine -- probably better, since
others have been able to learn from my mistakes. I'll try to learn
too, thanks to the many people who have spotted and reported problems
and made suggestions. I'll try to improve my documentation as well.
Quite a few people have written asking questions whose answers ought
to be in my README files (and some have had questions whose answers
are already in my README files).

I have enjoyed hearing from so many people, and have tried to write
back to all of them, but I'm afraid one or two letters have got
lost. One thing I'd like to do for the next 'interim' version is
to have an acknowledgments file, giving the names of all of the
people who have helped in one way or another. I should mention
especially Linus Torvalds, Theodore T'so, and Alan Clegg in
particular for putting up with a lot of e-mail and questions.

-- Owen
[email protected]


[next article]
From: [email protected] (A. V. Le Blanc)

Subject: Re: Sources (IMPORTANT to managers of LINUX ftp sites)
Message-ID: <[email protected]>
Date: 13 May 92 08:11:31 GMT
References: <[email protected]>
Followup-To: [email protected]
Organization: Computing Centre, University of Manchester
Lines: 15

In article <[email protected]> [email protected] ("Mark W
. Eichin") writes:
> Correct. The GPL says nothing about the *diffs themselves*; if
>the sites were only distributing patches, then the diffs would be
>sufficient. The problem is that binaries aren't made from the diffs,
>they're made from the main sources plus the diffs, and are thus
>covered by the GPL. One intent of the GPL is that you can find the
>sources to anything you use, and the person providing the binaries is
>responsible for making sure you can continue to get the sources.

I might mention that, at least for gcc, the FSF have been very careful
to include diffs which allow you to produce earlier versions of gcc
from the current source. This could be useful for other packages as well.

-- Owen
[email protected]


[next article]
From: [email protected] (Peter MacDonald)

Subject: What does "hiclock" really give you?
Message-ID: <[email protected]>
Date: 13 May 92 22:32:19 GMT
Sender: [email protected]
Organization: University of Victoria, Victoria, BC, CANADA
Lines: 23
Nntp-Posting-Host: sanjuan.uvic.ca


As many pointed out in my last note, it's
Vendor "hiclock"
not hicolor. But still, my clocks are reported to be about half of what
they should be. My dilema is this:

Without "hiclock" I found a "1024x768" mode (interlaced) that works on
my older monitor. With "hiclock" the normal "640x480" and "800x600"
modes work. But nary the twain shall meet.

What does "hiclock" really do for your. What I really want to do is
be able to switch to low res when using text, and high res with
graphics programs. This because scrolling is not speedy on my
older 386/25.

According to my manual, the modes used under DOS some clock timings
from the halved clocks reported by 'X', and some from the normal values.
ie: X says

25 28 36 30 23 20 32 22

or something.
while the clocks reported by the dmode program are double this.


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

Subject: Re: Security and "X" don't mix?
Keywords: X11, security, bugs
Message-ID: <[email protected]>
Date: 13 May 92 23:30:04 GMT
References: <[email protected]> tgers.edu> <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 40
Nntp-Posting-Host: ophelia.cs.colorado.edu

In article <[email protected]> [email protected] (Jay Maynard) writes:
>In article [email protected]
edu (Charles Hedrick) writes:
>>While I generally believe in open discussions of security issues,
>>posting code to break system security is probably a bad idea. If
>>someone uses this program to cause damage, I think a lot of people
>>would consider both the author and poster of the program at least
>>partly liable.

>They just had this discussion in comp.sources.d about a Unix password cracking
>program. The consensus is that any such program should be widely disseminated;
>if the bad guys don't have it already, they can either develop it themselves
>or get it from places that cater to bad guys.

There was a long thread about this under one of the comp.unix groups about
a month ago. The general consensus was that it was a good thing because it
increased awareness to the problem, and that acting on this new information,
sysadmins could protect themselves against what they hadn't known to be
a threat before. As far as any one who didn't act on this information,
that was their problem.

BTW : At our site, we run crack (just like anyone can), and the last
time it ran it caught ~150 passwords. The owners of those
accounts were mailed, and told to change them, which
could concievably protect us from a future attack by the
"bad guys"

BTW2 : Most of our machines also run a modified Perl COPS to spot problems with
suid programs, .rhosts, open home directory permissions,
changes in binary file checksums, etc. You should too
if your system supports setuid shell scripts - there is a whole in
most systems' kernel exec() code that is big enough to
drive a mac truck through.

BTW3 : Most modern systems use shadow passwords, making crack worth
less. There is still the getpwent(3) hole, which will let
users with access to your systems crack passwords if getpwent()
will return an encrypted password. BSD 4.3 Reno and later
will only return an offset into the /etc/master.passwd file,
which is not readable by normal users, eliminating this problem.



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

Subject: Re: Plenty o'Questions from a Clueless Newbie
Message-ID: <[email protected]>
Date: 13 May 92 23:34:39 GMT
References: <[email protected]>
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 43
Nntp-Posting-Host: ophelia.cs.colorado.edu

In article <[email protected]> [email protected] (Harrison Page)
writes:
>I took the time to grep through the FAQ and a well-written Beginner's Guide
>to Linux. I still end up with some leftover questions. My experience with
>Linux has been fantastic so far - I've been able to install everything on
>the first try with no major problems. Last night I got a terminal program
>called xc running with no problems and used zmodem to download all the
>gcc code. I also managed to get vile working as my editor of choice.
>
>My first project will be to port Waffle BBS to Linux.
>
>Questions, questions..
>
>1. I have two 40m ST-251's - drive 0 is running MS/DOS and drive 1 holds
> Linux. I access it by booting a patched boot image - done with pboot.exe
> from DOS. Since I only have a 4m 386, I'd like to configure Linux to use
> virtual memory. However, I partitioned *all* of drive 1 to use Linux.
> The swapping documentation says that a partition and a swap file are
> required - can I get around this and swap on the same partition?

Yes. This SHOULD be a FAQ. The basic process is to dd from a
"source" such as /dev/hd0 or a program 'yes very_long_foo' with
count * bs = the number of bytes you want, and then to do a mkswap
on that.


>
>2. Where can I find source or binaries for "who", "w", and "ps"? I poked
> around on tsx-11 and didn't see anything.

In the ps-0.95c patches.

>3. When I login as root, a number of environment variables are already
> pre-set, such as PATH. Where do these values get initialized? The
> root's home directory seems to be /, so I suppose I could change these
> values by assigning changing it to /usr/root and making a .login, but
> now I'm just curious.

Root's .profile for some, others are set by init.

>4. Has anybody managed to get Linux to accept outside connections through
> a modem?

yes.


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

Subject: Abysmal disk performance
Message-ID: <[email protected]>
Date: 13 May 92 23:44:57 GMT
Sender: [email protected] (The Daily Planet)
Organization: University of Colorado at Boulder
Lines: 10
Nntp-Posting-Host: ophelia.cs.colorado.edu

The string instructions (ie portread / portwrite in hd.c) are abysmally slow
in comparison to DMA. The same can be said for strcpy(). We should
modify the hd.c code, etc to use DMA to a track sized buffer, and DMA in
memory to memory mode to copy to the identity mapped buffercache.

The same can be used to copy pages where both pages < 16M.

I will start on this code after we get into our new apartment, and have
the data sheets out, or someone mails me a reference on the 8237's in
the AT.


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

Subject: Re: Plenty o'Questions from a Clueless Newbie
Message-ID: <[email protected]>
Date: 13 May 92 23:01:21 GMT
References: <[email protected]>
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, WIBSTR
Lines: 28

In article <[email protected]> [email protected] (Harrison Page)
writes:
>3. When I login as root, a number of environment variables are already
> pre-set, such as PATH. Where do these values get initialized? The
> root's home directory seems to be /, so I suppose I could change these
> values by assigning changing it to /usr/root and making a .login, but
> now I'm just curious.

root's $HOME should be the root directory, hence the name. ๐Ÿ™‚ The
'recommended' installation is to have at least two partitions - the
root partition, and a /usr partition.

Now, if your /usr partition becomes corrupted for some odd reason, and
root's home directory is /usr/root, you could run into problems.

I would strongly recommend keeping the root account configured as is
(/bin/sh as shell, / as $HOME, and add a password), and create another
account to do your work, and 'su' to root to do anything that requires
root privileges.

(There is an su program on tsx-11 someplace - poke around there to
find it.)

Oh, and $HOME and $SHELL are initialized from the /etc/passwd file.
--
+ Jim Winstead Jr. (CSci '95)
| Harvey Mudd College
| [email protected]
+ This is all my words. Honest!


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

Subject: Re: Oh oh, problem with .96 bootimage
Message-ID: <[email protected]>
Date: 13 May 92 23:29:14 GMT
References: <[email protected]> <23364
@dog.ee.lbl.gov>
Organization: Security Pacific Automation Co., LA, CA
Lines: 22

In article <[email protected]>, [email protected] (Robert J. McNamara) writes:
> In article <[email protected]> [email protected] writes:
>>
>>I'm using the 0.96 image without any problems... in fact, a large number of
>>problems I had with pre-0.96 are now gone. I don't have any SCSI devices,
>>though, so that might be the difference: the point at which you guys report
>>0.96 rebooting is when it is initializing and checking SCSI devices.
>>
>
> Nope, I've got an IDE drive and I've got the same thing happening.
> my system: 386/33 w/ OPTI chipset
> AMI bios 4/9/90

> Prodesigner II (4.01 vbios)
> Maxtor LXT200A
> Soundblaster
> is anyone else NOT having this reboot problem with 0.96?
>
Running .. no prob. on 386/40, IDE, AMI bios, Trident SVGA ...
will try tomorrow on 386sx/16, IDE, AMI bios, RealTech ?VGA .. ;( {home}
--
----> [email protected] OR [email protected]
David L.


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

Subject: Re: rc Shell Under Linux
Keywords: rc linux
Message-ID: <[email protected]>
Date: 13 May 92 23:14:40 GMT
References: <[email protected]> <[email protected]
.claremont.edu> <[email protected]>
Sender: [email protected] (The News System)
Followup-To: comp.os.linux
Organization: Harvey Mudd College, WIBSTR
Lines: 40

In article <[email protected]> [email protected] (The system o
perator) writes:
>I've read the Plan 9 paper about 'rc', and obtained the clone
>sources, but I do 'way too much Bourne shell work on my machine
>to make a change. What's your experience with rc under linux?

I have not worked with 'rc' under Linux extensively, although I use it
exclusively on my school's machine. I don't use it under Linux only
because I haven't had the time to set everything up - I've compiled it
successfully and done some limited testing and it seemed to work fine.

Part of the reason 'rc' is not on the standard root image is your
comment above (or the general idea behind the comment, not that
comment specifically :). People are far too used to the standard
Bourne shell to justify putting 'rc' as the only shell on the root
disk.

>Why wouldn't you put rc onto the linux boot disk, at least along
>with ash?

Here's at least one good reason: space. There's currently only 50KB
free on the root disk, and rc compiles to about 75KB or so. (Smaller,
I think, but I'm being conservative. ๐Ÿ™‚


This might change when I convert the root disk over to shared
libraries, which I expect to free up a good deal of space. In that
case, both 'ash' and 'rc' will appear on the root disk, with most of
the install stuff being written in 'rc' scripts because of it's vastly
superior syntax.

>What I like, from reading the paper, is that it's predictable.

Yes, 'rc' is easily the best shell I've ever seen.

Followup-To is set to comp.os.linux to redirect this to the right
newsgroup. ๐Ÿ™‚ (It's been crossposted there, too...)
--
+ Jim Winstead Jr. (CSci '95)
| Harvey Mudd College
| [email protected]
+ This is all my words. Honest!


[next article]
From: [email protected] (Kevin Brown)

Subject: Re: Oh oh, problem with .96 bootimage
Message-ID: <[email protected]>
Date: 14 May 92 02:21:24 GMT
References:
Sender: [email protected]
Reply-To: [email protected] (Kevin Brown)
Organization: Minimal.
Lines: 23

In article "Alex R.N. Wetmore" .cmu.edu> writes:
>Excerpts from netnews.comp.os.linux: 12-May-92 Oh oh, problem with .96
>boo.. by "Scott R. Grabowski"@and
>> It reboots after telling me that I have 8 virtual ttys (or whatever the
>> line is). I am guessing here that the problem is in the autodetect of
>> the SCSI card. I do not have any drives that Linux recognizes (since I
>> have an (actually, 2) 8 bit controllers for my MFM and RLL drives).
>
>I had the same problem, but cold booting (instead of cntrl-alt-del) from
>DOS seemed to fix it.

Seconded. Doing a hard reset, with the memory test, seems to have done the
trick.

However, this version *still* has the same problems that I described before
with respect to seeing SCSI target devices out there. Sigh...

>alex



--
Kevin Brown ([email protected])


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

Subject: X386, cant find server
Message-ID: <[email protected]>
Date: 13 May 92 23:09:37 GMT
Sender: [email protected] (Net Noise owner)
Organization: University of Illinois at Urbana
Lines: 17

Ok, I think other people have posted something similar, but after going
back over the old notes I did not see a solution.

I am useing the .96 version of linux. When starting up startx, I get

X386 Version..... etc

giving up.
xinit: Invalid argument (errno 22): unable to connect to X server
xinit: No such process (errno 3): Server error.

The other thing is that I cant get anything other than a blank screen
(I have the ATI 1024 card ?Wonder?) but perhaps that is because the server
never comes up. I did not apply the patches for sockets because I was under
the impression that that is one of the things the .96 release contained.

-Brian Chojnowski


[next article]
From: [email protected] (Derek Lieber)

Subject: Re: Linux rebooting
Message-ID: <[email protected]>
Date: 13 May 92 23:57:12 GMT
References: <[email protected]> <[email protected]
bnr.ca> <[email protected]>
Sender: [email protected] (NNTP News Poster)
Reply-To: [email protected] (Derek Lieber)
Organization: IBM T.J. Watson Research Center
Lines: 28
Disclaimer: This posting represents the poster's views, not necessarily those of
IBM
Nntp-Posting-Host: gumby.watson.ibm.com

In article <[email protected]> [email protected] (D
rew Eckhardt) writes:
>In article <[email protected]> [email protected] (Hamish Macdonal
d) writes:
>>>>>>> On 13 May 92 17:53:09 GMT,
>>>>>>> In message <[email protected]>,
>>>>>>> [email protected] (Drew Eckhardt) wrote:
>>
>>Drew> It is perfectly logical to use an unitialized global when doing
>>Drew> user code, because it will be in BSS which is zero filled.
>>Drew> However, in the kernel, anything that depends on this will
>>Drew> break.
>>
>>Why doesn't the kernel zero out the BSS as the first thing it does
>>before doing any real work?
>
>It's really sort of kludgy - the bootstrap loader loads X bytes
>(512K?) off of the disk when it boots up, period, irregardless
>of kernel size, and moves it to absolute address 0.
>
>Since the bootstrap loader does not load a normal a.out header file,
>like BSD's does, it doesn't know the size / location of BSS.
>

Ok, I'll bite. Why not zero out ALL memory (0-512K) before loading
the image? That's probably 3 instructions and fraction of a second.

--
Derek Lieber
[email protected]


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

Subject: Re: X finally up
Message-ID: <[email protected]>
Date: 13 May 92 23:17:14 GMT
References: <[email protected]> elsinki.FI>
Sender: [email protected]
Organization: Design Computing Unit, The University of Sydney
Lines: 48
Nntp-Posting-Host: dean.arch.su.oz.au

In article , [email protected]
.FI (Mika Pekka Liljeberg) writes:
|>
|>
|>
|> Nope. It's Vendor "hiclock".
|>
|> After this the clocks reported by clock.exe and X (less reliable) should
|> work. I had the same problem with my Boca Super X VGA with Sierra DACs.
|> Without the Vendor line the clocks seemed about double the ones reported.
|> I ended up with horrid resolutions like 1280x600 (blech!).
|>
|> I believe that doubling the clock values doesn't actually do anything.
|> I think that the purpose of the values is just to tie a certain mode
|> into a certain clock, numbered from 1 to 8 as displayed on the Clocks line.
|> The actual values just serve to remind the user about the frequency of
|> the actual clock (correct me, if I'm wrong).
|>
|> You can test this by changing the line Clocks to:
|>
|> Clocks 1 2 3 4 5 6 7 8
|>
|> ..and replacing the ModeDB clocks with these as appropriate.
|>
|> So, the most important thing is to know the proper clock frequencies AND
|> the order they are in and calculate syncs accordingly.
|>
|>
|>
|>
|>
|> Cheers!
|>
|> Mika
|>
|> --
|> Mika Liljeberg Email: [email protected]
|> Helsinki University [email protected]
|> Dept. of Computer Science

This works and was the most useful info I saw for getting X working on my
machine.

Thanks

--
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]

Subject: Modems that don't work
Message-ID: <[email protected]>
Date: 14 May 92 01:01:55 GMT
Sender: [email protected] (The News Administrator)
Organization: Rose-Hulman Institute of Technology
Lines: 23
Nntp-Posting-Host: manager.nextwork.rose-hulman.edu



In response to the post that I made earlier about modems that don't work, I
received 4 letters from people who's modems didn't work either. The only
thread that seems to run through all of the letters is possibly the type.

I beleive all of the problem modems were cheap modems.

Mine works fine under DOS, with lots of software that I have tried, but nothing

I or any of these other people could do makes these modems work under Linux.

One of the brands was Everex. Mine is Info(something).

Does anybody have any ideas? I can't afford a new modem--I don't have the
money.

--
Joseph T. Batt - deranged CS undergrad...
-----------------------------------------------------
\ [email protected] = &(Joseph T. Batt); \
\ \
\ Real programmers start counting with zero. \
------------------------------------------------------


[next article]
From: [email protected]

Subject: .96 ok-ish
Message-ID: <[email protected]>
Date: 14 May 92 02:41:21 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 11

I just got the .96 sources, applied the serial patch, set my rootdev and
keyboard in the Makefile, and did a make. 30 minutes later, i rebooted.
It did the run, press any key, reboot, run. I then did a reboot(so i could
read the messages that scroll by), and it ran fine. My next test will be a
cold power-on.
Equip: 386-33, SIS chipset, Boca ET4000AX, 4MB real, 21MB swap, 50MB root,
30MB /pub, 1.44, 1.2, IDE 100MB, NO SCSI...
Things are looking nice(ie, socket messages...)
Great Job!!!, Linus.
Jim
[email protected]


[next article]
From: [email protected] (Greg Lee)

Subject: chmod +x fm
Message-ID: <[email protected]>
Date: 14 May 92 02:59:07 GMT
Sender: [email protected] (News Service)
Organization: University of Hawaii
Lines: 6
Nntp-Posting-Host: uhunix.uhcc.hawaii.edu

The binary fm on tsx-11 in binaries/usr.bin/fm.tar.Z
(text formatting; nroff -man substitute) must be given
execute permission to run: chmod +x fm. Sorry about that.

--
Greg Lee


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

Subject: Re: Abysmal disk performance
Message-ID: <[email protected]>
Date: 14 May 92 01:00:44 GMT
References: <[email protected]>
Sender: [email protected] (NetNews)
Reply-To: [email protected] (Doug Thompson)
Organization: Novell, Inc.
Lines: 26
Nntp-Posting-Host: mobius.mobius.provo.novell.com

In article <[email protected]>,
[email protected] (Drew Eckhardt) writes:
> The string instructions (ie portread / portwrite in hd.c) are abysmally slow
> in comparison to DMA. The same can be said for strcpy(). We should
> modify the hd.c code, etc to use DMA to a track sized buffer, and DMA in
> memory to memory mode to copy to the identity mapped buffercache.
>
> The same can be used to copy pages where both pages < 16M.
>
> I will start on this code after we get into our new apartment, and have
> the data sheets out, or someone mails me a reference on the 8237's in
> the AT.

As I remember the 8237 DMA clicks along at 3 Mhz on the original AT
coupled with
10-15 wait states. I don't know how many wait states used on 386 boxes today
but I believe the DMA is still at 3Mhz. Starting with the 286 and beyond
doing a REP INS would transfer at bus speed.

The DMA circuits on the AT just could not provide very fast transfers, thus
the POLL input on the WD controller and childern. That is why Adaptec went
with the DMA master model. The 1542 would utilize the DMA handshake lines
itself and not use the on board 8237's.

I might be out of date, but thats what I remember from days working on 286s
and early 386s


[next article]
From: [email protected] (Tom Lightbody)

Subject: gcc 1.40 no 387: can we do math?
Message-ID: <[email protected]>
Date: 14 May 92 03:46:22 GMT
Sender: [email protected]
Organization: Computer Engineering and Science, Case Western Reserve University
Lines: 23
Nntp-Posting-Host: ida.ces.cwru.edu

linuxers,

if you are using 1.40 but lack a math coprocessor, tell me,
please, how you get a *functional* executable from pie.c below

the FAQ says

...if the program does any weird fp math (exp(), sin())
it'll die when you run it though....

tom ([email protected])

/* pie.c
* constants from Kernighan&Pike,p.234
* answers:
* 3.141593 (tc2.01 8086pc)
* 3.141593 (gcc1.40 sun3/60)
* 3.1415929203539823007 (c honeywell mainframe:-)
*/

#include
main(){float a,b;a=355.0;b=113.0;printf("%f\n",a/b);}


[next article]
From: [email protected] (goon)

Subject: Re: rc Shell Under Linux
Message-ID:
Date: 14 May 92 02:25:20 GMT
References: <[email protected]>
<[email protected]>
<[email protected]>
<[email protected]s.claremont.edu>
Sender: [email protected]
Followup-To: comp.os.linux
Organization: St. Olaf College, Northfield, MN USA
Lines: 18
In-Reply-To: [email protected]'s message of 13 May 92 23:14:40 GMT

In article <[email protected]> [email protected]
ont.edu (Jim Winstead Jr.) writes:
This might change when I convert the root disk over to shared
libraries, which I expect to free up a good deal of space. In that
case, both 'ash' and 'rc' will appear on the root disk, with most of
the install stuff being written in 'rc' scripts because of it's vastly
superior syntax.

Even if the root image is converted to shared libs, it is proabally a
good bet that the shell (whatever that may be) will be statically
linked because having a shared-library version of the shell requires
too much to be working for one to get a root prompt. Consider the
following nasty situations:

(1) the dynamic libs are on a different partition than the root
partition and this partition has been corrupted

(2) the dynamic-linker has had it's libs (or, on some systems cache)
corrupted.


[next article]
From: [email protected] (Court Demas)

Subject: X Problems - Help!
Message-ID:
Date: 14 May 92 03:12:53 GMT
Organization: Sophomore, Math/Computer Science, Carnegie Mellon, Pittsburgh, PA
Lines: 28


I've spent a lot of time in the past few days trying to get X to run on
my 486. At this point, after trying out 25-30 different Xconfig files,
computing all the clock frequencies, etc, etc, etc, that this may not be
a problem with the clocks or h/v frequencies or anything.

The best I could do at this point is bring up a 640x480 image which is
still on the screen. It's basically white vertical bars across the
screen, but at the top there's some garbage which looks like it could be
mangled (shifted) text. The image is steady and it looks like the mode
is set correctly - except that there's no cursor and no windows, etc.

What could be wrong? Playing around with the frequencies is NOT going
to change this. The picture is displayed squarely in the center of the
screen, just as a normal VGA picture should be. Could there be a
problem with the X server? With sockets? Something??!?

If there's absolutely anything else that could possibly cause this I'd
like to hear about it. I'm really anxious to get X up and running..


thanks!
-court



btw, I have a 486/33, 8mb, Tseng-4000 hicolor w/1mb (tried with
"hiclock" and without), Samsung multisync.


[next article]
From: [email protected] (Teresa Selling)

Subject: What does the order of the clocks do?
Message-ID: <[email protected]>
Date: 14 May 92 05:38:51 GMT
Sender: [email protected] (USENET News System)
Organization: Indiana University at South Bend
Lines: 4
Nntp-Posting-Host: peabody.iusb.indiana.edu

People have said it matters what order the clocks are in in the clocks
line for X. Why exactly does it matter?
--
Teresa Selling, IUSB, Indiana.


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

Subject: Re: rc Shell Under Linux
Message-ID: <[email protected]>
Date: 14 May 92 05:51:13 GMT
References: <[email protected]> <[email protected]
.claremont.edu>
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, WIBSTR
Lines: 17

In article [email protected]
du (goon) writes:
>(1) the dynamic libs are on a different partition than the root
> partition and this partition has been corrupted

This should *not* happen. /lib should be part of your root partition,
and even if you lose that - the root disk (floppy) would have it's own
copy of the shared libraries necessary for it's operation.

>(2) the dynamic-linker has had it's libs (or, on some systems cache)
> corrupted.

Dynamic linker? The libraries are shared, not dynamically linked.
--
+ Jim Winstead Jr. (CSci '95)
| Harvey Mudd College
| [email protected]
+ This is all my words. Honest!


[next article]
From: [email protected] (Wen-Chun Ni)

Subject: X library
Message-ID: <[email protected]>
Date: 14 May 92 05:46:39 GMT
Sender: [email protected]
Organization: Brown University Department of Computer Science
Lines: 20

I've seen people asking for the ftp site of libX*.a but no one
responded on the net. Would you please get it posted on the
net? I'm eager to port some programs and write some myself. Thanx!


BTW even after a couple of days since I successfully installed
the X11R5 on my machine, I am still very excited about it.
Partly because I just can't believe that the resource hog X
can be running on a 386/25 (no cache) so satisfactorily!


*BUG gone* When I use the Linux 0.96, the X window never wrap
like it had been using the bootimage supplied
or the testimage. Can any one figure out why?
There has got to be some bugs fixed in the final
version of Linux 0.96.
version.


Wen-Chun Ni


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

Subject: Documentation on shoelace
Message-ID:
Date: 14 May 92 06:38:44 GMT
Sender: [email protected] (Alex Liu)
Organization: None to Speak of
Lines: 481
NNTP-Posting-Host: aludra.usc.edu

This is posting as a request...

Since the Shoelace program ported for Linux did not come with any
documentation, a lot of peoples are missing out on certain shoelace
features.

Like "run file" command to execute DOS programs that only do BIOS calls.

And also, it is important to know that the shoelace fsck will only
work with old style minix filesystems (cyl<1024, etc)

Here are the docs.

----------------------------------------------------------------------


ShoeLace(8) Minix Programmers' Guide ShoeLace(8)



NAME
ShoeLace - a small bootstrap

DESCRIPTION
ShoeLace is a small bootstrap that replaces the original
awkward boot procedure. ShoeLace is not executable although
it does contain code that will be run at boot time.

BOOT PROCEDURE
When the volume is bootstrapped, ShoeLace will look for a
Minix image to load into memory. The strategy adopted is as
follows:

1 Search for the configuration file /etc/config. If the
configuration file is found, it is read otherwise the
default configuration is used.

2 Wait for a keystroke. If a keystroke is detected, the
character is discarded and steps 3-4 are skipped. Keys-
troke detection is abandoned if no keystroke is detected
after a small time.

3 Attempt to build a Minix system using separate system com-
ponents. If this succeeds, the newly built system is
started.

4 Search for a system image built using build(8). If such
an image is found, it is loaded and started.

5 All possibilities have been exhausted. The command inter-
preter is started.

When ShoeLace starts it will print a sign on message. Fol-
lowing this will be a summary of the hard disk partition
table if the bootstrap is being performed on a hard disk and
then a table showing where ShoeLace is loaded in memory.

BOOTING SYSTEM COMPONENTS
This is the new method of booting Minix systems. ShoeLace
mimics the functionality of build(8) and constructs and
loads a Minix system on the fly. Four components are
required to build the Minix system: the kernel, mm, fs and
init. Files with names /kernel, /mm, /fs and /init will be
sought on the boot volume. The names of the files can be
changed using the configuration file. The bootstrapped sys-
tem will be passed a pointer to the configuration parame-
ters. Old kernels will be oblivious to this and will boot as
usual. New kernels will be able to configure themselves
according to the boot parameters.

BOOTING PREBUILT SYSTEM IMAGES
Prebuilt system images are constructed using the program



Printed 12/6/89 Minix 1






ShoeLace(8) Minix Programmers' Guide ShoeLace(8)



build(8) with the output sent to a file. ShoeLace assumes
that this file will have the name /minix but this can be
changed using the configuration file. This is a clumsy way
of constructing a system and it is recommended that the sys-
tem files not be passed through build(8) since ShoeLace can
perform the same process at boot time. Furthermore, starting
Minix in this way causes the standalone fsck to be executed.
The `=' key must be pressed to make fsck start Minix.

The capability to load prebuilt system images provides some
backwards compatibility. In the future, when boot time
memory space becomes too valuable, this capability might be
removed.

CONFIGURATION FILE
The configuration file is named /etc/config. Text in this
file is scanned at boot time and is used to alter the
actions performed during the bootstrap.

Text within the configuration file is divided into lines.
Leading and trailing spaces are ignored. Empty lines and
lines beginning with `#' are ignored. Other lines are broken
up into a series of space separated words. The first word is
treated as the name of the command to invoke. All other
words are treated as arguments for the command.

The following commands are recognised:

boot name
Sets the name of the prebuilt Minix image. The default
name is minix.

kernel name
Sets the name of the file containing the Minix kernel
code. The default name is kernel.

mm name
Sets the name of the file containing the Minix memory
management code. The default name is mm.

fs name
Sets the name of the file containing the Minix file
system code. The default name is fs.

init name
Sets the name of the file containing the Minix init
process. The default name is init.

run file ...
Run the named files before attempting to boot Minix.





Printed 12/6/89 Minix 2






ShoeLace(8) Minix Programmers' Guide ShoeLace(8)



setenv name value
Set the named environment string to have the specified
value.

setdev name device
Set the named environment string to have the major and
minor device numbers of the specified device.

COMMAND LINE PROCESSOR
A small command line processor is built into ShoeLace. This
processor is started if all attempts to automatically start
a Minix system fail or if the key is pressed early at boot
time. The command line processor reads lines of input from
the console. Each line is broken into space separated words.
Empty lines are ignored. The first word names a command to
be executed. Subsequent words are used as arguments. The
following section lists the commands that are available.

help
Print a small summary of the commands which are avail-
able.

configure [-d] [file]
Alter the current configuration. If a file is speci-
fied, the alterations are taken from the named confi-
guration file, otherwise the alterations are performed
interactively. When the alterations are performed
interactively, responding with a single space or an
empty line will cause the current setting to be
retained. The switches have the following interpreta-
tions:

-d Reset the configuration to the default first.

display
Display the current configuration.

load [image]
Load a prebuilt Minix system and start it. If no name
is provided, the name is taken from the load setting in
the current configuration.

boot
Boot a Minix system by assembling all the components.
The components will be assembled according to the
current configuration setting. The relevant configura-
tion entries are kernel, mm, fs, and init. The current
environment is passed to the Minix kernel.

cat [file ...]
Print the contents of the named files on the console.




Printed 12/6/89 Minix 3






ShoeLace(8) Minix Programmers' Guide ShoeLace(8)



ls [directory ...]
List the contents of the named directories. If no
directories are specified, the contents of the root
directory on the boot volume are listed.

fsck [-acizrlsm]
Check the file system on the boot volume. The options
have the following meanings:

-a Automatically repair the file system and if ques-
tions are asked, provide the default response.

-c Ask for a list of inodes to be listed.

-i Ask for a list of inodes to be checked.

-z Ask for a list of zones to be checked.

-r Repair the file system if anything is found to be
wrong.

-l List the structure of the file system.

-s List the contents of the super block.

-m Make a file system. This option is capable of des-
troying the boot volume. You have been warned. It is
present to allow new floppy file systems to be made
at boot time just as in the original Minix boot pro-
cess.

printenv
Print the current environment strings.

run [file ...]
Load and run the named files.

setdev name device
Set an environment string to the major and minor
numbers of the named device.

setenv name value
Set the value of an environment string.

unsetenv name
Remove the named environment string.

DEVICE NAMES
The device names used with the setdev commands should be one
of /dev/hdn, /dev/fdn or /dev/atn, in which n is a small
number. The major and minor device numbers will be obtained
from the special files in the directory /dev on the boot



Printed 12/6/89 Minix 4






ShoeLace(8) Minix Programmers' Guide ShoeLace(8)



volume. If the a suitable special file is not available, a
guess will be made using the numbers compiled into the code
from shoelace.h. It is also possible to use the pseudo-
device name bootdev to specify the guessed major and minor
numbers for the boot volume.

ENVIRONMENT STRINGS
The kernel will be passed a set of environment strings.
ShoeLace knows about some important string names:

keyboard number
The keyboard scan code. With the old bootstrap, the
standalone fsck would prompt the user to start Minix by
pressing the `=' key. The resulting scan code would be
passed to the Minix console driver to set up the key-
board translation table. When ShoeLace boots Minix by
assembling the system components, fsck is no longer
run. The scan code is passed to Minix as the code for
the `=' key. The default value is 13.

processor number
Constrains the bootstrapped Minix system to run as if
it were using the specified processor. This option is
supported by Bruce Evans' kernel. Useful values are:

-1 Unconstrained processor.
386 Limit processor to 80386.
286 Limit processor to 80286.
186 Limit processor to 80186.
86 Limit processor to 8086.

rootdev device
Set the name of the root device that Minix will use.
The default is /dev/ram.

ramimagedev device
Set the name of the device on which the ram disk image
is stored. This name is used by the Minix to initialise
the ram disk if it is the root file system. The default
is /dev/fd0.

ramsize number
Set the size of the ram disk in blocks. This size will
only be used to allocate the ram disk if it is not
being used as the root device. If the ram disk is being
use as the root device, the size of the ram disk will
be set by the size of the ram disk image and the ram-
size command will have no effect. The default value is
zero.

RUNNING PROGRAMS
Programs run from either the configuration file or the



Printed 12/6/89 Minix 5






ShoeLace(8) Minix Programmers' Guide ShoeLace(8)



command line processor are loaded into memory at an arbi-
trary segment boundary. The files are assumed to be execut-
ables in the style of DOS TSR programs. Furthermore they
are assumed to be .com files stored with valid a.out header.
This option is currently present to support loading of Bruce
Evans' debugger. Future versions of ShoeLace will probably
incorporate a small DOS function emulator to make it easier
to use the same code for both DOS and ShoeLace. Only DOS
interrupt 0x21 is used. Other interrupts are not initialised
by ShoeLace. Split I&D programs may also be supported in
the future.

Programs exit using the DOS exit function or TSR exit func-
tion. ShoeLace will reclaim the memory allocated for the
program if termination is via the DOS exit function. Memory
will not be reclaimed if the program uses the TSR exit func-
tion. If the program uses the TSR exit function, ShoeLace
will examine the word at location 0:0x413. This is inter-
preted to be the amount of ram available (to the nearest
kb). If the program has not altered the value, ShoeLace will
reload the location to point at the base of the loaded pro-
gram. ShoeLace will then relocate to the top of the avail-
able memory. Thus programs can use 0:0x413 to determine
where the current top of memory is and can relocate them-
selves if necessary. Furthermore, they can reload 0:0x413 to
indicate how much memory.

LOADING STRATEGY
ShoeLace expects to be loaded into memory by BootLace and
will use the return address pushed onto the stack to locate
the BootLace corpse containing the boot parameter block and
disk code. This information is used by ShoeLace to access
the file system on the volume.

Low memory becomes crowded when Minix is loaded. For this
reason ShoeLace moves itself safely out of the way to the
very top of memory. The amount of memory available on the
system is determined by inspecting the word at location
0:0x413. ShoeLace relocates itself so that it occupies the
topmost 64kb of memory (aligned to a 64kb address boundary).
When running programs (such as the debugger) ShoeLace will
stack these programs starting at the top of memory. In order
to do this, ShoeLace will move itself down to make room.

FILES
/shoelace
/etc/laceup
/etc/bootlace
/etc/config
/minix
/kernel
/fs



Printed 12/6/89 Minix 6






ShoeLace(8) Minix Programmers' Guide ShoeLace(8)



/mm
/init

SEE ALSO
laceup(8), bootlace(8), build(8)


















































Printed 12/6/89 Minix 7




--
_____________________________________________________________________________
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] (Edward Wahl)

Subject: HELP! with Adaptec- scsi installation!
Keywords: crash and burn,bug, linux,scsi,adaptec,Adaptec1542B,Maxtor
Message-ID: <[email protected]>
Date: 14 May 92 06:45:46 GMT
Sender: [email protected] (NETnews )
Followup-To: comp.os.linux
Organization: The Ohio State University, Department of Computer and Information
Science
Lines: 61
Originator: [email protected]


Greetings. I am trying to install linux on my 386-33DX with 387,
with 32 megs of memory, and 2 Maxtor SCSI hard drives, running off an Adaptec
1542-B. (C is a XT8380S D is a XT8760S) I have 2 floppies: 1 5 1/4 HD, and
1 3 1/2 HD. I also have a 640x480x256 monitor with a Tseng ET4000 1meg.
Recently I tried 3 different times to install linux, and have
only recently had any luck getting any farther than bootong the floppy.

My last escapade invloved defining a 20 Meg partition with pfdisk
in DOS and then trying to make a file system inside linux.
I tried the scsi.fdisk getting NOTHING except the next prompt.
I then went and ftp'd and DL'ed the 96 bootimage. I tried scsi.fdisk
till I was blue.... Still nothing. So I went and did the following, and heres
what happened....: (oh, this was using the 95a root)
^L

#mknod /dev/sd0 b 8 1
#mkfs /dev/sd0 6144 (remember this is a 20 meg partition)

Linux then proceded to go:
280.....5something.....
Then the screen suddenly became filled with ANSI characters.. Every
one! (blocking the entire screen) After hitting a few keys, I locked up the
machine.
So I went back and went into the highest rez-graphics-text mode.
(thinking that maybe my ET4000 was somehow at fault) This time after the
wonderfull graphics garbeled my screen, the kernal dumped! I tried all the
text modes, thinking maybe something was wrong with my card. The higher
modes all dump, and produce similar results:
^L
scsidisk: free buffer being unlocked
general protection: 0000
EIP: 0008: 0000B7E8
EFLAGS: 00010283
fs: 0017
base: 1C000000, limit: 04000000
pid: 17 process nr: 7
83,3b,03,75,0f,68,b3,b7,00,00
memory fault
##
#

^L
Okay, now I am SERIOUSLY confused. I can make only about a 100 K
partition, and anything larger trashes memory. Can anyone out there offer
me some assistance? (drew?) Is the fact that I have 32 megs a prob?
Why does the scsi.fdisk STILL not report anything? Yes, I linked
hd0 the hda, etc... ARGHHH!!!!
(anyone know if any of the Adaptec's have bad ROMS in any rev?)

This is really starting to irk me, as 4 tiny partions and 2 floppies
just don't do it!

HELP!

Thanks in advance for any suggestions, etc....
--
----------------------------------------------------------------------
"Better to be silent and thought a fool, then to speak and remove all
doubt." --Abraham Lincoln
[email protected] !osu-cis!wahl CIS/ENG DPT. OSU


[next article]
From: [email protected] (Thierry Priol)

Subject: Linux on LocalBus 486DX33 motherboard
Message-ID: <[email protected]>
Date: 14 May 92 06:59:23 GMT
Sender: [email protected]
Organization: IRISA, Rennes (Fr)
Lines: 13

Does somebody has an experience in using a LocalBus 486DX33 motherboard with
Linux such as found in Zenon or Microlinks computers. Does X11 run with
the localbus ET4000 VGA board ?

Thanks in advance

Thierry Priol

--
Thierry PRIOL Phone: 99 84 72 10
IRISA / INRIA U.R. Rennes Fax: 99 38 38 32
Campus Universitaire de Beaulieu Telex: UNIRISA 950 473F
35042 RENNES CEDEX - FRANCE E-mail: [email protected]


[next article]
From: [email protected] (Joerg Pommnitz)

Subject: Re: Oh oh, problem with .96 bootimage
Message-ID:
Date: 14 May 92 06:50:55 GMT
References:
Sender: [email protected] (Owner of all binaries)
Organization: tu-chemnitz
Lines: 43

"Scott R. Grabowski" writes:

>Hi all,

>Well, since I am going on summer break tomorrow, I decided to grab the
>.96 bootimage (from tsx-11) ASAP. So I anxiously copied over the
>bootimage, and rebooted my machine. Unfortunately, the .96 bootimage
>caused my machine to reboot. I had no problem with the pre-.96 version,
>so it must be something new that Linus added.

>It reboots after telling me that I have 8 virtual ttys (or whatever the
>line is). I am guessing here that the problem is in the autodetect of
>the SCSI card. I do not have any drives that Linux recognizes (since I
>have an (actually, 2) 8 bit controllers for my MFM and RLL drives).

>Will the autodetect of SCSI hang if there are no hard drive cards
>available? Just a thought. I don't have the source, so I can't go
>trudging through it to find the problem.

>So, I guess that it is back to the pre-.96 kernel for now (and probably
>the summer, since I don't have FTP ๐Ÿ™ )

>Just thought that I would let you know. And, I'm open to suggestions
>until around 1PM or so tomorrow (DST, USA).

>AdvaThanksnce,

>-Scott

>PS - Thanks again Linus, this is the first problem that I've had with
>Linux so far (except for my hard drive problem, which is the fault of my
>soon to be replaced hardware ๐Ÿ˜Ž
>Scott Grabowski + "I don't pretend to understand women's
>[email protected] + little quirks, just one thing I know for
>Physics is Phun! + sure, chicks dig jerks." - great song!

This seems to be a general problem, because my machine shows the same
results.
I have uploadet the 0.96 sources from banjo and compiled it (no problems).
In his 0.96 announcement Linus says you " have to recompile the kernel with the
appropriate changes to config.h. ".
I wasn't able to figure out the settings to prevent any SCSI stuff
from my kernel. Any idea ?


[next article]
From: [email protected] (Eric_Munson)

Subject: Trident support for X?
Summary: has anyone started?
Message-ID: <[email protected]>
Date: 14 May 92 06:56:40 GMT
Organization: Western Washington University
Lines: 12


Has anyone started to put in the Trident support for
X windows? I saw that there was some stuff at banjo.concert.net
Also I am having problem with my modem causing
inux to hang when I boot it up. (ended up removing it from the
system.) It is a Supra 2400 internal.. I havn't had any problems
with previous versions..
--
*___________________________________________________________________________*
! Eric Michael Munson ! [email protected] ! Space intentionally !
! 1148 Fairhaven ! Journalism Major ! Left !
! Bellingham, WA 98225 ! 206-738-7181 ! Blank. !


[next article]
From: [email protected] (Teresa Selling)

Subject: GCC and X programming question
Message-ID: <[email protected]>
Date: 14 May 92 07:46:16 GMT
Sender: [email protected] (USENET News System)
Organization: Indiana University at South Bend
Lines: 16
Nntp-Posting-Host: peabody.iusb.indiana.edu

I am trying to learn to program with the X toolkits. The problem is when
compiling I get
Undefined symbol _XtInitialize referenced from text segment.
Undefined symbol _XtCreateManagedWidget referenced from text segment.
... and so on ...

I am including the proper include files, and using the right libraries
(actually I have tried including them all) but cannot get a simple program
to compile ๐Ÿ™ Any help appreciated.

Also how do I get on the GCC and X386 mailing lists?



--
Teresa Selling, IUSB, Indiana.


[next article]
From: [email protected] (Henning Spruth)

Subject: Re: Microsoft mouse works, Logitech doesn't..
Message-ID:
Date: 14 May 92 07:17:14 GMT
References: <[email protected]> <19[email protected]
sd.kodak.com>
Sender: [email protected]
Lines: 111


>just for reference, stty works on stdout, not stdin, so you need to use
>stty -istrip > /dev/ttys2 for it to take place.

>i am not happy with this -istrip stuff. the code which did this was to
>get around the fact that CS7 is not honored by the kernel. this is required
>by the microsoft mouse, which uses CS7. here are excerpts from the code:

> case DEVICE_ON:
> if ((x386Info.mseFd = open(x386Info.mseDevice, O_RDWR | O_NDELAY)) < 0)
> {
> Error ("Cannot open mouse");
> return (!Success);
> }

> if (x386Info.mseType != P_BM)
> {
> x386SetSpeed(9600, x386Info.baudRate, cflag[x386Info.mseType]);
> x386SetSpeed(4800, x386Info.baudRate, cflag[x386Info.mseType]);
> x386SetSpeed(2400, x386Info.baudRate, cflag[x386Info.mseType]);
> x386SetSpeed(1200, x386Info.baudRate, cflag[x386Info.mseType]);

> if (x386Info.mseType == P_LOGI)
> {
> write(x386Info.mseFd, "S", 1);
> x386SetSpeed(x386Info.baudRate, x386Info.baudRate, cflag[P_MM]);

****** shouldn't this read 'cflag[P_LOGI]' ????
****** Else, the cflags for the Mouseman will be used

> }

> if (x386Info.sampleRate <= 0) write(x386Info.mseFd, "O", 1);
> else if (x386Info.sampleRate <= 15) write(x386Info.mseFd, "J", 1);
> else if (x386Info.sampleRate <= 27) write(x386Info.mseFd, "K", 1);
> else if (x386Info.sampleRate <= 42) write(x386Info.mseFd, "L", 1);
> else if (x386Info.sampleRate <= 60) write(x386Info.mseFd, "R", 1);
> else if (x386Info.sampleRate <= 85) write(x386Info.mseFd, "M", 1);
> else if (x386Info.sampleRate <= 125) write(x386Info.mseFd, "Q", 1);
> else write(x386Info.mseFd, "N", 1);
> }

>this is used to turn the device on. if it's not a busmouse, it will try all
>the speeds from 9600 down (not quite sure why), but it obviously uses the
>cflags based on the mouse used.

>static void
>x386SetSpeed(old, new, cflag)
> int old, new;
> unsigned short cflag;
>{
> struct termio tty;
> char *c;

> (void) ioctl(x386Info.mseFd, TCGETA, &tty);
> tty.c_iflag = IGNBRK | IGNPAR ;
>#ifdef linux
> if (cflag & CS7)
> tty.c_iflag |= ISTRIP;
>#endif

**** This DOES NOT WORK!
**** In termios.h, CS7 ist defined to 40, and CS8 is defined to 60
**** So if cflag is set to CS8, the expression 'cflag & CS7'
**** evaluates to 40 !!!!
**** Therefore, ISTRIP is set for ALL mouse types!

> tty.c_oflag = 0;
> tty.c_lflag = 0;

>this is the setspeed function. nothing funny going on except for the
>ISTRIP handling which is done only if CS7 is specified.

>from this, i'm puzzled why people are having problems. by default, the
>serial devices come up with cs8 and -istrip, and the above logic doesn't
>do anything funny.

I think it's the bug I mentioned above.

Using 'stty -istrip >/dev/ttys1', my Logitech mouse works - almost
(jerky movement, mouse keeps pushing buttons on its own).

I played around a bit using "echo -n 'N' >/dev/ttys1' and found
that the character written to the mouse port which is derived
from the selected SampleRate greatly affects mouse behavior.

Using a SampleRate of 70 (i.e. writing a 'M' to /dev/ttys1) yields
the best results: the mouse almost works correctly. It is just
a bit jerky, making it difficult to select items from small
menus. I really don't know what I'm doing when writing to the mouse
port - I guess the mouse is initialized by some special character
sequences, right?

So right now I'm using

Logitech "/dev/ttys1"
BaudRate 1200
SampleRate 70

in my xconfig and a line

stty -istrip >/dev/ttys1

in my xinitrc, which makes my Logitech mouse almost useable.

Henning
--
------------------------------------------------------------------------
Henning Spruth [email protected]
Institute of Electronic Design Automation
Technical University of Munich, Germany


[next article]
From: [email protected]

Subject: mcc JANET host name
Message-ID: <[email protected]>
Date: 14 May 92 09:58:21 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 7

Can anyone tell me the janet host name for MCC ?
I've tried looking in all the documentation, FAQ, readmes, digests,
but can't see it.
Thanks JJ:: Jon Jagger [email protected]
:: Sheffield City Polytechnic, Pond Sreet, SHEFFIELD S1 1WB
:: Tel 0742 533802/432889 (work/home) Fax 0743 533840
:: Newspaper ad: Men wanted for expanding contracting company!


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

Subject: serial response problem in .96
Message-ID:
Date: 14 May 92 08:15:58 GMT
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 12

There's a probleml with the serial code in .96 that's serious enough
I'm going to have to back out. In both kermit and KA9Q, response
is very slow. It's like it take 1/2 second to wake up when a
character comes in. The first character echoes OK, but if I type
ahead, no others echo until some timeout happens. Since I'm a
fast typist, I can't live with it. It's worse for KA9Q, because
it causes problems with acks, etc.

On the good side, the select problem seems to be fixed. You
can now start emacs while doing an FTP in KA9Q without slowing
down the FTP. (However the FTP is already slow enough, because
of the problem just mentioned.)


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

Subject: serial problem: some progress
Message-ID:
Date: 14 May 92 09:18:59 GMT
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 8

I think the main problem is in serial.c, in routine send_intr.
The next to last line has LEFT where it should have CHARS.
Fixing that improves things greatly. However there may still
be a problem of some sort. I'll continue playing.

Like some other people I find that I have to do a hardware
reset when rebooting. Otherwise the system crashes near the
beginning of startup.


[next article]
From: [email protected] (Hillel Steinberg)

Subject: X pty problem (can't find any)
Keywords: X11, linux
Message-ID: <[email protected]>
Date: 14 May 92 12:14:28 GMT
Sender: [email protected]
Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
Lines: 16


Hey Xers,

I can get most X clients to work, but no xterm. Xterm reports something
like "child: cannot find any available pty". I know this from the stderr
that I redirect to a file. Now I have all 16 ttyp's and ptyp's, and they
are the correct major/minor numbers. I have also upped the number of
pty's to 16 as someone suggested. No go... the xterms just don't come
up. Why can't the xterm find the /dev/ptyp's or /dev/ttyp's ???? They
are right there !!!! (Permissions are fine ... I'm root)

Also, I compiled xconsole... all went ok ... but when I run it, it says,
can't open console.... xterm I downloaded from banjo, and xconsole I hardly
modified.... What am I doing wrong?

- Hillel


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

Subject: Re: Need SCSI install help (a lot!)
Message-ID: <[email protected]>
Date: 14 May 92 11:56:01 GMT
Sender: [email protected]
Lines: 17

DE> >I have been wanting to ask if anyone has attempted to
DE> > write code for the IN2000 SCSI controler, of if the Seagate one is
DE> >compati with the IN2000?
DE>
DE> 1. Not to my knowledge
DE> 2. No.
DE>
DE> Your best bet (if you can afford it) is the Adaptec 154x,
DE> if money is your
DE> only concern (ie : free SCSI disk and you want to use it), the
DE> Seagate has lousy performance, but can be found for $20 for
DE> a SCSI only controller.

Thanks for the reply. Guess I will try an IDE, the 40 meg MFM is just not
enough space for LINUX. And I want to try X11 at some point in time.

* Origin: Intermittent Connection Eugene, OR (1:152/35)


[next article]
From: [email protected] (Anthony Rumble)

Subject: HD corruption with new 0.96 kernel
Summary: My binaries are corrupting with the new kernel
Keywords: linux
Message-ID:
Date: 14 May 92 11:47:32 GMT
Sender: [email protected]
Organization: Sydney University Computing Service, Sydney, NSW, Australia
Lines: 33
Nntp-Posting-Host: extro.ucc.su.oz.au

Since I installed the new Kernel.. I have been getting some wierd
things happening with GCC 2.1..

It starts off working fine.. then after a bit.. things go haywire..
Like recieved signal 11 etc etc..

Under 95c+ it worked fine..

It ony starts happening after I do some heavy compiling with GCC..

If I reload the binaries if GCC.. The problem goes away.. (Unless I
do some more GNUing under .96.. Then the same happenes..


I recompiled the kernel.. Hoping for the best (Under 95+).. But it diddn't seem
to fix it.. And now im getting worse errors.. like "STDERR" undefined first use!
!!!!!.. So looks like the headers are corrupted now!

I have no idea what else is corrupted as yet..

I am running on a 486/33 with 4 Megs mem.. 2 Meg swapper..
With a Kylok 100 Meg IDE Hard Disk (BTW I have to comment out that "HD unexpecte
d interupt" message.. as this comes up all the time.. yet diddnt; seem to cause
any probs under .95c+)

Also.. One other thing.. I heard mention that Linux can only handle 32 Meg File
systems...

Is this true? because my df looks like this..
device "" "" "" "" "" "" "" blocks total blocks used
/dev/hda4 50820 28030

Could this be the cause of my problem?!?!?!?
(Having a 50Meg partition)

Hope someone can help me..
Linus.. keep up the good work..
Anthony Rumble


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

Subject: Don't use linked libs for crutial stuff [was:Re: rc Shell Under Linux]
Message-ID: <[email protected]>
Date: 14 May 92 13:04:24 GMT
References: <[email protected]> <[email protected]
.claremont.edu>
Sender: [email protected] (USENET News system)
Organization: University of Maryland, College Park
Lines: 23
Nntp-Posting-Host: next.wam.umd.edu

In article [email protected]
du (goon) writes:
>In article <[email protected]> [email protected]
mont.edu (Jim Winstead Jr.) writes:
> This might change when I convert the root disk over to shared
> libraries, which I expect to free up a good deal of space. In that
> [...]
>
>Even if the root image is converted to shared libs, it is proabally a
>good bet that the shell (whatever that may be) will be statically
>linked because having a shared-library version of the shell requires
>too much to be working for one to get a root prompt. Consider the
>following nasty situations:
> [two nasty situations deleted]

I think that anything really vital should be statically linked. As I
see it, this would include a shell, mount, rm, cp, sync (!) and
perhaps tar (that's a killer) or dd.

The idea is that if you inadvertantly delete the shared lib apparatus,
or mangle one or more of your partitions, you should be able to restore
enough to get shared libs working again.

-Joel
([email protected])


[next article]
From: [email protected] (Jorma Heimonen)

Subject: 8-bit-clean version of Curses??
Message-ID: <[email protected]>
Date: 14 May 92 10:46:32 GMT
Sender: [email protected] (Usenet pseudouser id)
Reply-To: [email protected] (Jorma Heimonen)
Organization: Helsinki University of Technology
Lines: 21
Nntp-Posting-Host: vipunen.hut.fi

Hi,
Is there anybody else interested in using 8-bit
characters with Linux curses. The version distributed with
gcc 2.1 seems to be the original BSD version, which by default
strips the 8th bit (ARRGH!!) and uses the 8th bit for standout.
I hacked the sources for an hour, and the best I got was text
with 8-bit characters in reverse video or alternatively, if
I commented out the _puts(SO) line in refresh.c, 8-bit text
but of course without any kind of standout. It didn't make
any difference to define _STANDOUT to 0400 instead of 0200.

So, before I start further hacking, I would like to know, if
anybody else has done it/is doing it. Finally, it would be
nice to extend curses to SYSV level, with UNDERLINE, BLINK,
BOLD, DIM, ALTCHARSET etc. attributes and decent line and box drawing
as well as the function key handling.

Jorma Heimonen
Helsinki University of Technology




[next article]
From: [email protected] (Richard Harry)

Subject: Re: Error: Job lost account erased. Find new account and try again.
Message-ID: <[email protected]>
Date: 14 May 92 10:19:15 GMT
References: <[email protected]>
Organization: University of Sussex
Lines: 13

[email protected] (Lawrence C. Foard) writes:
: I'm surprised that still so few people know about Linux, Linux is much
: more reliable than the initial OS/2 release, and yet I don't think any
: magazines have acknowledged its existance.

Here in the UK, the current issue ( June 1992 ) of Personal Computer World
has a write up of Linux ( in the unix section at the back of the mag,
for those who are interested ). Of course it's totally out of date.
e.g. the writer says you can only be root, no login etc..( How long ago
was that ??? 0.12 ! ). He does mention that by the time the article is
published, Linux should have evolved somewhat ( I'll say )

Richard


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

Subject: Gnuchess
Message-ID: <[email protected]>
Date: 14 May 92 12:12:53 GMT
Sender: [email protected] (Usenet poster for nntp (tj - 17/12/91))
Reply-To: [email protected]
Organization: Bristol Polytechnic
Lines: 12
Nntp-Posting-Host: kiwi

I'm just sending a update to the gnuchess that I put on tsx-11.mit.edu.
I forgot to include the Makefile, the Man page and a better Readme, ok
my spellings awful when Im in a hurry.

This version is compiled using shared libaries and will only work under
version 0.96 and gcc 2.1

----------------------------------------------------------------------------
Chris watts
Email: Internet [email protected]
Janet [email protected]
----------------------------------------------------------------------------


[next article]
From: [email protected] (Michael O'Reilly)

Subject: Re: HD corruption with new 0.96 kernel
Message-ID: <[email protected]>
Date: 14 May 92 13:20:59 GMT
References:
Sender: [email protected] (USENET News System)
Organization: University of Western Australia
Lines: 63
Nntp-Posting-Host: tartarus.uwa.edu.au

[email protected] (Anthony Rumble) writes:
: Since I installed the new Kernel.. I have been getting some wierd
: things happening with GCC 2.1..

Tell me about it. ๐Ÿ™ I have just finished re-building my partition
from scratch. Fsck gave me nearly 4000 lines of errors. ๐Ÿ™
At that point I decided to retrieve what I could and build a new file
system. This occurred sometime after my switch to the 0.96 kernel. I had
been doing a lot of compileing..

:
: It starts off working fine.. then after a bit.. things go haywire..
: Like recieved signal 11 etc etc..
Don't know this.I occasionally get IOT trap when resuming suspended
jobs.

:
: Under 95c+ it worked fine..
:
: It ony starts happening after I do some heavy compiling with GCC..
:
: If I reload the binaries if GCC.. The problem goes away.. (Unless I
: do some more GNUing under .96.. Then the same happenes..
:
:
: I recompiled the kernel.. Hoping for the best (Under 95+).. But it
: diddn't seem to fix it.. And now im getting worse errors..
: like "STDERR" undefined first use!!!!!!.. So looks like the headers
: are corrupted now!
:
: I have no idea what else is corrupted as yet..
:
: I am running on a 486/33 with 4 Megs mem.. 2 Meg swapper..
: With a Kylok 100 Meg IDE Hard Disk (BTW I have to comment out
: that "HD unexpected interupt" message.. as this comes up all the time..
: yet diddnt; seem to cause any probs under .95c+)

I am running 486/33 with 4 megs, 8 megs swap, 200 meg IDE hard disk.
I too , had to comment out the 'Unexpected hard disk interrupt'
message. It was comeing up about 25 times a second during a compile....
:
: Also.. One other thing.. I heard mention that Linux can only handle
: 32 Meg File systems...
Nahh. Linux currently handles up to 65535 blocks..
My partition is this size (all my linux stuff is on the one partition).
:
: Is this true? because my df looks like this..
: device "" "" "" "" "" "" "" blocks total blocks used
: /dev/hda4 50820 28030
:
: Could this be the cause of my problem?!?!?!?
: (Having a 50Meg partition)
:
: Hope someone can help me..
: Linus.. keep up the good work..
: Anthony Rumble
Well, I can't help. ๐Ÿ™ As an aside, does anyone have detailed info on
programming the IDE hard disk controller ???. This 'Unexpected HD error'
is bugging me. Damned if I know what is causeing it..

Michael.

[email protected]


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

Subject: Testers wanted... new 0.96 image
Message-ID: <[email protected]>
Date: 14 May 92 13:23:45 GMT
Organization: University of Helsinki
Lines: 15

I've put a new testimage on banjo.concert.net:

pub/Linux/Linus/boot92.05.14.Z

and I hope people that have problems with 0.96 could try it out. I've
mainly tried to remove some races in the serial code, but I hope this
image also corrects the bootup problems experienced by some people.

If you have problems booting the original 0.96, or are seeing more
errors with the new serial lines, please try it out. As with the
earlier testimage, it's no use to me if you can't test it out in a day
or two: by that time I'll probably release 0.96a or a new testimage
depending on the success of this one.

Linus


[next article]
From: [email protected] (Fisher S A)

Subject: Re: grrrrr RELAY
Message-ID:
Date: 14 May 92 13:44:29 GMT
References: <[email protected]>
Sender: [email protected]
Organization: Organised? Never!
Lines: 44
In-reply-to: [email protected]'s message of 13 May 92 17:12:32
GMT

In article <[email protected]> [email protected]
ac.uk (Jon Jagger) writes:

|Can anyone help ?
| I've been looking at the stuff at tsx-11.mit.edu and I managed to get
|enough stuff to make a root and boot disk. (doesn't it look good).
| I've decided that I'll put it on a hard disk that I just happen to
|have spare. My problem is ftp. I can only get the stuff via a janet
|relay, and our stream is so piddlingly small, that I have not been
|able to get back onto the relay, let alone, mit, for four days now!
| I desparately want the following
| a) /pub/linux/docs/linux-fs-standard
|b) /pub/linux/docs/linux-history
|c) /pub/linux/mirrors/GCCv2/*.*
|d) /pub/linux/mirrors/mcc-interim/0.95c+/*.* (and sub dirs)
|e) /pub/linux/ports/shoelace.tar.Z
|f) /pub/linux/sources/libs/README.curses
|g) /pub/linux/sources/libs/curses.tar.Z
| Can anyone help ?
|Or know a man/woman who can ?
| Thanks Jon Jagger

Not just you. At Essex we have the same problem. I think it is JANET wide.
I managed to get some stuff from uk.ac.ic.doc.src (via the PAD) but it is very
out of date.

People keep mentioning interim stuff at mcc.ftp. Is this Manchester Computing
Centre in England?? If so, it is not registered in the Janet database. I
mailed the postmaster but haven't had any reply yet.

Please, can somebody make stuff available to us in the UK. Imperial college
doesn't have 0.96 yet. the nsf.sun relay is hopeless. I can't get through
to nic.funet.fi. banjo.concert.net it has never heard of. Please can we
have some internet numbers or better yet a place in the UK that we can pad to.

Thanks

Steve Fisher

____ ____ ____ E-Mail [email protected]
/ / / / 3rd Year Computing Science Undergraduate
/___ /___/ /____ at the University of Essex
/ / / /
___/ / / / Earth shutting down in 5 minutes,
please save all files and log out


[next article]
From: [email protected] (Ken Corey - Operator)

Subject: Re: grrrrr RELAY
Message-ID: <[email protected]>
Date: 14 May 92 15:03:31 GMT
References: <[email protected]> ssex.ac.uk>
Sender: [email protected] (UNT USENet Adminstrator)
Organization: University of North Texas
Lines: 9

[email protected] (Fisher S A) writes:
>to nic.funet.fi. banjo.concert.net it has never heard of. Please can we
>have some internet numbers or better yet a place in the UK that we can pad to.

My host returns 192.101.12.6 as the IP address for banjo.

Hope this helps.

-Ken


[next article]
From: [email protected]

Subject: MTOOLS/X and du
Message-ID: <[email protected]>
Date: 14 May 92 15:23:51 GMT
Sender: [email protected]
Reply-To: [email protected]
Organization: Cornell University School of Chemical Engineering
Lines: 5
Nntp-Posting-Host: hermes.cheme.cornell.edu

Does anyone know where to get du or du.c for LINUX? Has anyone experienced
problems with mtools under X11R5. when I do a mdir or m??? my xterm locks
up - not even -c gets me anywhere...
Elan Feingold
Cornell University


[next article]
From: [email protected] (R.J.Newton)

Subject: Compiling Ghostscript 2.4.1
Message-ID: <[email protected]>
Date: 14 May 92 10:27:32 GMT
References: <[email protected]>
Sender: [email protected]
Organization: Cripps Computing Centre, University of Nottingham
Lines: 12

I downloaded the sources for Ghostscript 2.4.1 from wuarchive.wustl.edu last
night and attempted to compile them with gcc 2.1. After coaxing the compiler
around several errors it finally barfed in the linking with a message like:

In lib(something).a: Undefined symbol _connect referenced from text segment.

...or something like that, I'm doing this from memory. Has anybody got a
successfully compiled version that has the laserjet driver ? If they have then
could they please post it to tsx-11. Groff only outputs postscript code and
I don't have a postscript printer !

-Rich


[next article]
From: [email protected] (r b willmersdorf)

Subject: Help with video timing please (Have data, don't know what it mea~s
Message-ID: <[email protected]>
Date: 14 May 92 16:46:55 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 48


Hi,

I'm trying to get X working on resolutions higher than 640x480.
I suppose I'm luckier than some people, because it already does
work at this resolution, and my monitor came a timing chart. Only
I'm not fully confident on what their data means. I tried to
configure a higher resolution mode following the video tutorial
and failed abysmally. So if anyone can help me to relate the
monitor data to what is necessary for X386, I'll be forever gratefull.

The graphs are :


__________________________ ______________
video | | |
_______________| |______________|

C D E
|<---->|<---------------------->|<----->|
____ _________________________________________ _______________
sync | | | |
|____| |____|
B
|<-->|
A
|<------------------------------------------>|

This is for the horizontal scan, and the times are given in micro-seconds.
I can easyly convert them to ticks, because I have the scanning frequencies,
as long as I know which one is which.

These are my guesses, and I tremble with expectation for the wisdom of
linux people.

Frame lenght = A
width of sync pulse = B
Start of sync pulse = E (referred to end of D)
End of sync pulse = C (referred to the start of D)

I think I can pretty much tackle the configuration, as long as I know
what the data really means.

Many thanks,

--

Ramiro || [email protected]


[next article]
From: [email protected] (Alan B Clegg)

Subject: Linux available for FTAM (yes, full OSI protocol stack)
Message-ID: <[email protected]>
Date: 14 May 92 17:05:11 GMT
Sender: [email protected]
Organization: Concert Network -- Internet Operations Group
Lines: 19

Well, with everyone talking about FTP problems from the UK, I figured I could
throw another confusing protocol into the fray:

CONCERT Network is now fully connected with CLNP to the Internet, allowing me
to make Linux available for anonymous FTAM.

The entry for the ISODE/SunNET OSI compliant isoentities table is:

banjo default NULL #1/NS+47000580ffff0000000b0100010800200b49b000|\
Internet=192.101.21.6

The userid is "ANON" with no password.

If you actually use this facility, let me know how it works...

-abc
--
[email protected] Alan Clegg - Network Programmer
The Future: Live it or live with it! MCNC -- Center for Communications


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

Subject: gdb postmortem debugging...
Message-ID: <[email protected]>
Date: 14 May 92 16:51:06 GMT
Sender: [email protected] (News Database (admin-Mike Schwager))
Organization: University of Illinois, Dept. of Comp. Sci., Urbana, IL
Lines: 10

I have been using gdb for a little while, and it's working fine,
except that I can't do postmortem debugging: once the program
gets a SIGSEGV, it exits, and there is nothing left for gdb to
debug.

Is this a known bug? Or is my understanding of how gdb
works flawed?

I am using pre-0.96 kernel + everything else from mcc-interim
release, on a AMI 386/40 + st3096IDE.


[next article]
From: [email protected] (The Crouton Man)

Subject: Re: Oh oh, problem with .96 bootimage ANSWERS_AND_QUESTIONS
Summary: aanswer to get kernel without scsi and q's about 32mb, uucp and libs
Keywords: SCSI 32mb uucp uucico putpwent
Message-ID: <[email protected]>
Date: 14 May 92 16:05:32 GMT
References: e>
Sender: [email protected] (The News Manager)
Organization: Dartmouth College, Hanover, NH
Lines: 44

:In his 0.96 announcement Linus says you " have to recompile the kernel with the
:appropriate changes to config.h. ".
:I wasn't able to figure out the settings to prevent any SCSI stuff
:from my kernel. Any idea ?

you need to change 2 things. first in the file include/linux/config,
change it to:
#undef CONFIG_DISTRIBUTION

and then, becuase the SCSI stuff doesn't like being compiled with
nothing included (perhaps a fix next release?) in your global make file,
change it to:

DRIVERS =blahblah.........................chr_drv.a
# bhahbhah/scsi.a


(where you only deleted the \ from the first line and added the # to
the second).

i added the 32mb patchs (that applied) and the sock_init gave a
segv'd on bootup.. i added the lines
printk("ENTERING SOCKET INIT!\n");
at the beginning and
printk("LEAVING SOCKET INIT!\n");
and the end, and the segv went away.. sounds like something is
being optimized away or else is otherwize screwy in the socket
init routine.

anyway, after it boots, when i run a test program the simply
malloc's lots-o-memory (test the limits, yahknow) i get a
kernel panic about freeing a page..

i have 20Mb, any ideas? (i will keep looking)

also, has anyone gotten uucico to run (from taylor 1.3) without
segv'ing after the initial login?

which libraries do i need to recompile all my libraries..
putpwent() from the mcc-interim-095c+ release doesn't write
the home path field and screws up the passwd file.

croutons.
[email protected]


[next article]
From: [email protected] (Ari Lemmke)

Subject: Re: Problems compiling UMoria 5.4: please join GCC
Message-ID:
Date: 15 May 92 00:26:14 GMT
References: <[email protected]> <[email protected]>
Sender: [email protected] (Usenet pseudouser id)
Organization: Helsinki University of Technology, Finland
Lines: 27
In-Reply-To: [email protected]'s message of 9 May 92 22:12:56 GMT
Nntp-Posting-Host: cardhu.cs.hut.fi


In article <[email protected]> [email protected] (H.J
. Lu) writes:
: P.S. To join GCC, send mail to
:
: [email protected]
:
: with only one line
:
: Mn-Admin: join GCCC

Mn-Admin: join GCC

Would do it!

I was wondering why many people tried to join gccc channel ...

Mn-Info: channel

To get channel list.

arl


-----
I have been constantly taking out user "[email protected]",
because *.uucp address is not valid, maybe in the states,
but that address is definitely not in Finnish uucp system ๐Ÿ˜‰


[next article]
From: [email protected] (Peter MacDonald)

Subject: 0.96 praise
Message-ID: <[email protected]>
Date: 14 May 92 17:35:23 GMT
Sender: [email protected]
Organization: University of Victoria, Victoria, BC, CANADA
Lines: 5
Nntp-Posting-Host: sanjuan.uvic.ca


Amidst the hail of criticism 0.96 has received, I thought I should
report that it has worked fine for me. In particular, X seems to
be much more stable, and the serial code no longer loses characters,
even in X with the largest font scrolling on my 386/25. Excellent.


[next article]
From: [email protected] (Arno Schmitt)

Subject: linux 0.96
Message-ID: <[email protected]>
Date: 14 May 92 15:51:07 GMT
Sender: [email protected]
Lines: 19


Hello,


I've some trouble using linux version 0.96.

After removing the scsi-stuff from the sources and defining German
keyboard, I've managed to compile a boot image. For compiling I
used gcc 1.40.

When booting this image, I get the following error message:

Kernel panic: Trying to free up swapper space
In swapper task - not syncing

What does this mean and how to avoid it?

Thanks for your help
Arno
([email protected])


[next article]
From: [email protected] (goon)

Subject: Re: rc Shell Under Linux
Message-ID:
Date: 14 May 92 17:03:51 GMT
References: <[email protected]>
<[email protected]>

<[email protected]>
Sender: [email protected]
Organization: St. Olaf College, Northfield, MN USA
Lines: 13
In-Reply-To: [email protected]'s message of 14 May 92 05:51:13 GMT

In article <[email protected]> [email protected]
nt.edu (Jim Winstead Jr.) writes:

>(2) the dynamic-linker has had it's libs (or, on some systems cache)
> corrupted.

Dynamic linker? The libraries are shared, not dynamically linked.

How does one share libs without dynamic linking? How does the system
make sure that the function call in the application jmp's to the right
address? There has to be some sort of mechanism for setting this up at
runtime (dynamic linking) unless (a) calls to the lib are really traps
(b) the libs are at a fixed location in memory. or (c) the kernel
translates each call on-the-fly.


[next article]
From: [email protected] (Michael K. Johnson)

Subject: Re: Don't use linked libs for crutial stuff [was:Re: rc Shell Under Linux]
Message-ID: <[email protected]>
Date: 14 May 92 18:13:43 GMT
References: <[email protected]> <[email protected]
.claremont.edu> <1992May14.130424.16
[email protected]>
Sender: [email protected]
Organization: St. Olaf College; Northfield, MN USA
Lines: 18
In-Reply-To: [email protected]'s message of Thu, 14 May 1992 13:04:24 GMT

In article <[email protected]> [email protected] (Joel M. Hoffma
n) writes:

I think that anything really vital should be statically linked. As I
see it, this would include a shell, mount, rm, cp, sync (!) and
perhaps tar (that's a killer) or dd.

The idea is that if you inadvertantly delete the shared lib apparatus,
or mangle one or more of your partitions, you should be able to restore
enough to get shared libs working again.

I think that the regular rootimage disk should use shared libs, but
that an emergency rootimage disk that uses only staticly linked stuff
should be available for fixing things. Jim -- could you make this as
well? I hate to volunteer you, but you've done a good job so far...
๐Ÿ™‚

michaelkjohnson
[email protected]


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

Subject: Just Curious: Anybody Compile Netrek?
Message-ID: <[email protected]>
Date: 14 May 92 18:38:30 GMT
Sender: [email protected] (USENET News Service)
Organization: University of Delaware, Newark
Lines: 9
Nntp-Posting-Host: strawberry.cns.udel.edu

It's an almost horrendous thought, but out of curiosity, now that X is
available and there's somebody with a WD8003 patch, has anybody managed
to compile netrek? (-: (-: (-:

--
-- 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] (Johan Myreen)

Subject: Latin-1 keyboard driver for Linux 0.96
Message-ID:
Date: 15 May 92 01:11:14 GMT
Sender: [email protected] (Usenet pseudouser id)
Distribution: comp
Organization: Helsinki University of Technology, Finland
Lines: 60
Nntp-Posting-Host: psykoman.cs.hut.fi


A while ago there was some talk about using the Latin-1 character set
on Linux. My updated patches to the 0.96 driver didn't make it into
the official distribution, so I have uploaded a new driver to
nic.funet.fi.

Here's the README:

--
This directory contains a new keyboard driver for Linux 0.96 using the
Latin-1 character set. The file keyboard.c replaces keyboard.S in
linux/kernel/chr_drv. 'patches' contains patches to linux/Makefile and
linux/kernel/chr_drv/Makefile.

Features of the new driver:

- It is now written in C

- Fully compatible with the old driver

- The old keyboard mappings are still there, supplemented with
new Latin-1 maps for

* German (KBD_GR_LATIN1)
* French (KBD_FR_LATIN1)
* Danish (KBD_DK_LATIN1) and
* Finnish (KBD_FINNISH_LATIN1) keyboards

- Support for so called dead keys. Hitting a dead key does not
produce a character immediately but instead the driver waits
for another character to be typed. If possible, the two
keystrokes are then combined into an accented character. For
instance, pressing the backquote key followed by an 'a'
produces the character 'a accent grave'. (Pressing a dead
key twice or followed by a space produces the diacritical
mark itself.)

- Customizable: with a compile time option you can choose
whether the Alt key sets the high bit or prepends an escape
character. (Essential for using the driver with 8-bit
characters.)

- Also with a compile time option you can choose whether
certain keys are dead or not. See linux/Makefile for info
on changing these settings.

- The above settings are now stored in a global variable, which
makes it possible to write an interface with which you could
change the settings at run time.

- Characters not found on your keyboard can be entered using
the numeric keypad and the Alt key. For example, Alt 1 6 3
produces a pound sign.

Problems remain with certain applications like emacs and bash which do not
understand 8-bit characters. Elvis seems to be "8-bit clean", though.

--
Johan Myreen
[email protected]


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

Subject: Re: rc Shell Under Linux
Message-ID: <[email protected]>
Date: 14 May 92 19:36:17 GMT
References: <[email protected]
uddcs.claremont.edu>
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, WIBSTR
Lines: 23

In article [email protected] (
goon) writes:
>In article <[email protected]> [email protected]
ont.edu (Jim Winstead Jr.) writes:
>
> >(2) the dynamic-linker has had it's libs (or, on some systems cache)
> > corrupted.
>
> Dynamic linker? The libraries are shared, not dynamically linked.
>
>How does one share libs without dynamic linking? How does the system
>make sure that the function call in the application jmp's to the right
>address? There has to be some sort of mechanism for setting this up at
>runtime (dynamic linking) unless (a) calls to the lib are really traps
>(b) the libs are at a fixed location in memory. or (c) the kernel
>translates each call on-the-fly.

(d) binaries are linked against a stub library that uses the shared
library mechanism to load the correct shared library - there has been
a new shared library with each release of GCC 2.1 (by hlu, not by GNU).
--
+ Jim Winstead Jr. (CSci '95)
| Harvey Mudd College
| [email protected]
+ This is all my words. Honest!


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

Subject: Re: Don't use linked libs for crutial stuff [was:Re: rc Shell Under Linux]
Message-ID: <[email protected]>
Date: 14 May 92 19:40:02 GMT
References: <[email protected]
wam.umd.edu> <[email protected]>
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, WIBSTR
Lines: 26

In article <[email protected]> [email protected]
(Michael K. Johnson) writes:
>In article <[email protected]> [email protected] (Joel M. Hoffm
an) writes:
>
> I think that anything really vital should be statically linked. As I
> see it, this would include a shell, mount, rm, cp, sync (!) and
> perhaps tar (that's a killer) or dd.
>
> The idea is that if you inadvertantly delete the shared lib apparatus,
> or mangle one or more of your partitions, you should be able to restore
> enough to get shared libs working again.
>
>I think that the regular rootimage disk should use shared libs, but
>that an emergency rootimage disk that uses only staticly linked stuff
>should be available for fixing things.

I still think some people aren't thinking this through enough. ๐Ÿ™‚

When you use the root image, it should only be to give Linux a try,
install Linux, or recover from some heinous mistake. It shouldn't
matter whether the root disk uses shared libs or statically-linked
binaries - it is completely self-contained.
--
+ Jim Winstead Jr. (CSci '95)
| Harvey Mudd College
| [email protected]
+ This is all my words. Honest!


[next article]
From: [email protected] (Ames Grisanti)

Subject: How to set up swap space?
Message-ID: <[email protected]>
Date: 14 May 92 20:02:40 GMT
Organization: University of North Dakota EERC=
Lines: 19

Hello,

I am running Linux 0.96, and have a single file system for it
set up on a 65 MB HD. Could someone tell me how to set up a swap
partition on the HD? Sorry if this is an elementary question but
I tried to set one up once already and ended up having to relaod
the system. Any help would be apprecieted.

thanks,

Ames Grisanti

University of North Dakota
Energy and Environmental Research Center
[[email protected] -- Internet]
[[email protected] or ndsuvax -- Bitnet ]
[uunet!plains!grisanti -- UUCP ]
"Quantum Mechanics, the dreams that stuff is made of."
-unknown


[next article]
From: [email protected] (Steven Samuel Jr. Seiden)

Subject: Elvis & X11, also libobz.a
Message-ID: <[email protected]>
Date: 14 May 92 20:15:38 GMT
Organization: Univ. of Calif., Irvine, Info. & Computer Sci. Dept.
Lines: 11
Nntp-Posting-Host: anita.ics.uci.edu

You can kludge X so that Elvis works by adding xterm to the list of
alternative names for vt100, and commenting out the entry for xterm.
BTW linux and X work great, now that I have it all set up right.
I didn't even have any reboot problems with .96. Does anybody know
where libobz.a is? I want to compile makedepend, (so I can compile
other X11 apps.) and it needs it. The only problem remaining is to
restore the normal text screen when X is exited.


--
Steve Seiden Dept of Info & Computer Science UC Irvine


[next article]
From: [email protected] (Daniel Brahneborg)

Subject: SYS V IPC in Linux?
Message-ID:
Date: 14 May 92 20:58:31 GMT
Sender: [email protected] (News Administrator)
Organization: Dep. of Info.Proc, Umea Univ., Sweden
Lines: 6
Originator: [email protected]

What are the plans for Shared Memory, Semaphores and Message Queues?
I have a databaseprogram that uses all of these things, which I would
LOVE to use in Linux.
So, is anybody working on them?

/Basic


[next article]
From: [email protected] (Magnus Alvestad)

Subject: Re: Trident support for X?
Message-ID:
Date: 15 May 92 03:19:18 GMT
References: <[email protected]>
Sender: [email protected] (Bergen University Newsaccount)
Reply-To: [email protected]
Organization: AlveSoft
Lines: 12
In-Reply-To: [email protected]'s message of 14 May 92 06:56:40 GMT

> Has anyone started to put in the Trident support for
> X windows? I saw that there was some stuff at banjo.concert.net

We are working on it. Will announce it here when finished. Be warned
that there are some problems with the speed of the cards. There may be
some damage to the higher resolution.

The more I do with the Trident, the more I want to buy a Tseng.. ๐Ÿ™‚

--
| Magnus Alvestad | I am a prolog program. I get all my opinions
| [email protected] | from a small organization in Birmingham.


[next article]
From: [email protected] (Mike Sanders)

Subject: Re: Small emacs
Message-ID: <[email protected]>
Date: 14 May 92 21:32:39 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 20

Someone asked recently about a small emacs for linux, more
like gnu emacs than uemacs.

I prefer mg. (/pub/linux/binaries/usr.bin/mg is the executable.)
To make this useful requires a startup file called .mg in the
HOME directory. Mine, which maps the arrow, home, end, page-up,
and page-down keys is:

global-set-key "\eg" goto-line
global-set-key "\e[A" previous-line
global-set-key "\e[B" next-line
global-set-key "\e[C" forward-char
global-set-key "\e[D" backward-char
global-set-key "\e[H" beginning-of-line
global-set-key "\e[Y" end-of-line
global-set-key "\e[5~" scroll-down
global-set-key "\e[6~" scroll-up
auto-indent-mode

I haven't figured out if it's possible to vary the tab width.


[next article]
From: [email protected]

Subject: Re: Oh oh, problem with .96 bootimage
Message-ID: <[email protected]>
Date: 14 May 92 22:41:28 GMT
References: <[email protected]> <23364
@dog.ee.lbl.gov>
Sender: [email protected] (Uutis Ankka)
Organization: University of Helsinki
Lines: 27

In article <[email protected]>, [email protected] (Robert J. McNamara) writes:
> In article <[email protected]> [email protected] writes:
>>
>>I'm using the 0.96 image without any problems... in fact, a large number of
>>problems I had with pre-0.96 are now gone. I don't have any SCSI devices,
>>though, so that might be the difference: the point at which you guys report
>>0.96 rebooting is when it is initializing and checking SCSI devices.
>>
>
> Nope, I've got an IDE drive and I've got the same thing happening.
> my system: 386/33 w/ OPTI chipset
> AMI bios 4/9/90
> Prodesigner II (4.01 vbios)
> Maxtor LXT200A
> Soundblaster
> is anyone else NOT having this reboot problem with 0.96?
>
I've just tested booting my machine up with 0.96-boot and 0.95a-root
-images, and I have no problems. (Still, I've not transferred anything
to hard disk yet, just played a bit with the floppies.)

386SX/16, very old (89? 90?) Phoenix BIOS
Old 8-bit ET3000
Conner CP3204F
2serial/1parallel

.Wolf/[email protected]


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

Subject: 0.96 works great for me!
Message-ID: <[email protected]>
Date: 14 May 92 22:38:55 GMT
Sender: [email protected] (USENET News system)
Organization: University of Maryland, College Park
Lines: 11
Nntp-Posting-Host: next.wam.umd.edu

Some people have asked, so I'd like to let everyone know that 0.96
booted the first time round with me, via bootlin. I'm logged into the
mainframe right now via kermit and have GCC running on another VT.
The response time is indeed better than it was before.

I have a Dell 325D ('386 at 25MHz) with 6M and 4M swap, 125M EDI HD.

Thanks again Linus!!

-Joel
([email protected])


[next article]
From: [email protected]

Subject: Re: grrrr RELAY
Message-ID: <[email protected]>
Date: 14 May 92 23:10:25 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 13

Hi, access to the sites carrying Linux from the UK seems fine to me. Anyway,
the IP addresses for the ones which I just connected to ok are :-

hpb.mcc.ac.uk 130.88.200.7
banjo.concert.net 192.101.21.6
tsx-11.mit.edu 18.172.1.2
nic.funet.fi 128.214.6.100
puffin.doc.ic.ac.uk 146.169.3.7

Hope this helps,

John.
([email protected])


[next article]
From: [email protected] (Tim Martin; FSO; Soil Sciences)

Subject: question: kernel vs. image
Message-ID:
Date: 14 May 92 22:22:29 GMT
Sender: [email protected] (News Administrator)
Organization: University of Alberta, Edmonton, Canada
Lines: 12
Nntp-Posting-Host: menaik.cs.ualberta.ca

I use shoelace to boot from my hard disk, and have the Image file in
the /etc directory. But I see from the shoelace "config" file that one
might instead have the separate system parts - kernel, mm, fs, init -
in /etc/system. When I boot my system, it looks there first, can't
find kernel, and then loads the Image. My question to the *Real Linux
Hackers*: do you simply use the Image file, as I have done, or do you
make the separate parts? And if the latter, how? I confess, I'm
a rank beginner when it comes to compiling the system. So far, I have
learned to type "make", and now even "make clean", "make dep", and
"make Image", then watch the messages fly bye. ๐Ÿ™‚

Tim.


[next article]
From: [email protected] ("Mark W. Eichin")

Subject: Re: Security and "X" don't mix?
Message-ID: <[email protected]>
Date: 14 May 92 23:51:14 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected] ("Mark W. Eichin")
Organization: The Internet
Lines: 15

Simon Marlow writes:
| This is not an infallible solution, especially if other users can
| rlogin to your machine and completely circumvent the xauth mechanism.
Once you've given the server a key, simply having non-root
access to the local machine will not be enough to circumvent it (I get
bitten by this occasionally; I su, run emacs, and it fails because I
haven't set XAUTHORITY to point to my .xauthority file.) So it *is* a
reasonable protection against people sharing a machine, as long as
they don't have root.
It's still not a protection against someone monitoring the
net; the "magic-cookie" is passed in the clear... I haven't looked at
the DES mode of xauth, but it should solve that problem.
_Mark_
MIT Student Information Processing Board
Cygnus Support


[next article]
From: [email protected] (Mika Pekka Liljeberg)

Subject: Re: Abysmal disk performance
Message-ID:
Date: 15 May 92 00:12:14 GMT
References: <[email protected]>
<[email protected]>
Sender: [email protected] (Uutis Ankka)
Organization: Department of Computer Science, University of Helsinki, Finland
Lines: 31
In-Reply-To: [email protected]'s message of 14 May 92 01: 00:44 GMT

In article <[email protected]> [email protected] (Doug Thomps
on) wrote:
> As I remember the 8237 DMA clicks along at 3 Mhz on the original AT
> coupled with
> 10-15 wait states. I don't know how many wait states used on 386 boxes today
> but I believe the DMA is still at 3Mhz. Starting with the 286 and beyond
> doing a REP INS would transfer at bus speed.
>
> The DMA circuits on the AT just could not provide very fast transfers, thus
> the POLL input on the WD controller and childern. That is why Adaptec went
> with the DMA master model. The 1542 would utilize the DMA handshake lines
> itself and not use the on board 8237's.
>
> I might be out of date, but thats what I remember from days working on 286s
> and early 386s

Well, I've got a 25 MHz C&T chipset in my machine and a fancy setup (AMI bios),
which lets me play with the DMA clock, I/O waits and some other great stuff.

My DMA clock is currently set to Processor clock/4, which makes it about
6.25 MHz. I don't recall the DMA waits, but the longest wait the setup
offers ANYWHERE is 4 wait states, so there.

I'm only saying this to point out that there might be SOME people, who could
use DMA with a hard disk controller. It would be great, if somone gave it a
shot.

Mika
--
Mika Liljeberg Email: [email protected]
Helsinki University [email protected]
Dept. of Computer Science


[next article]
From: [email protected] (Rick Kelly)

Subject: Re: If it walks like a duck, quacks like a duck.....
Message-ID: <[email protected]>
Date: 14 May 92 17:32:40 GMT
References: <[email protected]> <[email protected]
dcs.claremont.edu> <[email protected]> <[email protected]
laremont.edu>
Reply-To: [email protected] (Rick Kelly)
Organization: The Man With Ten Cats
Lines: 72

In article <[email protected]> [email protected]
ont.edu (Jim Winstead Jr.) writes:
>In article <[email protected]> [email protected] (Rick Kelly) writes:
>>The GNU suite of of free software will all be available for Coherent 4.0
>>at some point.
>
>I'm sure they will be - they've even made it as far as DOS, so there's
>no holding them back from a real OS like Coherent 4.0. ๐Ÿ™‚

Yes. Finally no more small model code limit.

>>The ability to run COFF binaries is a big win for Coherent 4.0.
>
>I'm still not sure about this one - as someone pointed out, I think
>Coherent would be better served by an ANSI compiler. In many ways, I
>think shrink-wrapped applications for Unix(tm) machines are silly. If
>you want something to load WordPerfect and Lotus 1-2-3, use DOS.

Coherent 4.0 can also be used as a development environment for SVR3.2
systems. This is important to me because it means I can create binaries
that will run on SCO UNIX. I am a self-employed computer consultant,
and I would rather have a machine devoted to SVR4 than SCO UNIX. This
new version of Coherent will make that possible.

An ANSI compiler is important, but that will be GCC 2.1.

>>>I really would hate to see this sort of petty bickering transpire
>>>between Coherent and Linux users, because in many ways Linux and
>>>Coherent are kindred spirits.
>
>>Just out of curiosity, why have the Linux newsgroups been so quiet
>>lately? Have a lot of people jumped to the Jolitz BSD port?
>
>There's only one Linux newsgroup, now that alt.os.linux is obsoleted
>by comp.os.linux. It's been a little quieter lately, I believe,
>because people are going off for summer vacation, and it's been a
>while since a major Linux release. That is, until just a couple of
>days ago, when 0.96 came out. SCSI support is now part of the
>standard kernel, among other things

There are still postings every now and then in alt.os.linux. It will
probably take years to get rid of it. You guys should have gone the
comp route from the start.

>Also, now that Linux is more stable (well, at least for the releases
>with an alphabetic character after them :), there's less of the flood
>of bug reports. It's really amazing the number of times the same bug
>is reported sometimes.

True.

>>They have always had a 1200 page manual, which has to be updated for
>>the latest release.
>
>Yes, I knew this, but it just seemed humorous the press release
>mentioned the 1200 page 'critically acclaimed' manual that hadn't been
>printed yet. I'm sure there's lots of overlap between the old and new
>manuals.

And, of course, it is typical "marketing speak".

>Followup-To is set to comp.os.linux so as to divert this conversation
>where it belongs. I don't want to be blamed of cheating MWC of any
>sales.... ๐Ÿ™‚

I expect there will be plenty of sales. ๐Ÿ™‚

And there are still people out there, with a 286 hanging around, who will
want the 286 version. That may have a couple of years left in it.

--

Rick Kelly [email protected] unixland!rmkhome!rmk [email protected]


[next article]
From: [email protected] (Martin Rausche)

Subject: X, Speedstar, NEC II Problems.
Message-ID: <[email protected]>
Date: 14 May 92 12:32:31 GMT
Organization: CSD., University of Erlangen, Germany
Lines: 47

Hi there!

I have really great problems with my equipment and X. Let me tell you, what
I've done.

I installed the interim version of 0.95c+ two weeks ago. This week I got
X and the 0.96 bootimage. Now I patched the bootimage at offset 508h to
boot from my harddisk. All things are all right, sometimes my computer
boots once again after printout of "8 virtual ...", but that's not the
problem. Now I want to startx ๐Ÿ™‚
First problem: No /dev/tty0, hmm, I said:
mknod /dev/tty0 c 4 0
and now it works.
But I cannot get some useful information on my screen. The monitor is
complete out of sync :-X I get a lot of chessboard-like patterns on my
screen and some shifted text. ARGH. OK, I thought, no problem. I
tried a few clock settings and monitor settings, but I cannot get it
to work. My monitor always shows rubbish.
A cannot type anything or move my mouse around, because the machine
freezes. Maybe a problem mith my logitech mouse, I will check this out,
but I want to see a xterm or something like that ๐Ÿ™‚

My equipment:
486/33 AMI Bios 5/5/91
Diamond Speedstar VGA (no +, no hicolor !!!)
NEC MultiSync II (not 2A or something like that)
Quantum LPS 240A
4MB (+8MB swap)
Soundblaster 1.0
Logitech Pilot Mouse

Please tell me, where I can get the program clock.exe (Does it help?) or
can someone tell me the right clock settings for my VGA Card? Can
someone mail me a copy of a working Xconfig file? I really need help,
because I don't want to shoot my monitor ๐Ÿ™‚

Please answer by mail, because our News expire system is very strict and
I cannot read new postings for the next 4 days.

Thanx. Martin.

---
------------------------------------------------------------------------------
| Martin Rausche |
| EMail: [email protected] |
| Trust me, I know what I'm doing -- Sledge Hammer |
------------------------------------------------------------------------------


[next article]
From: [email protected] (Steven Kraft)

Subject: Re: Linux on LocalBus 486DX33 motherboard
Message-ID: <[email protected]>
Date: 15 May 92 01:41:54 GMT
References: <[email protected]>
Reply-To: [email protected]
Lines: 26
Nntp-Posting-Host: perdita.gac.edu

In article <[email protected]> writes:
>Does somebody has an experience in using a LocalBus 486DX33 motherboard with
>Linux such as found in Zenon or Microlinks computers. Does X11 run with
>the localbus ET4000 VGA board ?
>Thanks in advance
>Thierry PRIOL Phone: 99 84 72 10
>IRISA / INRIA U.R. Rennes Fax: 99 38 38 32
>Campus Universitaire de Beaulieu Telex: UNIRISA 950 473F
>35042 RENNES CEDEX - FRANCE E-mail: [email protected]

I am using Linux on an Orchid Superboard 486/33, and it has a direct bus video
slot in which I use an Orchid Fahrenheit 1280/D. It runs 0.12 with no
problems, and I am in the process of ftping the 0.96 files and will know in a
while how that works.

X11 should run fine on a localbus ET4000 card.

But X11 won't run on my S3 accelerated Fahrenheit 1280/D until I or someone
else willing to work at it writes an X11 S3 video driver. I will be getting
going on this project after I am done with final exams next week.

--
Steven Kraft [email protected] Micro/MiniComputer Consultant
Gustavus Adolphus College, St. Peter, MN 56082 Comp Sci major
Note: Due to the nature of life, the universe, and everything, the
opinions expressed above may be changed at any time without notice.


[next article]
From: [email protected] (A. V. Le Blanc)

Subject: Re: search path problem -- help.
Message-ID: <[email protected]>
Date: 14 May 92 06:17:12 GMT
References: <[email protected]>
Followup-To: [email protected]
Organization: Computing Centre, University of Manchester
Lines: 21

In article <[email protected]> [email protected] writ
es:
>version. in any given current working directory the programs can
>only be executed by typing ./test and not test (as one would expect).
>the /etc/profile has the line
> export PATH=:/bin://usr/bin:/usr/local/bin

Note first that this is the correct behaviour for root. Anything else
is dangerous.

Note second that, as there is a shell builtin named 'test', even changing
your path will not free you from needing './' in this case.

You may, if you wish, change the line in /etc/profile to

export PATH=/usr/bin:/bin:/usr/local/bin:.

(where I would recommend having the '.' last), but be sure that root
changes this in its own .profile to exclude the '.'.

-- Owen
[email protected]


[next article]
From: [email protected] (Lars Peter Fischer)

Subject: Re: Security and "X" don't mix?
Message-ID:
Date: 15 May 92 01:12:37 GMT
References: <[email protected]>
<[email protected]>

<[email protected]>
Sender: [email protected] (UseNet News)
Organization: Mathematics and Computer Science, Aalborg University
Lines: 31
In-Reply-To: [email protected]'s message of 14 May 92 17:04:36 GMT

>>>>> "Simon" == Simon Marlow ([email protected])

Simon> This is not an infallible solution, especially if other users can
Simon> rlogin to your machine and completely circumvent the xauth mechanism.

How does and rlogin allow one to "completely circumvent the xauth
mechanism"?

With xauth, the server generates a code (called a magic cookie) when
it starts and writes the code to a file (~/.Xauthority) with mode 600.
Other X programs read this file and send the code to the X server
right after connection is established. The server will refuse
connections that do not provide the correct code. The server does not
consider the originating host at all.

Reading the .Xauthority file and sending the code is handled in
XOpenDisplay in XLib, so the use of this is transparent to
programmers.

This system is neat, since in an environment with NFS, I can connect
to my X server right away without using xhost or anything like it.
What's more, for just about all users all this is completely
transparent.

None of this is really relevant to Linux since few persons would use
Linux for multi-user work.

/Lars
--
Lars Fischer, [email protected] | It takes an uncommon mind to think of
CS Dept., Aalborg Univ., DENMARK. | these things. -- Calvin


[next article]
From: [email protected] (Lars Peter Fischer)

Subject: Re: Security and "X" don't mix?
Message-ID:
Date: 15 May 92 01:20:17 GMT
References: <[email protected]>
<[email protected]>

Sender: [email protected] (UseNet News)
Organization: Mathematics and Computer Science, Aalborg University
Lines: 26
In-Reply-To: [email protected]'s message of 13 May 92 21:38:47 GMT


>>>>> "Charles" == Charles Hedrick ([email protected])

Charles> While individuals can protect themselves using
Charles> something like this, in practice setting it up is hard enough that I
Charles> think system managers really need to do it.

In an environment where NFS is used for user homedirectories, 95% of
the time using authentication requires no work at all. We just enabled
it by adding a flag to startx / ow3 scripts -- end of story.

Charles> While I generally believe in open discussions of security issues,
Charles> posting code to break system security is probably a bad idea.

Since all this program does is point out that if you don't use
authentication you will probably have problems and since these
problems are easily avoided, I think posting it is a Good Thing. When
you tell people that they really should use authentication with X,
they will go "um, yeah, I'll look into that" and do nothing. This kind
of program will mae people react.

Don't make the mistake of believing that the bad guys have not had
such programs for years. Security through obfuscation will *not* work.
--
Lars Fischer, [email protected] | It takes an uncommon mind to think of
CS Dept., Aalborg Univ., DENMARK. | these things. -- Calvin


[next article]
From: [email protected] (T. Morley on grange.cl.cam.ac.uk)
Newsgroups: comp.windows.x,comp.os.linux

Subject: New Drivers for X386/Generic X11R5 servers.
Message-ID: <[email protected]>
Date: 14 May 92 12:05:19 GMT
Sender: [email protected] (The news facility)
Reply-To: [email protected] (T. Morley)
Followup-To: comp.os.linux
Organization: U of Cambridge Computer Lab, UK
Lines: 14


I'm thinking of upgradeing my lowly pc-xt at home to 386SX, and running Linux
(the free unix clone) with X11R5, but I already have a Wyse high res grahpics
board, from what has been said onsaid on comp.os.linux, X386 has 8bits of
display fairly well hardcoded into it. Unfortunatley he Wyse is only
monochrome though it is very high res (1280x800). Has anybody written a
driver for this?? If not how difficult would it be to convert either X386 or
the server software that comes with the standard X11R5 distribution.

Thanks.

Tim M.

Please reply by mail as I can't guarentee to read the reply on the 'net


[next article]
From: [email protected] (Daniel D Deavours)

Subject: Re: Abysmal disk performance
Message-ID: <[email protected]>
Date: 14 May 92 21:52:50 GMT
References: <[email protected]> <[email protected]
ay.novell.com>
Sender: [email protected] (Net Noise owner)
Organization: University of Illinois at Urbana
Lines: 17

>> The string instructions (ie portread / portwrite in hd.c) are abysmally slow
>> in comparison to DMA. The same can be said for strcpy(). We should

>As I remember the 8237 DMA clicks along at 3 Mhz on the original AT
>coupled with
>10-15 wait states. I don't know how many wait states used on 386 boxes today
>but I believe the DMA is still at 3Mhz. Starting with the 286 and beyond
>doing a REP INS would transfer at bus speed.

Never programmed a DMA before, but if I'm correct, the advantage may not be
so much speed in data transfer, but the ability to do disk i/o in the back-
ground. It's annoying, especially on my slow hd/controller for everything to
wait until the hard drive is done (especially when syncing.)

I'm very much for that happening.

Dan Deavours


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

Subject: Re: Microsoft mouse works. ALSO, DRDOS 6 users please read.
Message-ID: <[email protected]>
Date: 15 May 92 02:43:06 GMT
References: <[email protected]> <[email protected]
sd.kodak.com>
Organization: Pyramid Technology Corp PL, Melb., Australia
Lines: 28

In article [email protected]
tu-muenchen.dbp.de (Henning Spruth) writes:
>
>>just for reference, stty works on stdout, not stdin, so you need to use
>>stty -istrip > /dev/ttys2 for it to take place.


[discussion of serial mouse stuff deleted]

Will there be support for bus mouse at any stage? Anyone working on this?

2nd. bit:

I am intending to run linux in tandem with DRDOS 6 on my 386. However, it
occurs to me that there is no way mtools will work if the SSTOR drive is
used, since I would guess (without having read the mtools code) that it
goes directly to a disk block and expects vanilla MSDOS hd layout. In which
case it is going to be sorely disappointed.

Anyone have any comments? For example, if anyone has mtools working with
either SSTOR or stacker I would like to know.

I have reloaded my linux partition completely since 0.96 seems too different
to 0.95 to bother upgrading. This entails heavy duty file transfers and if I
can avoid having to rawrite zillions of floppies I will...:-).

Chuck
--
email: [email protected]


[next article]
From: [email protected] (Yuan-Yin Wu)

Subject: GCC 2.1 Compilation Error
Summary: Got fatal signal 6 during compiling
Keywords: GCC 2.1, Gnuplot, Linux
Message-ID: <[email protected]>
Date: 15 May 92 02:51:36 GMT
Sender: [email protected]
Followup-To: comp.os.linux
Organization: Univ. of Texas at Dallas
Lines: 38
Nntp-Posting-Host: navajo.utdallas.edu

Hi, Linuxers:
I was compiling Gnuplot the other day and GCC told me
gcc: Internal compiler error: program cc1 got fatal signal 6
and CC1 said something like "IOT trap". This happened for both the
file graphics.c and graph3d.c (the largest among the files in the
package). I have no clue what all these mean. Could someone please
help. Thanks in advance.
My system:
386sx/16, 8mb ram, 5mb swap, 120mb IDE
runing 0.95c+ (and then 0.96), X

yuanyin
In article <[email protected]> [email protected] (A. V. Le Blanc) wr
ites:
>In article <[email protected]> [email protected] wri
tes:
>>version. in any given current working directory the programs can
>>only be executed by typing ./test and not test (as one would expect).
>>the /etc/profile has the line
>> export PATH=:/bin://usr/bin:/usr/local/bin
>
>Note first that this is the correct behaviour for root. Anything else
>is dangerous.
>
>Note second that, as there is a shell builtin named 'test', even changing
>your path will not free you from needing './' in this case.
>
>You may, if you wish, change the line in /etc/profile to
>
> export PATH=/usr/bin:/bin:/usr/local/bin:.
>
>(where I would recommend having the '.' last), but be sure that root
>changes this in its own .profile to exclude the '.'.
>
> -- Owen
> [email protected]


--
Yuan-yin Wu [email protected] (214) 690-2152


[next article]
From: [email protected] (Theodore Ts'o)

Subject: Re: Security and "X" don't mix?
Message-ID: <[email protected]>
Date: 15 May 92 03:13:45 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 25


From: [email protected] (Lars Peter Fischer)
Date: 15 May 92 01:20:17 GMT

Charles> While I generally believe in open discussions of security issues,
Charles> posting code to break system security is probably a bad idea.

Don't make the mistake of believing that the bad guys have not had
such programs for years. Security through obfuscation will *not* work.

Indeed. These programs have been around since the X10 days. In fact, I
remember writing such a program back when I was a freshman; it was a
great way to learn about Xlib programing! Several of my friends each
wrote their own version of this type of program as well ---
collectively, we called them "window mismanagers". ๐Ÿ™‚

The real solution is *not* to xhost hosts unless you absolutely have to,
and if possible, have only one person logged into a workstation at a
time. If that's the case, you're pretty much safe. Given how most
people will be using Linux, I seriously doubt this "security problem"
will be much of a problem under Linux. It is, however, a pretty generic
problem, and things like the magic cookie xauth do a pretty good job of
fixing it under X11.

- Ted


[next article]
From: [email protected] (Daniel Jimenez)

Subject: elvis
Message-ID: <[email protected]>
Date: 15 May 92 03:34:18 GMT
Organization: University of Texas at San Antonio
Lines: 6

I just got Linux working last week, and I'm trying to see which binaries
my computer will run. I have three megabytes on my 386/16. I can run
tcsh fine, but elvis (from tsx-11) just says "Memory fault." What's
the deal? Does elvis actually need more than 3 megs to run in?
--
Daniel Jimenez


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

Subject: Re: rebooting problem
Message-ID: <[email protected]>
Date: 15 May 92 00:07:32 GMT
References:
Sender: [email protected]
Organization: Design Computing Unit, The University of Sydney
Lines: 14
Nntp-Posting-Host: dean.arch.su.oz.au

In article , [email protected]
.edu (Charles Hedrick) writes:
|>
|>
|> Like some other people I find that I have to do a hardware
|> reset when rebooting. Otherwise the system crashes near the
|> beginning of startup.

Could this problem be caused by the keyboard input not being cleared and,
therefore, causing the control-alt-del to be read again.

--
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]

Subject: Re: grrrrr RELAY
Message-ID: <[email protected]>
Date: 14 May 92 15:35:36 GMT
References: <[email protected]>
Organization: Oxford University VAXcluster
Lines: 45

In article <[email protected]>, [email protected]
.ac.uk (Jon Jagger) writes:
> Can anyone help ?
>
> I've been looking at the stuff at tsx-11.mit.edu and I managed to get
> enough stuff to make a root and boot disk. (doesn't it look good).
>
> I've decided that I'll put it on a hard disk that I just happen to
> have spare. My problem is ftp. I can only get the stuff via a janet
> relay, and our stream is so piddlingly small, that I have not been
> able to get back onto the relay, let alone, mit, for four days now!
>

>...

>
> Thanks Jon Jagger
>
> --------------->8------------------>8----------------------
> Jon Jagger
> Sheffield City Polytechnic
> Tel 0742 533802
> Fax 0743 533840
> ๐Ÿ™‚
> --------------->8------------------>8----------------------

In the UK you can get stuff via Janet (or JIPS) from Ipmerial College
Janet :- UK.AC.IC.DOC.SRC
(JIPS) ftp src.doc.ic.ac.uk

look in directory /os/Linux/

Hope this helps.

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?
==================================================\=|=/=======================
\|/


[next article]
From: [email protected] (James Banks)

Subject: Logitech Bus Mice
Message-ID:
Date: 15 May 92 05:06:06 GMT
References: <[email protected]> nik.tu-muenchen.de> <[email protected]>
Organization: Brigham Young University, Provo UT USA
Lines: 24
Nntp-Posting-Host: fuji.ee.byu.edu



Hi, again!

My logitech bus mouse driver is coming along, but I've run into another
snag. The PD Logitech Bus mouse drivers' code for activating interrupts
doesn't seem to work. Does anyone know how to do this (i.e. have access
to actual manuals?) Once I know that, the rest should be easy (I hope ๐Ÿ™‚

Thanks to those who replied to my last post.

BTW: I'm guessing that the gate (in set_intr_gate) for IRQ 5 is 0x25.
Is this correct?

Thanks again,
James
[email protected]

()(())((()))(((())))((((()))))(((((())))))((((((()))))))(((((((())))))))(((((((
James A. Banks _____ ____
[email protected] _____|||_____|| |\
| | @ || | \ @@
(.sig under construction) //|\//|/////\/\/\/\\\/\\\//\
(()))))))))(((((((((())))))))))((((((((((()))))))))))(((((((((((())))))))))))((


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

Subject: Re: Sources (IMPORTANT to managers of LINUX ftp sites)
Message-ID: <[email protected]>
Date: 15 May 92 06:19:04 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 12

in article <[email protected]>, [email protected] (Theodo
re Ts'o) says:

> In priciple, I agree with what the FSF is trying to do --- and for
> programs where there have been serious changes made to do a "port", I
> might even agree that it would be a good thing to upload the sources

Like gcc itself?
--
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] (Jim Burns)

Subject: Re: source codes for libs and gcc 2.1
Message-ID: <[email protected]>
Date: 15 May 92 07:06:41 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 15

in article <[email protected]>, [email protected] (H
.J. Lu) says:

> BTW, if you read my FAQ for gcc 2.1, you may have known where the source
> codes of the old gcc 2.1 and libs are. Next time, please READ my FAQ
> and README for gcc 2.101. They are very important. If you didn't, you
> would lose.

I just got the FAQ from banjo.concert.net:pub/Linux/GCC, and it doesn't say
anything about source code. I see the lib-2.1.tar.Z in that dir, but nothing
for the compiler. Could you repost location for the compiler source? Thanx.
--
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] (Magnus Alvestad)

Subject: Re: GCC 2.1 Compilation Error
Message-ID:
Date: 15 May 92 08:02:19 GMT
References: <[email protected]>
Sender: [email protected] (Bergen University Newsaccount)
Reply-To: [email protected]
Followup-To: comp.os.linux
Organization: AlveSoft
Lines: 20
In-Reply-To: [email protected]'s message of Fri, 15 May 1992 02:51:36 GMT


Hi, Linuxers:
I was compiling Gnuplot the other day and GCC told me
gcc: Internal compiler error: program cc1 got fatal signal 6
and CC1 said something like "IOT trap". This happened for both the
file graphics.c and graph3d.c (the largest among the files in the
package). I have no clue what all these mean. Could someone please
help. Thanks in advance.
My system:
386sx/16, 8mb ram, 5mb swap, 120mb IDE
runing 0.95c+ (and then 0.96), X

First thing you should check is the presence of any ASCII NULs in the
source files or in an include file. Those cause that error. If not,
run the file first through cpp, then cc1 and then as and see what
happens at every step. Good luck.

--
| Magnus Alvestad | I am a prolog program. I get all my opinions
| [email protected] | from a small organization in Birmingham.


[next article]
From: [email protected] (BURNS)

Subject: Re: 0.96 (boot) and mcc-interim, shell problem help.
Message-ID: <[email protected]>
Date: 15 May 92 07:40:38 GMT
References: <[email protected]>
Distribution: na
Organization: Georgia Institute of Technology
Lines: 15

in article <[email protected]>, [email protected] sa
ys:

> In the login shell the command
> ls -l /dev | more
> does nothing prompt comes back.

Are you using bash? And is ls redefined as a function? I find I have to put
functions in a subshell to work, eg -

(ls -l /dev) | more
--
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] (BURNS)

Subject: Re: Linux rebooting
Message-ID: <[email protected]>
Date: 15 May 92 07:48:01 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 12

in article <[email protected]>, [email protected] (
Drew Eckhardt) says:

> Since the bootstrap loader does not load a normal a.out header file,
> like BSD's does, it doesn't know the size / location of BSS.

The bootstrap loader doesn't have to. That can be put in init code in the
kernel itself, which would know where the bss is (_edata?).
--
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] (Richard Sharman)

Subject: looking for source of xc or xcomm
Message-ID:
Date: 15 May 92 03:29:46 GMT
Sender: [email protected]
Distribution: comp
Organization: Mitel. Kanata (Ontario). Canada.
Lines: 9

Hi. I've not been able to get pcomm to work properly, and can't quite
get minicom (from Minix) to work under Linux. So I downloaded xcomm
from tsx-11 and it works well. But I'd like to make a few changes,
to have the BS/DEL key send a delete, swap ^S with ^\ etc, and so
would like the source. It doesn't seem to be at tsx-11; could anyone
please tell me where I could get it?

Thanks.
richard


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

Subject: Berkely Yacc runs without changes
Message-ID: <[email protected]>
Date: 14 May 92 21:05:02 GMT
Sender: [email protected] (Newsfiles Owner)
Reply-To: [email protected] (Michael Haardt)
Organization: Gandalf - a 386-20 machine
Lines: 12
Nntp-Posting-Host: messua

Just for completeness and because yacc tables are smaller than bison, I
got berkely yacc from the free bsd sources and it compiles and works on
a 0.95a kernel with ps and lp patches and gcc 1.40 without changes.

I know that this is an old configuration, but the last weeks there are
problems with the FDDI ring in Aachen, so that I can't ftp the new stuff
๐Ÿ™ mail and news work, but with lags.

I compiled rc and it also works wonderful. I think I found my favourite
shell, since rc has support for command line editing. ๐Ÿ™‚

Michael


[next article]
From: [email protected] ("Michael Zacherl")

Subject: What SCSI Controllers are supported (FAQ)
Message-ID: <[email protected]>
Date: 15 May 92 08:49:29 GMT
Sender: [email protected]
Reply-To: [email protected]
Organization: Mail to News Gateway
Lines: 9

Hi *!

Reading through this board I can't really figure out , what SCSI-Controllers
are supported. I got a WD 7000 FASST with a 650MB Disk.
Can Linux deal with that ??

I think that's a FAQ .. Couldn't find an answer ๐Ÿ™
Thanks,
nichael


[next article]
From: [email protected] (Anthony Rumble)

Subject: Re: HD corruption with new 0.96 kernel
Message-ID:
Date: 15 May 92 08:58:51 GMT
References: <[email protected]
.uwa.edu.au>
Sender: [email protected]
Organization: Sydney University Computing Service, Sydney, NSW, Australia
Lines: 36
Nntp-Posting-Host: extro.ucc.su.oz.au

[email protected] (Michael O'Reilly) writes:

>[email protected] (Anthony Rumble) writes:
>: Since I installed the new Kernel.. I have been getting some wierd
>: things happening with GCC 2.1..
>Tell me about it. ๐Ÿ™ I have just finished re-building my partition
>from scratch. Fsck gave me nearly 4000 lines of errors. ๐Ÿ™
>At that point I decided to retrieve what I could and build a new file
>system. This occurred sometime after my switch to the 0.96 kernel. I had
>been doing a lot of compileing..
.rK! yeah.. well strangely enuf.. fsck diddn't come up with any errors!
But things were definately corrupted!

>I am running 486/33 with 4 megs, 8 megs swap, 200 meg IDE hard disk.
>I too , had to comment out the 'Unexpected hard disk interrupt'
>message. It was comeing up about 25 times a second during a compile....

Yes.. Same with me! And.. Even AT&T SYS V rel 3.2 had the SAME kind of errors (U
nexoected HD interupt)!! So.. Im unsure if this is a software or hardware proble
m..

>: Also.. One other thing.. I heard mention that Linux can only handle
>: 32 Meg File systems...
>Nahh. Linux currently handles up to 65535 blocks..
>My partition is this size (all my linux stuff is on the one partition).

Oh.. thats a relief.. (Bit late tho.. I just repartitioned my HD.. Oh well.. i n
eeded a clean out anyways :))

>: Hope someone can help me..
>: Linus.. keep up the good work..

>Well, I can't help. ๐Ÿ™ As an aside, does anyone have detailed info on
>programming the IDE hard disk controller ???. This 'Unexpected HD error'
>is bugging me. Damned if I know what is causeing it..

I mentioned the problem to Kylok (Through my dealer) and I havent recieved the r
emply as yet.. I think i might do some more enquiries.. Maby try and even get th
rough to Kylok themselves direct for an answer..

Anthony Rumble


[next article]
From: [email protected] (Duco Fijma)

Subject: Re: 0.96 bugs?
Message-ID: <[email protected]>
Date: 15 May 92 10:00:14 GMT
References: <[email protected]
a.edu.au>
Sender: [email protected]
Organization: University of Twente, Dept. of Computer Science
Lines: 23
Nntp-Posting-Host: utis143

In article <[email protected]>, you write:
|> "Alex R.N. Wetmore" writes:
|>
|> >I have had a few problems with the 0.96 kernel. I can't login as anyone
|> >other then root (or su), I get a general protection fault. Also, I get
|> >bogus do_no_page errors all the time.
|>
|> Unless I miss my guess, this is the same problem that I had.
|> It was solved by [email protected] The problem was that my
|> shared libraries were only rx by root. Making them og+rx saved
|> the seg vees.

I'm having the same problem as the original poster. This hints did not helped
much. My shared libs are perfectly readable by everyone, and still the general
protection errors when logging in as non-root.

Also, 'ps' and 'free' give 'lseek' errors on 'kmem' under 0.96. Everything worke
d
fine under 0.95c+. Any help??

Duco Fijma
University of Twente
Enschede
the Netherlands


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

Subject: libobz
Message-ID: <[email protected]>
Date: 13 May 92 04:16:48 GMT
Sender: [email protected]
Organization: Computing Services, UC Davis
Lines: 10

Is there a copy of libobz.a someplace? Even souces would be sufficient.
Without this lib none of the X stuff will compile!!!



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



[next article]
From: [email protected] (Michael O'Reilly)

Subject: Re: 0.96 bugs?
Message-ID: <[email protected]>
Date: 15 May 92 11:27:11 GMT
References: <[email protected]>
Sender: [email protected] (USENET News System)
Organization: University of Western Australia
Lines: 37
Nntp-Posting-Host: tartarus.uwa.edu.au

[email protected] (Duco Fijma) writes:
: In article <[email protected]>, you write:
: |> "Alex R.N. Wetmore" writes:
: |>
: |> >I have had a few problems with the 0.96 kernel. I can't login as anyone
: |> >other then root (or su), I get a general protection fault. Also, I get
: |> >bogus do_no_page errors all the time.
: |>
: |> Unless I miss my guess, this is the same problem that I had.
: |> It was solved by [email protected] The problem was that my
: |> shared libraries were only rx by root. Making them og+rx saved
: |> the seg vees.
:
: I'm having the same problem as the original poster. This hints did not helped
: much. My shared libs are perfectly readable by everyone, and still the general
^^^^^^^^ They MUST be executable as
well. Damned if I know why. ๐Ÿ˜‰ If things are readable, youcan always
execute them. Ahh.comprehension strikes. Setuid programs can be readable,
but not executable, and this makes sense. setuid libraries?? The mind
boggles.....


: protection errors when logging in as non-root.
:
: Also, 'ps' and 'free' give 'lseek' errors on 'kmem' under 0.96.
: Everything worked fine under 0.95c+. Any help??

Yup. Compile the kernel and type
'ps U /src/linux/tools/system'. Assuming the your kernel sources are in
/src/linux. This rebuilds the ps database. You must do this every time
you change the kernel.

:
: Duco Fijma
: University of Twente
: Enschede
: the Netherlands


[next article]
From: [email protected] (Michael O'Reilly)

Subject: Re: HD corruption with new 0.96 kernel
Message-ID: <[email protected]>
Date: 15 May 92 11:34:18 GMT
References:
Sender: [email protected] (USENET News System)
Organization: University of Western Australia
Lines: 26
Nntp-Posting-Host: tartarus.uwa.edu.au

[email protected] (Anthony Rumble) writes:
: [email protected] (Michael O'Reilly) writes:
:
: >Tell me about it. ๐Ÿ™ I have just finished re-building my partition
: >from scratch. Fsck gave me nearly 4000 lines of errors. ๐Ÿ™
: >At that point I decided to retrieve what I could and build a new file
: >system. This occurred sometime after my switch to the 0.96 kernel. I had
: >been doing a lot of compileing..
: .rK! yeah.. well strangely enuf.. fsck diddn't come up with any errors!
: But things were definately corrupted!
:
: >I am running 486/33 with 4 megs, 8 megs swap, 200 meg IDE hard disk.
: >I too , had to comment out the 'Unexpected hard disk interrupt'
: >message. It was comeing up about 25 times a second during a compile....
:
: Yes.. Same with me! And.. Even AT&T SYS V rel 3.2 had the SAME kind of
: errors (Unexoected HD interupt)!! So.. Im unsure if this is a software or
: hardware problem..

SIGH!!!!. I was hopeing it was a software problem, but if a different OS
shows it too... ๐Ÿ™ doesn't look good.


: Anthony Rumble
Anthony, LEARN TO HIT RETURN WHEN YOU HAVE TYPED 72 CHARS!!. Now that I
have that off my chest. Long lines are very annoying when your
terminal's automarging is slightly broken...


[next article]
From: [email protected] (Fisher S A)

Subject: Re: grrrrr RELAY
Message-ID:
Date: 15 May 92 10:37:46 GMT
References: <[email protected]>
<[email protected]>
Sender: [email protected]
Organization: Organised? Never!
Lines: 49
In-reply-to: [email protected]'s message of 14 May 92 15:35:36 GMT

In article <[email protected]> [email protected] writes
:

|In article <[email protected]>, [email protected]
y.ac.uk (Jon Jagger) writes:
|> Can anyone help ?
|>
|> I've been looking at the stuff at tsx-11.mit.edu and I managed to get
|> enough stuff to make a root and boot disk. (doesn't it look good).
|>
|> I've decided that I'll put it on a hard disk that I just happen to
|> have spare. My problem is ftp. I can only get the stuff via a janet
|> relay, and our stream is so piddlingly small, that I have not been
|> able to get back onto the relay, let alone, mit, for four days now!
|>
| >...
| >
|> Thanks Jon Jagger
|>
|> --------------->8------------------>8----------------------
|> Jon Jagger
|> Sheffield City Polytechnic
|> Tel 0742 533802
|> Fax 0743 533840
|> ๐Ÿ™‚
|> --------------->8------------------>8----------------------
| In the UK you can get stuff via Janet (or JIPS) from Ipmerial College
|Janet :- UK.AC.IC.DOC.SRC
|(JIPS) ftp src.doc.ic.ac.uk
| look in directory /os/Linux/
| Hope this helps.
| 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?
|==================================================\=|=/=======================
| \|/
No you can't. Imperial does not have the MCC interim stuff. It does
not even have 0.96 on it. It is very out of date.


____ ____ ____ E-Mail [email protected]
/ / / / 3rd Year Computing Science Undergraduate
/___ /___/ /____ at the University of Essex
/ / / /
___/ / / / Earth shutting down in 5 minutes,
please save all files and log out


[next article]
From: [email protected]

Subject: Shared libs
Message-ID: <[email protected]>
Date: 15 May 92 13:26:44 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 14

|In article <[email protected]> [email protected] (Joel M. Hoffm
an) writes:
|
| I think that anything really vital should be statically linked. As I
| see it, this would include a shell, mount, rm, cp, sync (!) and
| perhaps tar (that's a killer) or dd.

This is something that Sun do on SunOS. Funnily enough, they don't include
login, getty, su, etc in the statically-linked list either. Or halt. So
you can get in to fix your system, but you've got to batter you way in to
do it:-(.
-------------------------------------------------------------------------------
Steve Kilbane, Systems Design, CEGELEC PROJECTS Ltd, Boughton Road, Rugby,
CV21 1BU, England. Tel: +44 788 563563 x3493. email:[email protected]
main(void){int i;char c;for(i=0;c = i["] && c != i["]; i++)(void)putchar(c);}


[next article]
From: [email protected] (Orest Zborowski COMP)

Subject: Re: libobz
Message-ID: <[email protected]>
Date: 15 May 92 12:21:30 GMT
References: <[email protected]>
Organization: Printer Products Eastman Kodak
Lines: 22

[email protected] (Steve Ansell) writes:
>Is there a copy of libobz.a someplace? Even souces would be sufficient.
>Without this lib none of the X stuff will compile!!!
>

with hlu's latest gcc2.1 stuff you don't need libobz.a.

for those without, its in the pre0.96-obz cdif stuff (right on the
top level). just compile it and make a symlink from /usr/lib.

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

zorst
--
zorst (orest zborowski)
[reply to [email protected]]


[next article]
From: [email protected] (Erik J. Arthur)

Subject: Segmentation faults
Message-ID: <[email protected]>
Date: 15 May 92 15:09:56 GMT
Sender: [email protected] (Usenet News Administration)
Organization: University of Minnesota
Lines: 21
Nntp-Posting-Host: dexter.psych.umn.edu

Hi,

Thanks for all the help with the pcomm problems.

When I try to run xv or xcalendar I get a "Segmentation fault"
error. What's happening? All of the rest of the X stuff that
came from zorst (kudos, by the way) runs fine. I set all of the permissions
on the libs to 755. My am running .96 with 4 mb RAM, 8 mb Swap, AMI bios.
I also upped the number of ptys to 8. I hope this is a trivial problem, and
I'm sorry to waste the bandwidth, but I have no clue how to solve it.

Thanks in advance,

Erik

-----------------------------------------
Erik Arthur [email protected]
-----------------------------------------
Phone : (612) 626 - 7521
(612) 626 - 1063
-----------------------------------------


[next article]
From: [email protected] (Kevin Cummings)

Subject: Re: Linux on LocalBus 486DX33 motherboard
Message-ID: <[email protected]>
Date: 15 May 92 15:54:31 GMT
References: <[email protected]> <[email protected]
du>
Sender: [email protected] (Kevin Cummings)
Organization: Prime Computer R&D
Lines: 15

AS the proud owner of a Diamond Stealth VRAM card, PLEASE don't make your
S3 X server Orchid specific!

=================================================================
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] (Ian Wells)

Subject: Re: Abysmal disk performance
Message-ID:
Date: 15 May 92 13:24:47 GMT
References: <[email protected]>
<[email protected]>
<[email protected]>
Sender: [email protected] (The news facility)
Organization: CompScis Anonymous, Cambridge branch
Lines: 29
In-reply-to: [email protected]'s message of 14 May 92 21:52:50 GMT

In article <[email protected]> [email protected]
(Daniel D Deavours) writes:

Never programmed a DMA before, but if I'm correct, the advantage may not be
so much speed in data transfer, but the ability to do disk i/o in the back-
ground. It's annoying, especially on my slow hd/controller for everything
to wait until the hard drive is done (especially when syncing.)

Yes, but everything doesn't just stop when disk access happens. The
only process that does stop running is he one using the disk. This
means that if you run disk-intensive stuff in one VC you can leave it
going and get on with your editing, etc. Basically anything that
doesn't access your hard disk much will ruin quite fast if the other
process is I/O bound.

Oh yes, 2 more questions to ask, while I think about it:
1. When you load, say, gcc twice (sepearately, as opposed to a fork()
call, is what I'm getting at), do the code and const data sections get
shared?
2. Is there a technical name for when the hard disk sits
reading/writing two separate sections of the disk some distance apart
and spends most of its time seeking? This is something annoying I've
tended to notice with Linux, but if it now does readahead on files it
may be improved in this respect.

I don't have my computer with me this term, in the hope that I'll do
some work. It's not working.


Ian Wells.


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

Subject: Re: GCC-2.1 libm (& libsoft)
Message-ID: <[email protected]>
Date: 13 May 92 15:23:39 GMT
References: <[email protected]>
Sender: [email protected]
Reply-To: [email protected]
Organization: NRAO
Lines: 26

In article [email protected], [email protected] (DAVE EINSTEIN) writes:
> Has anybody banged HARD on the GCC-2.1 math routines?
>
> I have recently found time to upgrade to gcc 2.1 While recompiling
>all of the stuff that I had compiled using 1.40. Most things seemed to go
>OK (compiled with slight changes in include files etc. and ran without
>glaring errors) until i recompiled PARI-1.36 (A Number Theory package).
>It compiled smoothly, but would crash on some of the test cases, and give
>incorrect answers on others. Linking with libsoft instead of libm gave fewer
>crashes but more wrong answers. After some fiddling i found that the crashes
>were being caused by incorrect intermediate values being fed to library functin
s
>i.e. negative numbers being fed to log().
> What is puzzling is that PARI compiled with 1.40 worked flawlessly,
>and other programs compiled with 2.1 seemed to work ( I compiled gmp which
>and other programs compiled with 2.1 seemed to work ( I compiled gmp which
>also uses some of libm and it has worked for a few hours).
> In short have any of you people who have ported f2c, gnuplot, or other
>computationally intensive programs detected wierdness?

Does anyone run the ELEFUNT package on the maths libraries? This would be
rather more efficient at detecting errors and loss of precision (and locating
the cause) than running numerically intensive applications. Checking the
software floating-point package against paranoia might also be useful.

Chris Flatters
[email protected]


[next article]
From: [email protected] (Chris Flatters,208,7209,homephone)

Subject: Re: Compiling Ghostscript 2.4.1
Message-ID: <[email protected]>
Date: 14 May 92 19:10:11 GMT
References: <[email protected]>
Sender: [email protected]
Reply-To: [email protected]
Organization: NRAO
Lines: 8

In article [email protected], [email protected] (R.J.Newton) writes:
> Groff only outputs postscript code and I don't have a postscript printer !

Not true! groff -Tdvi will output TeX dvi. There is a dvi to LaserJet translato
r in
the Linux TeX distribution.

Chris Flatters
[email protected]


[next article]
From: [email protected] (Peter MacDonald)

Subject: X, oleo, and shoelace
Message-ID: <[email protected]>
Date: 15 May 92 17:25:07 GMT
Sender: [email protected]
Organization: University of Victoria, Victoria, BC, CANADA
Lines: 32
Nntp-Posting-Host: sanjuan.uvic.ca

Apologies in advance for the potpourri, but this newsgroup
is getting so busy, I am trying to minimize my postings.

First, seems I can not start X as root anymore. The video mode is
so screwed up, my screen shuts down. User mode works fine.
Also the "su" command doesn't work under X. I see the password
as I type it, but it rejects my login. Could be old SW, as I haven't
got the root disk lately.

Next, has anyone else who has no coprocessor, tried running "xv"?
It works fine for me until I go into the "coledit" area, then
my machine locks up tight. Could this be a floating point problem?

Oleo has a big problem. Copying areas doesn't usually adjust cells correctly.
I notice copying above the current area works, but below only adjusts the
column address and not the row address (aaargh). I will look at it this
weekend.

Now shoelace. The port I did of shoelace is partial, at best. Loading
kernel by parts is definitely a Minix anachronism. Also setting
bootdev, rootdev, etc may or may not work since device numbers do
not always map exactly from Linux to Minix. I have used shoelace since
I ported it, but don't get to attached to it. It reads the file system
directly and only understands Minix FS type. When VFS comes, and
you use a big partition as your boot partition (to get long filenames),
it will fail.

And of course, I provided almost no documentation with the port, mainly
because most of the features don't work. The only reason I ported it at
all was that at the time, many utilities were missing from Linux, so I
had to keep my Minix partition around, which meant I couldn't repartition
my drive for a Linux boot partition.


[next article]
From: [email protected]

Subject: Weird terminals
Message-ID: <[email protected]>
Date: 15 May 92 17:53:33 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 13

I have this weird terminal that can only be configured for even, odd, or mark pa
rity.
If set for even, it automatically goes to 7bit, if odd, 8bit. I have it hooked
to my 386, (and yes, I did set up a getty on ttys2), and I would like to use
it. Unfortunately, I don't know how to tell linux, getty, or ttys2 which
parity to use. If someone can help, I would appreciate it. Thanks. ๐Ÿ™‚

BTW, I have cross-mailed this to both the digest, and to the linux-serial list.
I am working on a serial.FAQ for linux, and this will be in it(hopefully)
I have tried to mail to [email protected] but I am getting some
heinous and awful smtp error messages from the attmail postmaster acct.
If you can contact me, I would appreciate it. Merci, Jim.

Jim Gifford [email protected]


[next article]
From: [email protected] (Peter Belding)

Subject: Re: GCC 2.1 Compilation Error
Keywords: GCC 2.1, Gnuplot, Linux
Message-ID: <[email protected]>
Date: 15 May 92 18:01:46 GMT
References: <[email protected]>
Sender: [email protected] (The News System)
Organization: Harvey Mudd College, Claremont, CA 91711
Lines: 16

In article <[email protected]> [email protected] (Yuan-Yin
Wu) writes:
>Hi, Linuxers:
> I was compiling Gnuplot the other day and GCC told me
> gcc: Internal compiler error: program cc1 got fatal signal 6
>and CC1 said something like "IOT trap". This happened for both the
>file graphics.c and graph3d.c (the largest among the files in the
>package). I have no clue what all these mean. Could someone please
>help. Thanks in advance.

This is actually a gcc bug, not a problem with the source files. Sorry, I
don't remember precisely where the difficulty occured.

-Peter

[email protected]
uunet!jarthur!peter


[next article]
From: [email protected] (Brad c Midgley)

Subject: customising X/hardwarej?
Message-ID: <[email protected]>
Date: 15 May 92 18:10:13 GMT
References: <[email protected]>
Sender: [email protected] (netnews admin account)
Organization: Nyx, Public Access Unix at U. of Denver Math/CS dept.
Lines: 8
X-Disclaimer: Nyx is a public access Unix system run by the University
of Denver for the Denver community. The University has neither
control over nor responsibility for the opinions of users.

I have read the files included with x386 which talk about setting up
video information in Xconfig, but I found it rather difficult to fathom.
Has anyone else had the problem? Where can I find some more documentation?
(It would be very nice too if someone wrote a program which generated the
xconfig information given the clocks and synchronisation of the card and
monitor)

thanks


[next article]
From: [email protected] (Adam Goldberg)

Subject: Re: Abysmal disk performance
Message-ID:
Date: 15 May 92 18:27:31 GMT
References: <[email protected]> <[email protected]
gateway.novell.com> <[email protected]> [email protected]>
Sender: [email protected] (USENET News System)
Organization: Iowa State University, Ames IA
Lines: 13
Originator: [email protected]

In [email protected] (Ian Wells) writ
es:

>2. Is there a technical name for when the hard disk sits
>reading/writing two separate sections of the disk some distance apart
>and spends most of its time seeking? This is something annoying I've
>tended to notice with Linux, but if it now does readahead on files it
>may be improved in this respect.

I'd call that 'Thrashing'. Is that a technical term?
--
Adam Goldberg ! Ask me how IOWA STATE UNIVERSITY is
[email protected] ! censoring my usenet access!!
..!uunet!iastate.edu!goldberg ! >>I don't need a BIG BROTHER!!<<


[next article]
From: [email protected] (Kevin Cummings)

Subject: Re: Don't use linked libs for crutial stuff [was:Re: rc Shell Under Linux]
Message-ID: <[email protected]>
Date: 15 May 92 17:49:21 GMT
References: <[email protected]> <[email protected]
.claremont.edu> <1992May14.130424.16
[email protected]> <[email protected]>
Sender: [email protected] (Kevin Cummings)
Organization: Prime Computer R&D
Lines: 45

In article <[email protected]>, [email protected]
u (Michael K. Johnson) writes:
> In article <[email protected]> [email protected] (Joel M. Hoff
man) writes:
>
> I think that anything really vital should be statically linked. As I
> see it, this would include a shell, mount, rm, cp, sync (!) and
> perhaps tar (that's a killer) or dd.
>
> The idea is that if you inadvertantly delete the shared lib apparatus,
> or mangle one or more of your partitions, you should be able to restore
> enough to get shared libs working again.
>
> I think that the regular rootimage disk should use shared libs, but
> that an emergency rootimage disk that uses only staticly linked stuff
> should be available for fixing things. Jim -- could you make this as
> well? I hate to volunteer you, but you've done a good job so far...

Assumptions:
1) /bin is for programs that are needed to cold start the system,
and must be capable of running even if /usr is not yet mounted.
1a) other programs in /bin are to help recover from system
catastrophies (like /usr died) and must also be self contained.
2) shared libs are installed in /usr/lib.
3) programs in /usr/bin can have dependancies on /usr/lib.
programs in /bin can't.
4) There should be no LARGE libraries in /lib.

Conclusions:
1) programs in /bin must be statically linked and be able to
be run without /usr yet mounted.
2) programs in /usr/bin can depend on shared libs.

Shouldn't the root disks be set up in this way?

=================================================================
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] (Gregory R. Weiss)
Newsgroups: comp.sys.ibm.pc.hardware,comp.os.linux

Subject: IBM PS/2 (not clone) Disk Controller - registers needed for Linux
Summary: IBMs are not IBM compatible
Keywords: PS/2, hard disk drive, controller, IBM, register, Linux
Message-ID: <[email protected]>
Date: 26 May 92 22:15:43 GMT
Sender: [email protected] (USENET News System)
Organization: Princeton University
Lines: 15
Originator: [email protected]
Nntp-Posting-Host: phoenix.princeton.edu


After being frustrated at not being able to run Linux on the PS/2s here
at school (I don't have a computer here!), I found out that IBM's PS/2s use
a different set of registers for accessing the hard disk controller.
Linux doesn't work on any PS/2s apparently, because the hard disk driver
in Linux only supports the clones.

Does anyone know what registers IBM uses? Is anyone aware of a lookup/
translation table between the IBM and clone HD controller mechanisms?

BTW, the fact that the PS/2s can only run Linux off floppy should be in
the FAQ for comp.os.linux in the section on "What do I need to run Linux."

--Greg
[email protected]


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

Subject: Re: Problem with gcc 2.1 (why do I have to flush stdout?)
Message-ID: <[email protected]>
Keywords: gcc flush stdout
Organization: Occidental College, Los Angeles, CA 90041
References: <[email protected]>
Date: Wed, 27 May 1992 05:11:35 GMT
Lines: 35

[email protected] (Budi Rahardjo) wrote:
>I have problem with gcc 2.1 with a simple (test) program :
>
>#include
>char line[255];
>main(argc, argv)
>int argc; char *argv[];
>{
> printf("Testing gcc 2.1\n");
> printf("Enter a string: ");
> scanf("%s", line);
> printf("You entered: %s\n", line);
> printf("Enter another string: "); /* I have to flush stdout */
> scanf("%s", line);
> printf("You entered %s\n", line);
>}
>
>When I try to run it, it won't display the "Enter another string:".
>I have to flush stdout explisitly to display it. What gives ?
>Is this a gcc bug or somehere something is buffering my stdout...

Welcome to ANSI C. ANSI C says that you can't rely on printf() flushing
your output. Lots of people are getting bitten my this with the new
IBM compiler for OS/2 also.

I guess you have to add fflush() calls here and there. Seems awfully
kludgy to me, but who ever said the standard way was the best way?

-Dave

--
David Giller, Box 134 | Q: How many Oregonians does it take to screw in a light
Occidental College | bulb? A: Three. One to replace the bulb, and two to
1600 Campus Road | fend off all the Californians trying to share the
Los Angeles, CA 90041 | experience. [email protected]


[next article]
From: [email protected] (Ajay Shah)

Subject: shutdown procedure?
Date: 26 May 1992 22:53:19 -0700
Organization: University of Southern California, Los Angeles, CA
Lines: 17
Sender: [email protected] (Ajay Shah)
Message-ID:
NNTP-Posting-Host: almaak.usc.edu

My /etc/rc sets up a swap and mounts a 2nd partition into /usr/home.

Does that mean a few syncs and powerout are still all we need by
way of a shutdown procedure?

(Why is Sun's /usr/etc/shutdown program 25k long (and world unreadable??)?)

What is a decent fsck (not toothless while still not being too slow)
command to put into /etc/rc? I would be happier if the filesystem was
verified at each boot.

Thanks,

-ans.
--

Ajay Shah, (213)749-8133, [email protected]


[next article]
From: [email protected] (David Tucker / KC4ZGO)

Subject: Re: Problem with GCC 2.11a curses.h
Message-ID: <[email protected]>
Date: 27 May 92 06:27:45 GMT
Sender: [email protected]
Organization: Digital Equipment Corporation
Lines: 12


In article , [email protected] (A
nthony Rumble) writes...
>Something along the lines of /usr/include/curses.h:77: parse error before '100'

Someone was kind enough to answer this a couple months ago when I was having
problems. I just found my note about it. He said:

put: #define HZ HZ

before the: #include

David Tucker @ATO / KC4ZGO / [email protected]


[next article]
From: [email protected] (Humberto Ortiz-Zuazaga)

Subject: Re: memory hungry uncompress?
Message-ID: <[email protected]>
Summary: get a new compress!
Keywords: compress, memory, error.
Organization: Univ. of Cincinnati
References: <[email protected]> <[email protected]> <[email protected]>
Date: Wed, 27 May 92 03:48:31 GMT
Lines: 18

In article <[email protected]> [email protected] (Jeff Blanchet) writes:
>> in article <[email protected]>, [email protected] (Timothy Cullip
) says:
>> > try to uncompress a 1.3 Mbyte file
>> > Memory Fault
>> > Out of Swap memory
>
>I recently tried to uncompress the boot image for 0.96a and got the same
>error. The compressed file is only 97461 bytes. I could uncompress it fine
>on a unix system at work.

The compress on the 0.95 rootimage is BAD. Get a new one. Better yet,
get the FAQ.

P.S. This is what causes the parse error before "(" messages from gcc
also.
--
Humberto Ortiz-Zuazaga [email protected]
Dept. of Physiology & Biophysics University of Cincinnati


[next article]
From: [email protected] (Henning Spruth)

Subject: xfig works in synchronous mode only - why?
Message-ID:
Date: 27 May 92 06:48:51 GMT
Sender: [email protected]
Lines: 22

I finally managed to make xfig work. As mentioned elsewhere, you
have to start it in synchronous mode, i.e. using
'xfig -sync'. Performance is rather slow, and there seem to be
some server bugs - you get interesting results when drawing
splines with width>1, for example.

When starting xfig without the -sync flag, the server hangs
while drawing the command panels, i.e. the mouse freezes.
I managed, however, to kill the xfig process from a serial
terminal, and the server continued working as if nothing
had happened.

What is the reason for this behaviour? I'm using 8 MB of RAM
with no swap, which is ample even for big compile jobs
with X Windows running.

Henning
--
------------------------------------------------------------------------
Henning Spruth [email protected]
Institute of Electronic Design Automation
Technical University of Munich, Germany


[next article]
From: [email protected] (Ajay Shah)

Subject: When do we escape 64 Meg partitions?
Date: 27 May 1992 00:01:53 -0700
Organization: University of Southern California, Los Angeles, CA
Lines: 12
Sender: [email protected] (Ajay Shah)
Message-ID:
NNTP-Posting-Host: almaak.usc.edu

I'm sure many others feel this, but I would have a couple of hundred
meg on a system ``to do real work''. In such a system the hard limit
of 64 Meg is a pain ... I often have single files > 100 Meg.

So: is eliminating this problem on the cards?

Thanks,

-ans.
--

Ajay Shah, (213)749-8133, [email protected]


[next article]
From: [email protected] (Wen-Chun Ni)

Subject: Re: Problem with gcc 2.1 (why do I have to flush stdout?)
Message-ID: <[email protected]>
Keywords: gcc flush stdout
Sender: [email protected]
Organization: Brown University Department of Computer Science
References: <[email protected]> <[email protected]
shire.oxy.edu>
Date: Wed, 27 May 1992 06:58:33 GMT
Lines: 39

In article <[email protected]> [email protected] (
David Giller) writes:
>[email protected] (Budi Rahardjo) wrote:
>>I have problem with gcc 2.1 with a simple (test) program :
>>
>>#include
>>char line[255];
>>main(argc, argv)
>>int argc; char *argv[];
>>{
>> printf("Testing gcc 2.1\n");
>> printf("Enter a string: ");
>> scanf("%s", line);
>> printf("You entered: %s\n", line);
>> printf("Enter another string: "); /* I have to flush stdout */
>> scanf("%s", line);
>> printf("You entered %s\n", line);
>>}
>>
>>When I try to run it, it won't display the "Enter another string:".
>>I have to flush stdout explisitly to display it. What gives ?
>>Is this a gcc bug or somehere something is buffering my stdout...
>
>Welcome to ANSI C. ANSI C says that you can't rely on printf() flushing
>your output. Lots of people are getting bitten my this with the new
>IBM compiler for OS/2 also.
>
>I guess you have to add fflush() calls here and there. Seems awfully
>kludgy to me, but who ever said the standard way was the best way?
>
>-Dave
>

Weird. I've tried something similar but gotten a different result.
printf() did flush the output. One thing should be mentioned is:
I used gcc 2.12.



Wen-Chun Ni


[next article]
From: [email protected] (Ajay Shah)

Subject: Success story on Linux troubleshooting (or how to get questions answered)
Date: 27 May 1992 00:05:09 -0700
Organization: University of Southern California, Los Angeles, CA
Lines: 24
Sender: [email protected] (Ajay Shah)
Message-ID:
NNTP-Posting-Host: almaak.usc.edu

Today I encountered the dreadful error "unable to open lock-file"
when trying to mount a partition.

Here is how I solved it:

(At a USC Sun server:)

$ cd /usr/spool/news/comp/os/linux
$ less `egrep 'lock-file' * | awk -F\: '{print $1}'

Sure enough -- there were three articles on it: one asking a question and
two telling the answer (rm -f /etc/mtab~ first).

It's a irritating and avoidable bug, but it was sure easy getting the
answer!

Thanks guys,

-ans.

ps: where does one get 'df'?
--

Ajay Shah, (213)749-8133, [email protected]


[next article]
From: [email protected] (Ian McCloghrie)

Subject: vt100 scrolling windows in linux?
Message-ID:
Summary: the vt100 appears not to function properly with scrolling windows
Keywords: vt100
Sender: [email protected] (USENET News System)
Organization: University of Washington
Date: Wed, 27 May 1992 07:05:30 GMT
Lines: 22

Running kermit from 0.96a, the vt100 emulation appears to not be
functioning properly. In particular, when an application sets up
a scrolling window. Whenever this application (tinyfugue in visual
mode if anyone is familiar with it) switches windows to print
something, it prints extra blank lines above and below the line it
prints. It also doesn't wrap input in the bottom window properly,
anything off the end of one input line wraps onto the beginning of
the same line, rather than scrolling down one.

I'm running the 0.96a kernel image, modified only to use hda1 as
root device. not sure if this is useful or not, but I'm running on a
386-25 with 4megs, JAX-8094 Trident 8800 SVGA card, and a seagate
st-1100 with a 47 meg linux partition and 30 meg DOS.

So is there anything I can set to correct the terminal problems,
or do I just have to live with it? I've tried playing with the linewrap
option, it didn't do anything.

--
/~> Ian McCloghrie | Time Traveller, betz.biostr.washington.edu 4096
< < /~\ |~\ |~> | | <~ | email: [email protected] <-> (128.95.10.119 4096)
\_> \_/ |_/ |~\ |__| _> | Card Carrying Member, UCSD Secret Islandia Club


[next article]
From: [email protected] (Theodore Ts'o)

Subject: Re: breakout lists/discussion channels
Message-ID: <[email protected]>
Date: 27 May 92 07:40:44 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 24


From: [email protected] (Michael K. Johnson)
Date: 27 May 92 03:47:37 GMT

The main lists for linux are the [email protected] lists
-- They are really one list with several "channels", or at least that
is the way the guy who wrote the software (Ari Lemke,
[email protected]) thinks of it. To get information on how to use
this list, send a note to [email protected], and
a daemon will auto-respond, telling you how to use the automagical
features of the list, including joining and leaving the list.

Well, given the sheer amount of traffic, I think the argument could be
made that comp.os.linux (mailing list digest form) are more of the "main
lists" for Linux. But I suppose it's all in the matter of definitions
and semantics.

One problem I have with the [email protected] setup is that (1)
the discussions aren't archived anywhere, not that I can tell, anyway,
and (2) using the channels are a real pain in the tuckus, both in terms
of joining a channel and remembering to put in the MN-Key: field....
Gripe, gripe.... ๐Ÿ™‚

- Ted


[next article]
From: [email protected] (Theodore Ts'o)

Subject: Re: comp.sources.linux & comp.binaries.linux
Message-ID: <[email protected]>
Date: 27 May 92 07:50:03 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 18


From: [email protected] (Alan Hourihane)
Date: 26 May 92 12:35:42 GMT

What could the voting be on the above newsgroups. I personally would vote
again and again for the above groups. It would make life much easier to
keep track of fixes to linux.

Maxims about voting early and voting often aside :-), you're allowed to
vote once as far as newsgroup votes are concerned. ๐Ÿ™‚

I personally like things lumped together in one newsgroup. Among other
things, it makes handling the mailing list digest much easier. And
while the traffic is a bit on the high side, it's not clear to me how
much splitting apart of the newsgroup would help, due to things like
cross-postings.

- Ted


[next article]
From: [email protected] (Kuo-Chun Su)

Subject: What files to grab ?
Message-ID: <[email protected]>
Date: 27 May 92 07:24:15 GMT
Sender: [email protected] (USENET PROXY ACCOUNT)
Organization: NTUTaiwan
Lines: 19
Nntp-Posting-Host: ccds2

Dear All,

I am new to Linux. When I connect to tsx-11, I find a lot of files
which I don't quite know. Would anyone tell me what files are needed if
I want to install Linux, including Linux's sources and executables, X11,
GCC, and every other things required to do X programming ?

Another question. Could anyone describe the pros and cons of Linux
and Coherent 386 for me ? Also, if Linux more reliable and stable than
BSD386 ?

Any information will be appreciated !

Thanks in advance !

Sincerely Yours
Kuo-Chun Su




[next article]
From: [email protected] (Johan Myreen)

Subject: Keyboard lights (Was: 0.96 and 0.96a comments (and a bug report))
In-Reply-To: [email protected]'s message of Mon, 25 May 1992 04:33:07 GMT
Message-ID:
Sender: [email protected] (Usenet pseudouser id)
Nntp-Posting-Host: conan.cs.hut.fi
Organization: Helsinki University of Technology, Finland
References:
Date: 27 May 92 10:58:24
Lines: 22

In article [email protected] (Michael Pereckas)
writes:

>I have just started using 0.96a. I have noticed no major problems
>yet, but I have noticed a minor keyboard problem---the keyboard LEDs
>don't work.

My bug, sorry. An error slipped in when I translated keyboard.S to C.
In the file kernel/chr_drv/keyboard.c, function kb_ack() change the
line

if (inb(0x64) == 0xfa)
to
if (inb(0x60) == 0xfa)

I didn't notice this bug when testing the C version because the lights
worked on my keyboard (but with a small delay caused by the loop).

Hope this helps.

--
Johan Myreen
[email protected]


[next article]
From: [email protected] (Gregor Schmid)

Subject: longer filenames dilemma
Message-ID:
Date: 27 May 92 09:55:12 GMT
Sender: [email protected] (News Manager)
Organization: FB3, TU-Berlin, Germany.
Lines: 32
Nntp-Posting-Host: fb3-s2.math.tu-berlin.de


Hello Linuxers,

a lot of people have complained about the 14 chars file name length
restriction and common answers are: take the kernel sources and fix
it, the only incompatibility you get is that you are no longer able
to use image distributions.

That's fine with me, so I went ahead and tracked the problem down
(I think) to
#define MINIX_NAME_MAX 14
in include/linux/minix_fs.h

Changing this value to s.th. else (preferred 2^n - 2) and recompiling
should give you a kernel that handles longer filenames (right Linus ?).
The problem: You have to rebuild mkfs with the new header and remake
your file system. That's inconvenient but unavoidable.

But how do I get a bootable rootdisk with the new file system ?
I can use the new mkfs to build the fs on the disk, but the old kernel
can't handle the disk and the new one won't boot without a root file
system. Some intermediate system or a file sytem conversion program
(maybe usig DOS - urgh) seems to be required.

Any suggestions ?

If all this is redundant because we'll get a new file system soon
(VFS ?) please let me know. I don't really like to reinstall 30M
over and over again.

Regards,
Greg


[next article]
From: Andrew Haylett (GEC-Marconi Research Centre)

Subject: Re: Problem with gcc 2.1 (why do I have to flush stdout?)
Message-ID: <[email protected]>
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Date: Wed, 27 May 1992 09:08:14 GMT
Lines: 38

*** someone said:

| I have problem with gcc 2.1 with a simple (test) program :

[code omitted]

| When I try to run it, it won't display the "Enter another string:".
| I have to flush stdout explisitly to display it. What gives ?
| Is this a gcc bug or somehere something is buffering my stdout...

*** someone else replied:

| Welcome to ANSI C. ANSI C says that you can't rely on printf() flushing
| your output. Lots of people are getting bitten my this with the new
| IBM compiler for OS/2 also.
|
| I guess you have to add fflush() calls here and there. Seems awfully
| kludgy to me, but who ever said the standard way was the best way?

*** and someone else chipped in:

| Weird. I've tried something similar but gotten a different result.
| printf() did flush the output. One thing should be mentioned is:
| I used gcc 2.12.

This is an interesting problem. I experimented a couple of weeks back with
some similar code. I compiled it with gcc2.1, then ran the *same* binary under
0.95c+ and pre-0.96. Under 0.95 output was flushed before reading; under
pre-0.96 (and now 0.96a), it wasn't. So it looks like a kernel feature, rather
than purely a compiler/library bug. Other programs like the mcc fdisk and top
do the same thing. I reported it to Linus, who was surprised, but I don't know
if he tracked down a kernel problem. Sure, the new behaviour is quite valid,
but I'm curious to know why it has changed.

----
Andrew Haylett | Inet: [email protected] | Fax: +44 245 75244
GEC-Marconi Research | Tel: +44 245 73331 x.3283 | Telex: 995016 GECRES G



[next article]
From: [email protected] (Ajay Shah)

Subject: Re: longer filenames dilemma
Message-ID:
Date: 27 May 92 09:20:17 GMT
References:
Sender: [email protected] (Ajay Shah)
Organization: University of Southern California, Los Angeles, CA
Lines: 22
NNTP-Posting-Host: almaak.usc.edu

[email protected] (Gregor Schmid) writes:

>a lot of people have complained about the 14 chars file name length
>restriction and common answers are: take the kernel sources and fix
>it, the only incompatibility you get is that you are no longer able
>to use image distributions.

>That's fine with me, so I went ahead and tracked the problem down
>(I think) to
>#define MINIX_NAME_MAX 14
>in include/linux/minix_fs.h

It's amazing that a restriction like 14 characters has been used
when its ``so trivial'' to avoid ๐Ÿ™‚

I vaguely know about work on VFS. Will it eliminate short filenames?
Or should we go about assembling bootimage etc. with 254 character filenames
at tsx?

--

Ajay Shah, (213)749-8133, [email protected]


[next article]
From: [email protected] (H.J. Lu)

Subject: Re: Problem with GCC 2.11a curses.h
Message-ID: <[email protected]>
Date: 27 May 92 08:23:16 GMT
References: <[email protected]>
Sender: [email protected] (USENET News System)
Organization: Washington State University
Lines: 19

In article <[email protected]> [email protected] (David Tucker /
KC4ZGO) writes:
>
>In article , [email protected] (
Anthony Rumble) writes...
>>Something along the lines of /usr/include/curses.h:77: parse error before '100
'
>
>Someone was kind enough to answer this a couple months ago when I was having
>problems. I just found my note about it. He said:
>
>put: #define HZ HZ
>
>before the: #include
>
>David Tucker @ATO / KC4ZGO / [email protected]

In the next release, HZ is undefined in curses.h. I think this one is
better. I will change to this.


H.J.


[next article]
From: [email protected] (H.J. Lu)

Subject: Re: Problem with gcc 2.1 (why do I have to flush stdout?)
Message-ID: <[email protected]>
Keywords: gcc flush stdout
Sender: [email protected] (USENET News System)
Organization: Washington State University
References: <[email protected]> <[email protected]
shire.oxy.edu>
Distribution: world
Date: Wed, 27 May 92 08:20:15 GMT
Lines: 35

In article <[email protected]> [email protected] (
David Giller) writes:
>[email protected] (Budi Rahardjo) wrote:
>>I have problem with gcc 2.1 with a simple (test) program :
>>
>>#include
>>char line[255];
>>main(argc, argv)
>>int argc; char *argv[];
>>{
>> printf("Testing gcc 2.1\n");
>> printf("Enter a string: ");
>> scanf("%s", line);
>> printf("You entered: %s\n", line);
>> printf("Enter another string: "); /* I have to flush stdout */
>> scanf("%s", line);
>> printf("You entered %s\n", line);
>>}
>>
>>When I try to run it, it won't display the "Enter another string:".
>>I have to flush stdout explisitly to display it. What gives ?
>>Is this a gcc bug or somehere something is buffering my stdout...
>
>Welcome to ANSI C. ANSI C says that you can't rely on printf() flushing
>your output. Lots of people are getting bitten my this with the new
>IBM compiler for OS/2 also.
>
>I guess you have to add fflush() calls here and there. Seems awfully
>kludgy to me, but who ever said the standard way was the best way?
>

>-Dave
>

Please wait for the next release of gcc for this test.

H.J.


[next article]
From: [email protected] (FIGUEIVEDO, CARLOS)

Subject: X11-binaries
Sender: [email protected] (The Usenet-News System)
Message-ID: <[email protected]>
Date: Wed, 27 May 1992 11:06:55 GMT
X-News-Reader: VMS NEWS 1.20
Nntp-Posting-Host: br1.hrz.th-darmstadt.de
Organization: Technical University Darmstadt, Germany
Lines: 2

Where can I find the X11-binaries ?
Thanks


[next article]
From: [email protected] (Mark Evans)
Newsgroups: comp.sys.ibm.pc.hardware,comp.os.linux

Subject: Re: IBM PS/2 (not clone) Disk Controller - registers needed for Linux
Message-ID: <[email protected]>
Date: 27 May 92 10:45:37 GMT
References: <[email protected]>
Sender: [email protected] (Usenet administrator)
Organization: Aston University
Lines: 26
Nntp-Posting-Host: uhura

[email protected] (Gregory R. Weiss) writes:
:
: After being frustrated at not being able to run Linux on the PS/2s here
: at school (I don't have a computer here!), I found out that IBM's PS/2s use
: a different set of registers for accessing the hard disk controller.
: Linux doesn't work on any PS/2s apparently, because the hard disk driver
: in Linux only supports the clones.
:
: Does anyone know what registers IBM uses? Is anyone aware of a lookup/
: translation table between the IBM and clone HD controller mechanisms?
:
: BTW, the fact that the PS/2s can only run Linux off floppy should be in
: the FAQ for comp.os.linux in the section on "What do I need to run Linux."
:
Runs absolutly fine from floppy, both 3.5" and 5.25", also drivers the printer
ok, just completly ignores the hard disk.
Anyone got any ideas, if it is just a matter of different registers, then
adding a new device driver should work.
However getting the kernal sources, a c compiler, and the root file system onto
2 floppy disks is rather tricky.

--
-------------------------------------------------------------------------
Mark Evans |[email protected]
+(44) 21 565 1979 (Home) |[email protected]
+(44) 21 359 6531 x4039 (Office) |


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

Subject: Re: misc questions(gcc2.11, filenames, this group)
Message-ID:
Date: 27 May 92 10:58:57 GMT
References: <[email protected]>
<[email protected]>
Sender: [email protected] (Kayvan Sylvan)
Organization: Sylvan Associates
Lines: 23
In-Reply-To: [email protected]'s message of 26 May 92 19:42:29 GMT

In article [email protected] (Al Clark) writes:

(Yes, I know about kill files, but they use
up so much time and space resources that I avoid them.)

But using the mailing list topics as the initial set of breakouts is
probably a good start.

I see. You don't want to use the technology available (kill files and
miscellaneous other methods) because it's personally inconvenient. Yet
you don't hesitate to break up the newsgroup, which would be an
inconvenience to a lot more people.

I vote we keep the group the same. I read it all and I don't want to
be on a gazillion mailing lists each devoted to one corner of the
Linux world.

---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] (Linus Benedict Torvalds)

Subject: Re: What files to grab ?
Message-ID: <[email protected]>
Date: 27 May 92 11:30:53 GMT
References: <[email protected]>
Organization: University of Helsinki
Lines: 47

In article <[email protected]> [email protected] (K
uo-Chun Su) writes:
>
> I am new to Linux. When I connect to tsx-11, I find a lot of files
> which I don't quite know. Would anyone tell me what files are needed if
> I want to install Linux, including Linux's sources and executables, X11,
> GCC, and every other things required to do X programming ?

X and gcc are in a constant flux right now: before getting the binaries
it's a good idea to join the X11 and gcc mailing-lists. The newest
version of gcc is a pre-release of gcc-2.2, and the libraries change
almost daily if you really want to keep up. It's a bit of a bother, but
the result of it all is that most of the library bugs seem to have been
fixed.

Other than the compiler and X11, the easiest way to install linux right
now is to get the mcc-interim version, and then upgrading to the newest
kernel image. The mcc-interim version had some problems with the login
binary, but I think they got fixed. The mcc distribution contains most
of the basic unix utilities.

> Another question. Could anyone describe the pros and cons of Linux
> and Coherent 386 for me ?

Well, coh386 isn't available for a couple of weeks yet, but this is how
I've understood all the hype:

Linux pros:

- virtual memory, page sharing. VERY BIG pro.
- free, full source.
- X11 and a lot of other good software already ported.
- good support from hackers

Coh386 pros:

- manual. Easier installation. BIG pro.
- runs standard text-mode 386-unix and coh286 binaries.
- support from MWC

> Also, if Linux more reliable and stable than
> BSD386 ?

Haven't tested 386BSD, but I assume it will be pretty good in a month or
two, and it's probably quite useable already. If you want NFS and full
networking, it's the way to go. Comments, anyone?

Linus


[next article]
From: [email protected] (Kari E. Hurtta)

Subject: Re: Problem with gcc 2.1 (why do I have to flush stdout?)
Message-ID:
Date: 27 May 92 12:36:49 GMT
References: <[email protected]>
Sender: [email protected] (Uutis Ankka)
Organization: Department of Computer Science, University of Helsinki, Finland
Lines: 14
In-Reply-To: [email protected]'s message of 27 May 92 09: 08:14 GMT

In article <[email protected]> [email protected] (GEC-Marcon
i Research Centre) wrote:
> This is an interesting problem. I experimented a couple of weeks back with
> some similar code. I compiled it with gcc2.1, then ran the *same* binary under
> 0.95c+ and pre-0.96. Under 0.95 output was flushed before reading; under
> pre-0.96 (and now 0.96a), it wasn't. So it looks like a kernel feature, rather
> than purely a compiler/library bug. Other programs like the mcc fdisk and top
> do the same thing. I reported it to Linus, who was surprised, but I don't know
> if he tracked down a kernel problem. Sure, the new behaviour is quite valid,
> but I'm curious to know why it has changed.

You ran the *same* binary under 0.95c+ and pre-0.96
with or without shared library?

- K E H "Well - I don't know anyting about linux"


[next article]
From: [email protected] (Joerg Pommnitz)

Subject: Re: PROLOG available ???
Keywords: Linux,PROLOG
Message-ID:
Date: 27 May 92 11:06:04 GMT
References:
Sender: [email protected] (Owner of all binaries)
Organization: tu-chemnitz
Lines: 23

[email protected] (Joerg Pommnitz) writes:

>Well, I think the title says it all. I am looking for a working PROLOG system
>for Linux.
>I think it is mentioned in the FAQ, but I haven't found it.
>Could somebody post a hint ? Thanks in advance.

Today I've got some answers, so I'll post a short summary.

There is a PROLOG system called swi-prolog available on tsx-11 in the
directory /pub/linux/sources/usr.bin.
It was ported to LINUX by Pieter Olivier ([email protected]).

Another way could be to use a PROLOG written in LISP.
I think this is only a half solution, but people who are
mainly interested in LISP and only want to try out PROLOG should
check this way. It was suggested by Jim Burnes ([email protected]).
More informations about this solution may be found in the comp.lang.lisp FAQ.

Thanks to everybody for the answers. I hope my summary is useful for
some people who had the same problem.

Joerg


[next article]
From: [email protected] (Joerg Pommnitz)

Subject: Re: Bug in Linux 0.96a with patch 1?
Keywords: Linux, Bug
Message-ID:
Date: 27 May 92 11:21:49 GMT
References: <[email protected]>
Sender: [email protected] (Owner of all binaries)
Organization: tu-chemnitz
Lines: 14

[email protected] (Jaime Jofre) writes:

>I have applied patch 1 to the Linux 0.96a distribution, but now I'm experiencin
gkeyboard lockup once in a while. I haven't noticed any pattern to the problem,
>but I have noticed that I can switch VCs, but I still can't get any input from
>the keyboard. Has anyone else experieced this problem?

>Jaime Jofre

Well I have the same problem, but I think I found the missing pattern.
Do you have a terminal connected to ttys[12] ?
If so try to remove the relating getty. This solved (?) the problem
for me, but now I can't use my VT100.

Joerg


[next article]
From: [email protected] (Joerg Pommnitz)

Subject: Re: 0.96 and 0.96a comments (and a bug report)
Keywords: kernel bug 0.96 0.96a
Message-ID:
Date: 27 May 92 10:55:46 GMT
References:
Sender: [email protected] (Owner of all binaries)
Organization: tu-chemnitz
Lines: 15

I'll add some comments on 0.96a (with patches).
First at all it solved my reboot problem. Now I can restart my
system without doing a hard reset.
Because I have never had any problem with my harddisk, I can't
comment about this.
But I think there is a new problem.
I have an old VT100 compatible terminal connected to ttys1.
If running a getty on ttys1, the console (VGA display) dies !
I can switch between the different virtual consoles but I can't
do any input (not even login).
I don't know if this problem is related to the changes in console.c
keyboard.[Sc] or anything else.
If somebody can fix this problem, I would be interested in the solution.

Joerg


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

Subject: Re: Sources of utils....
Message-ID: <[email protected]>
Date: 27 May 92 13:11:49 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 17

in article <[email protected]>, [email protected] (Dan Miner) say
s:

> I have found sources for many for programs for linux, even
> linux itself. But there is one thing I have not found...
> The sources for tar, ls, chmod, etc... The binaries that come with
> the rootimages for new releases. Where can I find these little
> utils that make life livable.... ? ๐Ÿ™‚

Try anon ftp to ftp.mcc.ac.uk:dsk/pub/linux/mcc-interim/0.95c+. The sources
directory (I think - mcc is unavailable right now) are mostly links to
pub/gnu, and the patches directory contains our patches to same for 0.95c+,
+ a misc.tar.Z w/login, etc.
--
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] (Jim Burns)

Subject: Re: Problem with gcc 2.1 (why do I have to flush stdout?)
Message-ID: <[email protected]>
Date: 27 May 92 13:17:01 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 16

in article <[email protected]>, [email protected] (GEC-Marco
ni Research Centre) says:

> *** someone else replied:

> | Welcome to ANSI C. ANSI C says that you can't rely on printf() flushing
> | your output. Lots of people are getting bitten my this with the new
> | IBM compiler for OS/2 also.

On most systems, tty i/o is line-buffered, so you need to have a \n in your
format, BUT output also gets flushed by the next input, as in his scanf.
This is a tty driver/line discipline thing, not necessarily ANSI.
--
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] (Jim Burns)

Subject: Re: shutdown procedure?
Message-ID: <[email protected]>
Date: 27 May 92 13:22:14 GMT
References:
Organization: Georgia Institute of Technology
Lines: 20

in article , [email protected] (Ajay Shah) sa
ys:

> What is a decent fsck (not toothless while still not being too slow)
> command to put into /etc/rc? I would be happier if the filesystem was
> verified at each boot.

I just say:

echo fscking ... '(no write)' # use fsck -a for writes
fsck /dev/whatever

in the rc, and if I get errors, I manually do the fsck -a. (I want to know
what I'm getting into before I do it!) Obviously, you can't fsck the root
before mounting it. What is the best place to fsck /usr (or wherever) -
before the mount or after?
--
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] (Jim Burns)

Subject: Re: What files to grab ?
Message-ID: <[email protected]>
Date: 27 May 92 13:44:13 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 33

in article <[email protected]>, [email protected] (
Kuo-Chun Su) says:

> I am new to Linux. When I connect to tsx-11, I find a lot of files
> which I don't quite know. Would anyone tell me what files are needed if
> I want to install Linux, including Linux's sources and executables, X11,
> GCC, and every other things required to do X programming ?

See my recent posts about gnu util sources & patches for mcc-interm 0.95c+
(today), and re-compiling the kernel when you need to update the ps -U
database (yesterday?), which explains about getting the kernel sources
(mainly, see tsx-11.mit.edu:pub/linux/docs/*, especially gcc_kernel.notes).
I don't use X, but that's mostly on banjo.concert.net:pub/Linux/X11.
The mcc-interim stuff on ftp.mcc.ac.uk:dsk/pub/linux/mcc-interim/0.95c+,
and read & follow the readmes, are the best way to get binaries. Note,
their boot disk uses the 0.95c+ kernel, which doesn't support SCSI. If you
need it, use their xdisk, boot of of the standard bootimage-0.96a, and
follow the mcc instructions.

> Another question. Could anyone describe the pros and cons of Linux
> and Coherent 386 for me ? Also, if Linux more reliable and stable than
> BSD386 ?

I don't use coherent, but it's system 7 programming interface w/64k code +
64k data, like minix, but no sources, unlike minix & linux. Linux is
definitely more stable than 386BSD at his point (can't even boot it!) Note
there is a commercial (read: $1000) offering called BSDI's BSD/386, which
is beta, and much more stable, comes from the same source tree as 386BSD,
but also has X, SCSI, and other goodies.
--
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] (Jim Burns)

Subject: Re: What files to grab ?
Message-ID: <[email protected]>
Date: 27 May 92 13:54:15 GMT
References: <[email protected]>
Organization: Georgia Institute of Technology
Lines: 22

in article <[email protected]>, [email protected]
FI (Linus Benedict Torvalds) says:

> kernel image. The mcc-interim version had some problems with the login
> binary, but I think they got fixed. The mcc distribution contains most

Not on the boot & util images. You still have to get the login source from
misc.tar.Z (ftp.mcc.ac.uk:dsk/pub/linux/mcc-interim/0.95c+ in the sources
or patches dir) and recompile it.

> Coh386 pros:

> - manual. Easier installation. BIG pro.
> - runs standard text-mode 386-unix and coh286 binaries.
> - support from MWC

Oh well, my previous post got the difference between 386BSD and BSD/386
right, but not between coherent & coherent 386!
--
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] (JAMES LEWIS NANCE)

Subject: Need help with internal modem
Message-ID: <[email protected]>
Originator: [email protected]
Lines: 17
Sender: [email protected] (USENET News System)
Reply-To: [email protected] (JAMES LEWIS NANCE)
Organization: North Carolina State University, Project EOS
Date: Wed, 27 May 1992 13:48:36 GMT


Hello all,
I have just installed linux.96 on a PC wiht an internal modem, and I have
no idea how to talk to the modem. Can anyone help me?

Thanks,
Jim Nance
[email protected]

PS if it helps:

25 MHz 386
8 Megs w/ 8 Megs swap
/dev/hda 105Meg Dos
/dev/hdb1 30Meg Linux
/dev/hdb2 8Meg swap
AMI BIOS


[next article]
From: [email protected] (Remy CARD)

Subject: Re: longer filenames dilemma
Message-ID: <[email protected]>
Sender: [email protected] (Mr Background)
Reply-To: [email protected] (Remy CARD)
Organization: The Internet
Date: Wed, 27 May 1992 14:01:09 GMT
Lines: 39


[email protected] (Gregor Schmid) writes:

|a lot of people have complained about the 14 chars file name length
|restriction and common answers are: take the kernel sources and fix
|it, the only incompatibility you get is that you are no longer able
|to use image distributions.

|That's fine with me, so I went ahead and tracked the problem down
|(I think) to
|#define MINIX_NAME_MAX 14
|in include/linux/minix_fs.h

You can use the VFS layer to add a new filesystem type with
less restrictions than the Minix one. I have made it and, on my
machine, Linux uses two filesystems :
- a minix filesystem for the root,
- an "extended-minix" filesystem for the users filesystem.

My "extended-minix" filesystem is a quick and dirty hack from
the Minix one but I am currently working on a "real filesystem". This
filesystem will allow several new things :
- filesystems up to 4096 Gb (no more 64Kb limit),
- file names up to 255 characters,
- file size up to 16 Gb (files are limited to 256 Mb with the
Minix file system),
- a few other modifications to get a faster file system.

Unfortunately, I don't know when I will be able to post this
new file system (I have my real work to do and I can work on Linux only
during my free time). I hope that I will have finished it soon but ...

If there are people working on a similar project, I would be
happy to hear from them.

-----
Remy CARD
[email protected]
Tel : +33 (1) 44.27.34.23


[next article]
From: Andrew Haylett (GEC-Marconi Research Centre)

Subject: Re: Problem with gcc 2.1 (why do I have to flush stdout?)
Message-ID: <[email protected]>
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Date: Wed, 27 May 1992 14:18:21 GMT
Lines: 19

I wrote:

| This is an interesting problem. I experimented a couple of weeks back with
| some similar code. I compiled it with gcc2.1, then ran the *same* binary under
| 0.95c+ and pre-0.96. Under 0.95 output was flushed before reading; under
| pre-0.96 (and now 0.96a), it wasn't....

and [email protected] (Kari E. Hurtta) said:

| You ran the *same* binary under 0.95c+ and pre-0.96
| with or without shared library?

Using the (same) shared library on each try, just rebooting the kernel
each time. No other differences.

----
Andrew Haylett | Inet: [email protected] | Fax: +44 245 75244
GEC-Marconi Research | Tel: +44 245 73331 x.3283 | Telex: 995016 GECRES G



[next article]
From: [email protected] (Robert Lockwood)

Subject: Need Help Printing
Message-ID: <[email protected]>
Date: 27 May 92 14:21:27 GMT
Sender: [email protected]
Organization: University of Virginia Computer Science Department
Lines: 8

I am getting error 19 when I try `cat foo > /dev/lp0`. I am running the
0.96 boot image and I created /dev/lp0 with the command
`mknod /dev/lp0 c 6 0`. Is there something else I need to do??

Thanks,

Bob
[email protected]


[next article]
From: [email protected] (Dylan Smith)

Subject: Re: DDJ? (was: Something else)
Message-ID: <[email protected]>
Date: 27 May 92 13:00:34 GMT
References: <[email protected]> <[email protected]>
Sender: [email protected] (Usenet poster for nntp (tj - 17/12/91))
Organization: Transputer Centre, Bristol Polytechnic.
Lines: 21
Nntp-Posting-Host: hal-9000

In article <[email protected]> [email protected] (Tony Tra
vis) writes:

>Personal Computer World has shown a great deal of interest in the
>technically competent/hobbyist readership that exists in the UK. I
>have read it since 1978 and I think it would be an ideal place to
>publish articles about Linux (there is a regular Unix section).

I did mail the person who writes the UNIX section in the days of 0.12 (his
mail address is [email protected]). Unfortunately, he didn't seem at all
interested at the time and wrote me a very hostile reply (he dosen't seem
to like free software for some reason, and tends only to write about
commercial stuff). However, in the current edition of Personal Computer
World, he did give Linux a reasonable mention, so I suppose I can let
him off ๐Ÿ˜‰




--
Email : JANET [email protected] | Everywhere else [email protected]
[email protected] | [email protected]


[next article]
From: [email protected] (H.H. Bergman)

Subject: QIC-02 tape driver
Message-ID: <[email protected]>
Keywords: Mailing list/kernel mods
Organization: Faculty of Arts & Letters, Groningen University, The Netherlands
Date: Wed, 27 May 1992 14:08:13 GMT
Lines: 16

It's taken me longer than I expected, but here it goes...

For those of you with a QIC-02 tape drive:

I have created a mailing list 'TAPE' on linux-activists.
You can join by sending mail to [email protected]
Put 'X-mn-admin: join TAPE' in the header field, or as the first line in
your message.

I currently have a driver that sort of works. Even multiple files on one
tape. After I clean it up a bit, I'll try to upload it to
banjo.concert.net before June 3.

-Hennus Bergman
[email protected] [until July 3, 1992]
[email protected] [keep trying 'till it doesn't bounce :-(]


[next article]
From: [email protected] (Tim Bunnell)

Subject: X cannot find any screens...
Message-ID: <[email protected]>
Sender: [email protected] (USENET News Service)
Nntp-Posting-Host: oeo.asel.udel.edu
Organization: AI duPont Inst.
Date: Wed, 27 May 1992 15:33:42 GMT
Lines: 16

Can anyone suggest reasons why the X server cannot find any screens on
my system? It exits with a message to that effect (after grabbing and
clearing the screen!). I've read the info sheets and tried most of the
seemingly obvious things. I'm running 0.96a on a 486 system with 8 meg
of memory.

Thanks in advance,

Tim Bunnell


--
-------------------------------------------------------------------------------
H. Timothy Bunnell
A. I. duPont Institute, Speech Processing Laboratory
Univ. of Delaware, Computer and Information Sciences


[next article]
From: [email protected] (George Grimes)

Subject: Will linux run DOS applications??
Message-ID: <[email protected]>
Date: Wed, 27 May 92 15:01:52 GMT
Organization: Netcom - Online Communication Services (408 241-9760 guest)
Sender: George A. Grimes
Lines: 7

I would love to switch my PC over to linux but I MUST continue to run some
DOS software (some of them Windows versions!). Can linux handle this?
Or would I have to boot back and forth between linux and dos?

Thanks,
George



[next article]
From: [email protected] (Theodore Ts'o)

Subject: Re: What files to grab ?
Message-ID: <[email protected]>
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Date: Wed, 27 May 1992 15:35:18 GMT
Lines: 33

From: [email protected] (Linus Benedict Torvalds)
Date: 27 May 92 11:30:53 GMT

Haven't tested 386BSD, but I assume it will be pretty good in a month or
two, and it's probably quite useable already. If you want NFS and full
networking, it's the way to go. Comments, anyone?

Well 386BSD does have NFS and networking, but it has the following
problems (some of which may be solved with time):

* It wants to take over the entire disk (it doesn't use the
MS-DOS partition scheme). Also, in general, it seems to
require more disk space.

* It doesn't have shared libraries.

* It doesn't have X ported to it (although that will probably
eventually change).

* The kernel is a lot more complicated and has a lot more
hair in it.

I would like to see Linux eventually have NFS and full networking. It
is true that by adding networking, it will make the Linux kernel more
complicated, but I think we can keep the added complexity to a minimum.
Even in the worse case, I believe that the end result would be still be
simpler and better than what BSD is now.

I am starting to do some work about moving the networking code into the
kernel, although I don't have a lot of time to put into the project. If
someone would like to collaborate with me on this, that would be great.

- Ted


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

Subject: Re: Need Help Printing
Message-ID: <[email protected]>
Date: 27 May 92 15:35:19 GMT
References: <[email protected]>
Sender: [email protected] (Usenet administrator)
Organization: Aston University
Lines: 11
Nntp-Posting-Host: uhura

[email protected] (Robert Lockwood) writes:
: I am getting error 19 when I try `cat foo > /dev/lp0`. I am running the
: 0.96 boot image and I created /dev/lp0 with the command
: `mknod /dev/lp0 c 6 0`. Is there something else I need to do??
Was the printer recognised in the boot up sequence?
You should get a message to that effect from the boot up disk.
--
-------------------------------------------------------------------------
Mark Evans |[email protected]
+(44) 21 565 1979 (Home) |[email protected]
+(44) 21 359 6531 x4039 (Office) |


[next article]
From: [email protected] (Theodore Ts'o)

Subject: Re: longer filenames dilemma
Message-ID: <[email protected]>
Date: 27 May 92 15:56:34 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 66


From: [email protected] (Ajay Shah)
Date: 27 May 1992 02:20:17 -0700

[email protected] (Gregor Schmid) writes:

>a lot of people have complained about the 14 chars file name length
>restriction and common answers are: take the kernel sources and fix
>it, the only incompatibility you get is that you are no longer able
>to use image distributions.

>That's fine with me, so I went ahead and tracked the problem down
>(I think) to
>#define MINIX_NAME_MAX 14
>in include/linux/minix_fs.h

It's amazing that a restriction like 14 characters has been used
when its ``so trivial'' to avoid ๐Ÿ™‚

I vaguely know about work on VFS. Will it eliminate short filenames?
Or should we go about assembling bootimage etc. with 254 character filenames
at tsx?

The problem with the minix directory structure is that it uses fixed
length slots for your filenames. So if you have 254 character
filenames, and you have a directory with 500 files in it, that directory
will take up over 128k of space. It also means that searching that
directory will slower, since in order to find files at the end of the
directory, the kernel would have to read in 128k worth of disk blocks to
search the entire directory. So just #defining MINIX_NAME_MAX to a
large number is not an optimal fix to the "short characters" problem.
#defining MINIX_NAME_MAX also doesn't fix the other problem with the
minix filesystem, which is that it limits you to 64 meg filesystems.

The right way to do this would be to define a directory structure that
has variable length pathnames; this does make allocation of directory
slots more complicated though, since they would no longer be a fixed
size. I would invite people to study the Berkeley Fast Filesystem for
an example of one way of handling the allocation problem.

To solve the 64 meg limit, you would need to revamp the inode structure
to allow for 32 bit blocks. It would also be nice to put in an
allocation policy which attempts to spread files across the disk, and
which clusters blocks belong to one particular file close together.

What is VFS? Linux 0.96 already has VFS; VFS stands for Virtual
Filesystem, which means that it acts as a filesystem switch. It makes
it easy for someone to design another filesystem format and include it
into the Linux kernel along with the standard Minix filesystem format.
So it _enables_ someone to design a robust filesystem which would have
all of the nice properties described above, and it could be included in
the kernel in such a way that both Minix filesystem and the new,
improved filesystem could be mounted at the same time. This solves the
incompatibility problem, since the rootimage disk could still using the
Minix filesystem, while your hardisk could be using the new, improved
filesystem, and the Linux kernel could simultaneously access both of
them.

What's the catch? As far as I know, no one is working on this new
filesystem. If someone is interested in doing so, the VFS layer should
make it much easier to implement and test it. Another filesystem that
would be nice to have is a MS-DOS filesystem, so that people would be
able to access a DOS filesystem transparently from Linux, without
needing to use the mcopy, mdir commands.

- Ted


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

Subject: Re: longer filenames dilemma
Message-ID: <[email protected]>
Date: 27 May 92 16:46:49 GMT
References: <[email protected]>
Sender: [email protected]
Reply-To: [email protected] (Robert J. McNamara)
Organization: Lawrence Berkeley Laboratory
Lines: 16
Nntp-Posting-Host: cad4.lbl.gov

In article <[email protected]> [email protected] writes:
>What's the catch? As far as I know, no one is working on this new
>filesystem. If someone is interested in doing so, the VFS layer should
>make it much easier to implement and test it. Another filesystem that
>would be nice to have is a MS-DOS filesystem, so that people would be
>able to access a DOS filesystem transparently from Linux, without
>needing to use the mcopy, mdir commands.

And get stuck with the 8.3 filenames? Ugh....Howabout the OS/2 HPFS filesystem?
(or is that protected?)

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


[next article]
From: [email protected] (Ian Wells)
Newsgroups: comp.sys.ibm.pc.hardware,comp.os.linux

Subject: Re: IBM PS/2 (not clone) Disk Controller - registers needed for Linux
Message-ID:
Date: 27 May 92 16:59:56 GMT
References: <[email protected]>
<[email protected]>
Sender: [email protected] (The news facility)
Organization: CompScis Anonymous, Cambridge branch
Lines: 12
In-reply-to: [email protected]'s message of Wed, 27 May 1992 10:45:37 GM
T

One of the more recent upgrades for MINIX has a PS/2 ESDI driver in
it. It might be worth looking at that for help (although I wouldn't
recommend nicking any code, MINIX people seem to be a bit touchy
about that).
Also, the keyboard driver didn't work on the PS/2s I
tried. It completely ignores the backspace key, which is very
frustrating. I seem to remember that PS/2 keyboard controllers are
odd too, and that there's something in the MINIX patch to deal with
that. In any case, I've seen the same problem with DOS code. I can't
imagine that this will be too difficult to fix, though.

Ian.


[next article]
From: [email protected] (Darren Senn)

Subject: Re: Context diffs
Message-ID: <[email protected]>
Date: 27 May 92 17:05:14 GMT
References: <[email protected]>
Sender: [email protected]
Reply-To: [email protected]
Organization: Curiosity Confederacy
Lines: 21

In article <[email protected]>, [email protected] (Jim Burns) writes
:
> Thanx for the tutorial, but I've also seen 'patch -p1 ...'. What creates
> that?

Nothing "creates" 'patch -p1 ...'. That's the command (and one possible
argument) that you can use to _apply_ the patch that you learned how
to create in the patch tutorial. There is a program called "patch"
whose sole purpose is to apply patches. The -p argument says "strip
leading components off of all filenames in the patch." So if
you use -p1, and your diff says:
*** usr/src/linux/kernel/fuu.c Mon Feb 24 15:49:21 1992
--- usr/src/linux/kernel/bar.c Tue May 26 14:22:49 1992
< lines of diffs >
Then patch will apply the patch to the file src/linux/kernel/bar.c
(dropping the "usr/" from the front of the file name). This path
is (of course) relative to your current directory.

--
Darren Senn Phone: (408) 479-1521
[email protected] Snail: 1785 Halterman #1
Wasurenaide -- doko e itte mo soko ni anata wa iru yo. Santa Cruz, Ca 95062


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

Subject: Console refresh ?
Message-ID:
Date: Wed, 27 May 92 08:09:00 GMT
Organization: MausNet
Lines: 15
X-Gateway: MausGate/News 1.06

Hy folks,

Now I'v a running MausTausch (my Point-Programm) under Linux,
but it was very funny, because I was hunting a bug, what was'nt a bug ๐Ÿ™‚

Normaly ZModem gives a verbose on received and sended bytes and CPS
on stderr, but under Linux it doesnt ๐Ÿ™ I was hunting why.

Now I see that Linux only refreshes the console when a NL occures,
but the verbose is using only CR, so nothing happen on screen ๐Ÿ™

Can somebody tell me where to patch this ๐Ÿ™‚ Because I like the verbose
of ZModem ๐Ÿ™‚

By Michael.


[next article]
From: [email protected] (Robert Lockwood)

Subject: IPC
Message-ID: <[email protected]>
Sender: [email protected]
Organization: University of Virginia Computer Science Department
Date: Wed, 27 May 1992 16:03:52 GMT
Lines: 4

Are there any plans to add shared memory and semaphores to Linux?

Thanks,
Bob


[next article]
From: [email protected] (Richard Harry)

Subject: top, linux 0.96 etc..
Message-ID: <[email protected]>
Date: 27 May 92 12:05:16 GMT
Organization: University of Sussex
Lines: 42

Hi there linuxers, I've been lurking in this newsgroup and a.o.l since
0.12, only now do I have the urge to post a request.

top very gradually uses up available memory, so that I periodically have
to kill it and restart it. I think I remember reading the reason for
this and a possible fix, but I can't find it in any of my archives. So
does anyone have the answer ?

When building the kernel, if I do a 'make all', it fails with a general
protection error - segmentation violation. However if I 'make
tools/system' followed by a `make Image' it works ??

This only happens when making 0.96,0.96a or 0.96a+patches regardless
of the image I'm running at the time ( I'm using gcc 2.11c & gnu make )
Is this a bug in gcc or a bug in make ? or have I missed something?

( this has been mentioned before ) I get read error messages when
reading the last few blocks at the end of a floppy, but the files seem
to have been read ok. I don't know if the same happens on hard disks as
thankfully I haven't filled mine yet ๐Ÿ™‚ ( running 0.96a + Linus' first
patch )

oh, and running benchmarks shows hard disk performance up by 3x for
writes i.e 200Kbs to 600Kbs so thanks to Branko Lankester
(Is it possible to change the interleave on IDE drives ? )

Finally, any news on a driver for Trident SVGA cards for X11.
I may have to buy a new video card, can anyone send me a list of ET4000
cards that are known to work with X and linux ?

email or post please
Thanks in advance.

Richard

PS
Thanks to all the altruistic folks who made this possible. Linux
prompted me into buying a PC clone, I would have bought an Archimedes
instead!!

--
[email protected]


[next article]
From: [email protected] (Darren Senn)

Subject: Re: longer filenames dilemma
Message-ID: <[email protected]>
Date: 27 May 92 17:29:33 GMT
References:
Sender: [email protected]
Reply-To: [email protected]
Organization: Curiosity Confederacy
Lines: 19

In article , [email protected]
.math.TU-Berlin.DE (Gregor Schmid) writes:
> Changing this value to s.th. else (preferred 2^n - 2) and recompiling
> should give you a kernel that handles longer filenames (right Linus ?).
> The problem: You have to rebuild mkfs with the new header and remake
> your file system. That's inconvenient but unavoidable.

Here's the problem bothering me: why does the filesystem have to be rebuilt?
This implies that the full MAX_NAME bytes of space are allocated on the
filesystem for file names. This is very bad. BSD, for instance, uses
a linked-list of filename/inode pairs to represent directories on disk.
This is read into memory and converted to an array of char * for efficiencies'
sake.

[ Reference: Implementation of the BSD 4.3 Operating System, McGraw Hill ]

--
Darren Senn Phone: (408) 479-1521
[email protected] Snail: 1785 Halterman #1
Wasurenaide -- doko e itte mo soko ni anata wa iru yo. Santa Cruz, Ca 95062


[next article]
From: [email protected] (r b willmersdorf)

Subject: Mailing list channels
Message-ID: <[email protected]>
Date: 27 May 92 18:22:51 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 11


Hello,

Could someone please, please, e-mail to me (or post) a summary of the
channels that are active now in the linux mailing list ? Information
on how to join them would also be very welcome.

Thanks a lot,
--

Ramiro || [email protected]


[next article]
From: [email protected] (r b willmersdorf)

Subject: Latest X version ?
Message-ID: <[email protected]>
Date: 27 May 92 18:36:39 GMT
Sender: [email protected] (Mr Background)
Reply-To: [email protected]
Organization: The Internet
Lines: 21


Hi,

In a previous post, Linus mentioned the new X version, that we
should be using now because it's compiled with the shared libs
and therefore much smaller.

However, he failed to mention where to find it. I tried
banjo, tsx-11, and a mirror of nic, and couldn't find it, at least
in the obvious places.

I'd really appreciate any pointers.

PS: If everybody that's using X is supposed to join the mailing
list channel, you might as well post the whole thing here.

Bye, and many thanks

--

Ramiro || [email protected]


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

Subject: X11: olvwm dies with SHIFT-L-DRAG on title bar
Message-ID: <[email protected]>
Organization: Occidental College, Los Angeles, CA 90041
References:
Date: Wed, 27 May 1992 18:02:01 GMT
Lines: 22

Hi. I am having a problem with olvwm. Specifically, when I hold down
shift and do a left button drag on the title bar, olvwm dies.

I have tried capturing stout and stderr, but it dies silently.

I thought at first that it was a problem with the modified Logitech
Bus mouse driver I'm working on (one that maintains a count rather than
a queue of mouse events), but I went back and tried it with the new
queueing bus mouse driver, and it does the same thing there.

On the one occasion where I was left with an xterm focused, I was able to
start up the olvwm again, only for it to die on the next shift-l-drag.

Anyone else having this problem? I can't tell whether it is a new problem or
whether I just never noticed before.

-Dave
--
David Giller, Box 134 | Q: How many Oregonians does it take to screw in a light
Occidental College | bulb? A: Three. One to replace the bulb, and two to
1600 Campus Road | fend off all the Californians trying to share the
Los Angeles, CA 90041 | experience. [email protected]


[next article]
From: [email protected] (Al Clark)

Subject: Re: Sources of utils....
Message-ID:
Date: Wed, 27 May 92 18:22:00 GMT
Organization: Self
References: <[email protected]>
Lines: 27

In article <[email protected]> [email protected] (Dan Miner) writ
es:
>
> I have found sources for many for programs for linux, even
>linux itself. But there is one thing I have not found...
>The sources for tar, ls, chmod, etc... The binaries that come with
>the rootimages for new releases. Where can I find these little
>utils that make life livable.... ? ๐Ÿ™‚
>
>Dan
>
>[email protected]

The sources for ls and chmod are in the GNU fileutil package. I have
identified three GNU packages: filetutil, textutil, and shellutil.
These may have linux patches. I have obtained and compiled textutils-1.3,
but not the others. There is a diff for shellutils, but I haven't obtained
the shellutils source yet. Tar is a source directory of it's own.
In addition there are some linux unique sources like fsck. (Which I understand
has been lost.)

Also, there are libraries, like the gcc libc source.

Maybe we can cooperate in getting a complete set of sources that compile
under linux.
--
Al - [email protected] - My opinions are my own.
*** Practice random kindness and senseless acts of beauty! ***


[next article]
From: [email protected] (Mark Evans)
Newsgroups: comp.sys.ibm.pc.hardware,comp.os.linux

Subject: Re: IBM PS/2 (not clone) Disk Controller - registers needed for Linux
Message-ID: <[email protected]>
Date: 27 May 92 18:33:31 GMT
References:
Sender: [email protected] (Usenet administrator)
Organization: Aston University
Lines: 16
Nntp-Posting-Host: uhura

[email protected] (Ian Wells) writes:
: One of the more recent upgrades for MINIX has a PS/2 ESDI driver in
: it. It might be worth looking at that for help (although I wouldn't
: recommend nicking any code, MINIX people seem to be a bit touchy
: about that).
: Also, the keyboard driver didn't work on the PS/2s I
: tried. It completely ignores the backspace key, which is very
: frustrating.
The backspace problem is apparently a problem with the latest version of linux
switching to another VC usually cures it.

--
-------------------------------------------------------------------------
Mark Evans |[email protected]
+(44) 21 565 1979 (Home) |[email protected]
+(44) 21 359 6531 x4039 (Office) |


[next article]
From: [email protected] (Pete Chown)

Subject: Re: problems with ps, pcomm95c+1 and more...
Summary: Way of implementing ps functionality without reading kernel memory
Keywords: ps /dev/mem /dev/kmem
Message-ID: <[email protected]>
Date: 27 May 92 18:30:03 GMT
References: <[email protected]> <19
[email protected]> <[email protected]>
Sender: [email protected] (The news facility)
Reply-To: [email protected] (Pete Chown)
Organization: U of Cambridge Comp Lab, UK
Lines: 13

I have a ps working, based on a system call which allows any process
to read the process structure (you can even find out what a process
umask is set to!). Also a variant on the call allows a table to be
read which holds, in text form, the argv[] passed to the most recent
execve pertaining to a particular process. If people generally think
this is the right way to go, I will be happy to pass the kernel
patches and ps source on.


--
__________________________________________________ I look forward to the time
Pete Chown, mail [email protected] (Internet) | when machines do what we
or [email protected] (Janet ๐Ÿ™‚ _| want, not what we tell them




 December 26, 2017  Add comments

Leave a Reply