NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/comp/os/linux/9882

Text File | 1992-09-07 | 10.3 KB | 277 lines

Newsgroups: comp.os.linux
Path: sparky!uunet!gatech!darwin.sura.net!jvnc.net!primerd.prime.com!tiger1.prime.com!cummings
From: [e-mail] (Kevin J. Cummings)
Subject: Re: linux install experience, newbie question, and not so newbie question
Message-ID: <[e-mail]>
Keywords: FAQ, install, gcc
Sender: [e-mail] (Usenet)
Organization: Prime Computer, Inc.
References:  <[e-mail]>
Date: Fri, 4 Sep 1992 20:49:43 GMT
Lines: 264

In article <[e-mail]>, [e-mail] () writes:
>
> Sorry the first post didn't get out, saved to the wrong place.
>
> [long post]
>
> I am a newbie installing linux. Below are my experience, questions, etc.
>
> The experience may help another newbie; Also if you got an answer to my
> question please let me know.
>
> Hardware:
>
> 386-40 + hd +4mb:
>
> 40m dos
> 20m extended dos
> 24m not used
> 40m linux
>
> VGA+Trident --> Question: do I have no hope for X?

If you read this group diligently, you;d know that a Trident 8900 driver
is in ALPHA test.  If you have a Trident 8800 board, I beleive you are SOL.

> Software:
>
> .97 boot
> .97p1 root

Grrr   That's 97.1 root.

> Disk 3 to 10 of SLS package. tar'd to 3 inch floppies.
>
> GCC 2.2.2d stuff.
>
> Question: can some one tell me in brief why I want .97 rather than .96,
> and in what level things are incompatible?

GCC 2.2.2d won't run under 0.97 or before.  You need to run it under 0.97-pl1
or better kernel.  For the other changes, look for the RELEASE NOTES for
the 0.97 releases (I'm not promising they are available, but they should be).

> This installation is used primarily for someone to learn C programming
> on unix systems.
>
> Install experience and problems:
>
> 1. Dos partition the disk. (fdisk). one primary 40m, one extended 20m
> which has D: drive on it. (I have 130m total)
>
> 2. Boot 0.97 then 0.97p1 root floppy
> 3. Install on hard disk, /dev/hda4, 40m. no swap no user
>
> Question here: if I make the /dev/hda4 active, then DOS won't boot on C:
> saying that illegal partition table entry or some such. Apparently only
> one active partition is allowed in DOS. This is not covered in the FAQ.

Your "active" partition is the one in which the system reads the boot sector
from (after reading in the partition table of course).  It is primarily for
MS-DOS or OS/2 style systems.  LINUX tends to use other (more than one) method
of booting off the hard disk.  Of course only one partition can be active at a
time.  The boot software assume that you want to boot the active partition.
If you overwrite the boot sector of your hard drive with the LINUX boot image
(which doesn't fit in the space allowed for the boot sector anyways) you would
most likely llose you ability to boot DOS.  I use the BOOTLIN/BOOT142 combination
of boot managers.  Others use LILO or SHOELACE.  Some patch the boot floppy to
use the hard drive as root partition but boot the kernel from the boot floppy.
To each their own.

> 4. go back to DOS, fix the bood disk, boot to hard drive.
>
> 5. tar SLS package
>
> 6. Question: found that all my /dev/*fd* entries are changed!!.
> /dev/fd{0 1} use to have major minor of 2 0 or some such, now they are 0
> 0. Am I doing things wrong or is this the change between two versions?

devices don't change major numbers like that between versions.  Something else must
be wrong.

> 6. go back to 0.97p1 root floppy and do install again, this correct the
> /dev/fd* stuff. (just get the root dir files)
>
> 8. Now I have a 0.97 boot imag, with some 0.97p1 binaries + files, plus
> whatever the rest come from SLS which is .96
>
> Question -- am I in trouble? I really liked to have newer stuff, but I
> should probably have used all 0.96 stuff, given that the system is used
> for learning C programming on UNIX.

Your 0.97.1 root disk will work just fine with the 0.97 boot image.

You should instead be wondering about getting the 0.97-pl1 or better kernel so
GCC will work after you have installed it.  Actually, GCC will run under 0.96
(patchlevel something or other or better), but the code it produces will only
work right under 0.97-pl1 or better kernels.  I guess it's best to be clear
about this if you want to test out the code GCC generated!

> Question 2 -- I only found that ps didn't work, where can I find a
> better ps binary?

Somewhere on your favorite archive site!  Each archive site is set up slightly
differently, though many just mirrors others.

> Question 3 -- how can I make vi not insert blanks? get the .96 termcap
> file?

Can't help here, I'm and EMACS man myself.

> Extended Install
>
> 9. get 0.97 source and install
> 10. patch <.97p1.patch or some. Got whole bunch of "hunt" failures, some
> message complaining about that "this looks like new style diff to me" or
> some such. Remember that I probably have .96's patch command, am I ok at
> this point?

Probably better to do as  patch -c < patchfile.   The -c suppresses the "looks
like a new style diff" message by forcing patch to assume these are context
diffs.  0.96's patch should work just fine in this regard.

> 11. patch <.97p2.patch, same as above, but it even ask me why I am
> reversing the patch done above! I just asked it to go ahead.

Huh?

> 12. Question -- what have I got? 0.97 p2 source? really messed up 0.97
> source?

I don't know.  What files did you get, and WHERE/WHEN did you get them???

> 13. I forgot if I installed the .97p1inc that came from gcc2.2.2d or
> not, anyway I have it.

Install it when you install GCC.

> 14. Tested that gcc from SLS works

Well, you've got a starting point.

> Problems with SLS
>
> 15. as stated by other posters:  /tmp has wrong permissions, so is
> /dev/*fd* which is not writeable by other users, causing mtools to fail.

What ARE the permissions on /dev/*fd*????  I beleive the powers that be
decided that user root and group disk should be the only ones with write
access to the disk devices.  You get around this by making the mtools
programs sgid to group disk.  I beleive a similar thing was done to the tty
devices makeing them owned by user root and group tty.  Communications
programs (like kermit and ka9q) which need to open tty devices should be
sgid tty.

> emacs is missing a link from /usr/local/emacs to /usr/emacs ( path may
> be wrong I don't remember).
>
> Install GCC2.2.2d
>
> Question -- what is the difference, overall, between the gcc that's in
> SLS (2.2.2) and the 2.2.2d version? yes I can read the release note but
> I simply don't understand what is a jump table, let alone some change to
> it.

I thought there were more changes than just the jump table support.

> 16. Did as readme said, the gcc set of files are from a dos disk, and the
> name is really messed up, I have to restore them one by one as best as I
> could
>
> 17. install.2.x
>
> went through remarkably without major problems, as much as I can tell.
>
> 18. tested gcc works, this is the 2.2.2d gcc. Also, as this point I
> found out that my disk has much more files than before the gcc install.

Yeup.

> 19. emacs complain that libx11.blah isn't found

You have a newer version of emacs.  There were a few suggested fixes for this.
One was to get the library and install it (you'll do this if you ever install
X from the SLS distribution).  Another was to invoke emacs with the -nw option,
a third was to unset your DISPLAY environment variable.  I don't know if the
third method will work or not.

> 20. ln -s /lib/lib.so.2blah /lib/libX11.blah
>
> 21 emacs runs fine (remember I have to do a link from /usr/local/emacs
> to somewhere)
>
>
> Now I am in trouble, people with more experience please help me,
>
> 22. after a while, gcc would never work, gcc -v would produce an error:
>
> can't load /lib/lib.so.2blah,
> Version mismatch (or something like this)
>
> Is this somehow related to the stupid step 20? It worked for a while
> (gcc), then won't anymore

Shouldn't be.  Step 20 only created a symbolic link from the X11 library to your
lib.so library.  That should have had no effect on GCC being able to run.  You
might very well be suffering from the bleeding memory bug that Linus produced
a patch for.  (All the more reason to be sure your running 0.97-pl1.)

> 23. reinstall of gcc from SLS would not help, gave me the same error

What about a cold start of the machine.  Halt it using shutdown now.
reboot it.  Does GCC work again.  If not, you've done something to your disk.
(improper machine halt without syncing first?)

> 24. blow away everything, repeat the install again, but forget GCC2.2.2d
>
> 25. this time I am smart, I copied some /lib/libc.2.2.2 to
> /lib/libX11.blah, rather than doing a link.
>
> 26. I never dared to install GCC2.2.2d again
>
>
> Anyway it took me a while, but linux is a great system overall.
>
> But I think that in the FAQ, we should have an entry to help out the
> newbies (non testers who probably want to get a stable system to use,
> rather then installing things all day):

I guess I can't say this enough.  Even though many great people try and produce
simple to install version of LINUX, LINUX is primarily an OS for the UNIX
experienced, by the Unix experienced, so that they can enjoy themselves.

> 0.96 can be had by a SLS package, and is easier to install

That was SLS's intent.

> 0.97px is better than 0.96 because xxxx, also xxxx is not compatible
> between the two version.
>
> you can upgrade from 0.96 to 0.97, by swapping the files xxxx/and
> following the procedure yyyy...
>
> GCC 2.2.2d is different from SLS/gcc2.2.2(?) because ...., to install at
> your own risk because then your debugger needs to be changed too(?)
>
>
> This is a long post, if you read through here, thanks, and if possible,
> let me know if you have answer to any questions.

I hope it's been educational.  I know I still have a lot to learn about this
stuff myself.

=================================================================
Kevin J. Cummings                       PrimeService
20 Briarwood Road                       A Computervision Company
Framingham, Mass.                       500 Old Connecticut Path
                    Framingham, Mass.
Work:  [e-mail]
Home:  [e-mail]

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