Subject: v 0.11 boot disk problem
Date: Sun, 8 Dec 91 15:01:30 -0800
From: mpeters@polyslo.csc.calpoly.edu (Marc Peters)
To: Linux-activists@joker.cs.hut.fi

I just downloaded the new improved v0.11, but it doesn't want to load on my
computer.  It gives a couple errors similar to
 HD controller reset failed: 000
 Kernel Panic: HD controller not ready
and aborts the loading of the system.  My computer has an MFM HD and 
controller which may account for the problem, but v0.10 didn't seem to mind.
A friend has a machine with an IDE drive and controller, it gives the 
first error but loads the system and seems to work fine.

If you could direct me toward a cure I'd be happy again, even if the cure
involves replacing my HD and controller.  But I'm wondering what differences
could be in the new kernel that would obsolete my HD and if those changes
can't be re-changed.

Thank you for your time,
Marc Peters

Subject: RE: v 0.11 boot disk problem
Date: Mon, 9 Dec 1991 01:18:21 +0200
From: Ari Lemmke <arl@zen.cs.hut.fi>
To: Linux-activists@joker.cs.hut.fi
In-Reply-To: Marc Peters's message of Sun, 8 Dec 91 15:01:30 -0800 
< 9112082301.AA00933@polyslo.csc.calpoly.edu>


	Linus downloaded new stuff ..  so:

	Linux 0.11 should now be available at nic.funet.fi ...
	everything needed is in INSTALL directory (actually
	stuff in INSTALL directory is hard linked from other
	directories).

	The main README file changes constantly ...

	arl

Subject: hd_reset fails in 0.11
Date: Mon, 9 Dec 1991 02:57:40 +0200
From: Linus Benedict Torvalds < torvalds@cc.helsinki.fi>
To: Linux-activists@joker.cs.hut.fi

Marc Peters: "v 0.11 boot disk problem" (Dec  8, 15:01):
> I just downloaded the new improved v0.11, but it doesn't want to load on my
> computer.  It gives a couple errors similar to
>  HD controller reset failed: 000
>  Kernel Panic: HD controller not ready
> and aborts the loading of the system.  My computer has an MFM HD and 
> controller which may account for the problem, but v0.10 didn't seem to mind.
> A friend has a machine with an IDE drive and controller, it gives the 
> first error but loads the system and seems to work fine.

Thanks, this is why there are still debug-statements in the code. There
are in fact two solutions (I hope) to the problem. Both of them require
you to change the file "linux/kernel/blk_drv/hd.c" which does the
harddisk specific stuff. As 0.10 doesn't try to reset the controller at
bootup, it seems to work, and you can still use it to recompile 0.11.

1 - set static int reset to 0 at startup (currently 1).  This means
linux won't try to reset the controller in the beginning, but this is
just a temporary hack: if there are errors later on that need a reset,
you'll be in the same trouble.

2 - try to edit the timeouts in "reset_controller". If memory serves,
it's a line that goes like this:

	outb_p(4,HD_CMD);
>>	for (i=0 ; i<100 ; i++)
		nop();
	outb_p(xxx,HD_CMD);

Change the max value to 10000 or something, and try again. I have only
one controller to test, and timing-problems are always the worst kind.
There are a couple of other timing-values that are completely arbitrary
(guess a number between 1 and 100000, and if it works, it works :-), but
this is the first suspect.

I'd be very grateful if everybody who encounters this problem would try
out (2), and report to me how big the value had to be before the reset
made it (or indeed if it works even then). Don't do (1) unless nothing
else works. I'd also be interested to know what kind of machine you are
running (fast machine, slow disk?).

		Linus

PS. I hope even old linux-users read the INSTALL-file: there is a change
in how linux boots up. No more /bin/update, instead linux executes the
shell script /etc/rc, which may mount/fsck etc, and start up update in
the background.

Subject: hd_reset fails in 0.11
Date: Mon, 9 Dec 1991 03:51:38 +0200
From: Ari Lemmke < arl@zen.cs.hut.fi>
To: Linux-activists@joker.cs.hut.fi
In-Reply-To: Linus Benedict Torvalds's message of Mon, 
9 Dec 1991 02:57:40 +0200 <199112090057.AA06512@kruuna.helsinki.fi>


>2 - try to edit the timeouts in "reset_controller". If memory serves,
>it's a line that goes like this:

>		outb_p(4,HD_CMD);
>	>>	for (i=0 ; i<100 ; i++)
>			nop();
>		outb_p(xxx,HD_CMD);

>Change the max value to 10000 or something, and try again. I have only
>one controller to test, and timing-problems are always the worst kind.
>There are a couple of other timing-values that are completely arbitrary
>(guess a number between 1 and 100000, and if it works, it works :-), but
>this is the first suspect.

	I hacked one SCSI driver for Mach pc532 ... and there
	was a lot of static wait loops (aaaaaargh) .. what we did was
	changed wait loops as:

		Wait_uSec( 100 );

	and Wait_uSec function has loop .. where is (should be)
	computed one uSec loop.

	But we have all only 25MHz 32532 chips .. so it was
	hard coded (ugh). There are many PC 386s around ..
	like mine .. I'm running my 386/25 with 16 MHz CPU,
	and don't have guts to run it full speed (gets really
	hot, and don't have peltier element or any other way
	to cool my machine just now ;-), so I'm running it
	about 6 MHz speed (my board has dip for it).

	arl

Subject: linux-.11 cures unexpected interupt problem
Date: Mon, 9 Dec 91 19:56:31 PST
From: pmacdona@sol.UVic.CA (Peter MacDonald)
To: linux-activists@joker.cs.hut.fi

Just a word of praise for linux-0.11.  I still get unexpected HD interrupt
messages, but I could compile the system, and comment out that message.
Which means linux can live with my brain-dead IDE drive, something minix
couldn't do without my modifying the harddisk driver.

Subject: Seg Faults under 0.11
From: "LCDR Michael E. Dobson" <rdc30@nmrdc1.nmrdc.nnmc.navy.mil>
To: linux-activists@joker.cs.hut.fi (Linux Mailing List)
Date: Tue, 10 Dec 91 10:44:17 EST

I seem to be having some trouble with Out of memory errors under 0.11 that
I didn't have under 0.10.  Uncompress is failing with an:

Out of memory.
Segmentation Fault

error on even small files.

At bootup I show 3647 free blocks on the root hd device and 3145728 free
bytes of mem.

The cu from Will Rose's UUCP 1.2 package also fails with the above error
even though I now have it compiling and linking with only some incompatible
pointer type warnings, some unused variable warnings, and some implicit
function declaration warnings.

Can anyone shed some light as to what could be causing the above errors?
-- 
Mike Dobson, Sys Admin for      | Internet: rdc30@nmrdc1.nmrdc.nnmc.navy.mil
nmrdc1.nmrdc.nnmc.navy.mil      | UUCP:   ...uunet!mimsy!nmrdc1!rdc30
AT&T 3B2/600G Sys V R 3.2.2     | BITNET:   dobson@usuhsb or nrd0mxd@vmnmdsc
WIN/TCP for 3B2                 | MCI-Mail: 377-2719 or 0003772719@mcimail.com

Subject: More Seg Fault troubles
From: "LCDR Michael E. Dobson" < rdc30@nmrdc1.nmrdc.nnmc.navy.mil>
To: torvalds@kruuna.helsinki.fi (Linus Torvalds)
Date: Tue, 10 Dec 91 16:05:11 EST
Cc: linux-activists@joker.cs.hut.fi (Linux Mailing List)

Linus,

I've found another set of conditions where I can reproducibly generate the
out of memory
Segmentation fault

messages using a simple command.

Environment:

drwxr-xr-x   root   sys    /usr/spool
drwxr-xr-x   uucp   uucp   /usr/spool/uucp

ls -l /usr/spool  

from anywhere but within /usr/spool generates :

/usr/spool
total 1
out of memory
Segmentation fault

a plain ls /usr/spool generates:
uucp

with no errors

looks possibly like a EPERM problem during the stat of the directory.

-- 
Mike Dobson, Sys Admin for      | Internet: rdc30@nmrdc1.nmrdc.nnmc.navy.mil
nmrdc1.nmrdc.nnmc.navy.mil      | UUCP:   ...uunet!mimsy!nmrdc1!rdc30
AT&T 3B2/600G Sys V R 3.2.2     | BITNET:   dobson@usuhsb or nrd0mxd@vmnmdsc
WIN/TCP for 3B2                 | MCI-Mail: 377-2719 or 0003772719@mcimail.com

Subject: Re: More Seg Fault troubles
From: "LCDR Michael E. Dobson" < rdc30@nmrdc1.nmrdc.nnmc.navy.mil>
To: torvalds@cc.helsinki.fi (Linus Benedict Torvalds)
Date: Wed, 11 Dec 91 12:02:13 EST
Cc: linux-activists@joker.cs.hut.fi (Linux Mailing List)
In-Reply-To: < 199112102246.AA07956@kruuna.helsinki.fi>; 
from "Linus Benedict Torvalds" at Dec 11, 91 12:46 am

Linus,

I've done some more investigating and can give you detailed info that
should reproduce the bug on your system.

Entry from /etc/passwd that causes problem

uucp::5:5:/usr/lib/uucp   **NOTE NO ":"  on end of entry***

ls -l /usr/spool

drwxr-xr-x    uucp    uucp    48 11 DEc 10:53 uucp

/usr/spool
total 1
out of memory
Segmentation fault

ls -lR /usr/spool is ok!

cd /usr/spool

ls -l is ok!

Now for the interesting part:

adding ":" to end of uucp entry in /etc/passwd cures problem!!

So, you know the code involved, bug or "feature"?


-- 
Mike Dobson, Sys Admin for      | Internet: rdc30@nmrdc1.nmrdc.nnmc.navy.mil
nmrdc1.nmrdc.nnmc.navy.mil      | UUCP:   ...uunet!mimsy!nmrdc1!rdc30
AT&T 3B2/600G Sys V R 3.2.2     | BITNET:   dobson@usuhsb or nrd0mxd@vmnmdsc
WIN/TCP for 3B2                 | MCI-Mail: 377-2719 or 0003772719@mcimail.com