Re: Linux-2.4.15-pre5 - probably something wrong with /proc/cpuinfo.

Re: Linux-2.4.15-pre5 - probably something wrong with /proc/cpuinfo.

[Alexander Viro]

On Sun, 18 Nov 2001, hari wrote:

> Hello,
> 
> The following shell script (extracted from SuSE 7.1 /etc/rc.d/boot) would 
> execute fine on Linux-2.4.14, Linux-2.4.15-pre3 etc.. but not on 
> Linux-2.4.15-pre5 (AMD Athlon Computer). Couldn't check it on 
> Linux-2.4.15-pre4 as it would not compile successfully.

I can't reproduce it here.  Which version of bash it is?

Re: Linux-2.4.15-pre5 - probably something wrong with /proc/cpuinfo.

[Florian Schmitt]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Same problem here... I just removed this part of the boot script - it works 
for me now. But anyway, this is a kernel issue: a stable kernel shouldn´t 
break userspace apps.

Flo

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE79pL7H7Gei80C0lQRAp8xAKCTu1cUX1/yUovibbp2t9de7tqENwCgw/1x
pY09AxkOwbgB6NHFnmQLk3k=
=0wT2
-----END PGP SIGNATURE-----

Re: Linux-2.4.15-pre5 - probably something wrong with /proc/cpuinfo.

[Alexander Viro]

On Sun, 18 Nov 2001, hari wrote:

> On a related note redirecting the contents of /proc/cpuinfo using 'cat' 
> command to a file on my home directory and providing that as an input fixes 
> the problem. For eg:

Eww...  If anything, that's cat </proc/cpuinfo | while ...,
but that's quite ugly.  Try the patch I've posted on l-k
(Subject: [PATCH][CFT] seq_file and lseek).

Linux-2.4.15-pre5 - probably something wrong with /proc/cpuinfo.

[hari]

Hello,

The following shell script (extracted from SuSE 7.1 /etc/rc.d/boot) would 
execute fine on Linux-2.4.14, Linux-2.4.15-pre3 etc.. but not on 
Linux-2.4.15-pre5 (AMD Athlon Computer). Couldn't check it on 
Linux-2.4.15-pre4 as it would not compile successfully.

#!/bin/bash
# set and adjust the CMOS clock
if test "$HWCLOCK_ACCESS" != "no" ; then
echo -n Setting up the CMOS clock
CLOCKCMD=hwclock
while read line; do
        case "$line" in
                *MTX\ Plus*) CLOCKCMD="hwclock --mtxplus --directisa" ;;
                *PReP\ Dual\ MTX*) CLOCKCMD="hwclock --mtxplus --directisa" ;;
        esac
done < /proc/cpuinfo
fi

May someone be kind enough to let me know where the problem could be?
-- 
Thank you,
harisri@bigpond.com

Re: Linux-2.4.15-pre5 - probably something wrong with /proc/cpuinfo.

[hari]

On Sat, 17 Nov 2001 15:56, Alexander Viro wrote:
> I can't reproduce it here.  Which version of bash it is?
Hello,

It is: GNU bash, version 2.04.0(1)-release

On a related note redirecting the contents of /proc/cpuinfo using 'cat' 
command to a file on my home directory and providing that as an input fixes 
the problem. For eg:

#/bin/bash
# set and adjust the CMOS clock
if test "$HWCLOCK_ACCESS" != "no" ; then
echo -n Setting up the CMOS clock
CLOCKCMD=hwclock
cat /proc/cpuinfo > /root/2.4.15-pre5
while read line; do
        case "$line" in
                *MTX\ Plus*) CLOCKCMD="hwclock --mtxplus --directisa" ;;
                *PReP\ Dual\ MTX*) CLOCKCMD="hwclock --mtxplus --directisa" ;;
        esac
done < /root/2.4.15-pre5
fi

Thanks.
-- 
Hari.
harisri@bigpond.com

Re: Linux-2.4.15-pre5 - probably something wrong with /proc/cpuinfo.

[hari]

On Sun, 18 Nov 2001 00:09, Alexander Viro wrote:
> Eww...  If anything, that's cat </proc/cpuinfo | while ...,
> but that's quite ugly.  Try the patch I've posted on l-k
> (Subject: [PATCH][CFT] seq_file and lseek).
Hello,

Thanks a lot. I have seen your patch included in Linux-2.4.15-pre6, and it 
does fix the issue.
-- 
Hari.
harisri@bigpond.com