initial shell respawning too fast

initial shell respawning too fast

[Oliver Amft]

I am trying to boot Linux (linuxppc_2_4_devel 2.4.19-pre1) from a
Walnut,
405GP Rev D board - All works fine till INIT starts (see screen
printout below). The root fs (nfs boot) is an older version of the
HardHat filesystem. bash crashes somehow, but unforunatly it prints no
report. An old HHL 2.4.0-test2 kernel still works with this root fs.

INIT: version 2.77 booting
INIT: Entering runlevel: 3
INIT: Id "sh" respawning too fast: disabled for 5 minutes
INIT: no more processes left in this runlevel
INIT: Id "sh" respawning too fast: disabled for 5 minutes
INIT: Id "sh" respawning too fast: disabled for 5 minutes
INIT: Id "sh" respawning too fast: disabled for 5 minutes
INIT: Id "sh" respawning too fast: disabled for 5 minutes


Probably I am missing some trivial thing - any ideas welcome!

Oliver

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: initial shell respawning too fast

[Frank Haverkamp]

Hi,

I have seen that behaviour when I used a kernel without math
emulation on a filesystem which had binaries not compiled with
-mcpu=403 having floating point instructions in them.

The missbehavior was caused because sh/bash crashed due an invalid
instruction and init was trying to respawn it of course without
success.

#! /bin/bash
#
# Here's a useful little one-liner that lets you find out if there are any
# floating point instructions in an object file.
#

OBJDUMP=/opt/fsp/i586-pc-linux-gnu/local/bin/powerpc-linux-objdump

function usage()
{
    echo "Here's a useful little one-liner that lets you find out if there"
    echo "are any floating point instructions in an object file."
    echo
    echo "Usage: $(basename $0) ppc-objfile"
}

if [ -z $1 ]; then
        usage
        exit
else
        $OBJDUMP -S $1 | egrep \
        'mtfsf|lfd|lfs|stfd|stflwx|stfs|:[ \t]*(fc|fd|fe|ff|ec|ed|ee|ef) ';
fi

exit 0

I found this script which finds out if there are any floating point
instructions in you binaries/libraries. I copied it somewhere and
hope that the one you originally wrote it is ok with seeing this
published ;-)

Frank

Oliver Amft <oam@oamx.net> schrieb am 20.03.02:
> I am trying to boot Linux (linuxppc_2_4_devel 2.4.19-pre1) from a
> Walnut,
> 405GP Rev D board - All works fine till INIT starts (see screen
> printout below). The root fs (nfs boot) is an older version of the
> HardHat filesystem. bash crashes somehow, but unforunatly it prints no
> report. An old HHL 2.4.0-test2 kernel still works with this root fs.
>
> INIT: version 2.77 booting
> INIT: Entering runlevel: 3
> INIT: Id "sh" respawning too fast: disabled for 5 minutes
> INIT: no more processes left in this runlevel
> INIT: Id "sh" respawning too fast: disabled for 5 minutes
> INIT: Id "sh" respawning too fast: disabled for 5 minutes
> INIT: Id "sh" respawning too fast: disabled for 5 minutes
> INIT: Id "sh" respawning too fast: disabled for 5 minutes
>
>
> Probably I am missing some trivial thing - any ideas welcome!
>
> Oliver
>
>

--
Frank Haverkamp
f.haverkamp@web.de
______________________________________________________________________________
Die Nummer, die sich jeder merkt: Ihre 01212 Wunschrufnummer von WEB.DE!
Jetzt sichern: http://freemail.web.de/?mc=990001


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: initial shell respawning too fast

[Oliver Amft]

Hi Frank,

thanks for your answer - actually I have found this last night! Besides,
I encountered that INIT version 2.77 generates another problem: shell
does not receive any character typed on the console, while telnet works
fine. I am not sure what it is, but updating to INIT 2.84 helps.
Thanx for the script :-)
Oliver



> I have seen that behaviour when I used a kernel without math
> emulation on a filesystem which had binaries not compiled with
> -mcpu=403 having floating point instructions in them.
>
> The missbehavior was caused because sh/bash crashed due an invalid
> instruction and init was trying to respawn it of course without
> success.
>
> #! /bin/bash
> #
> # Here's a useful little one-liner that lets you find out if there are any
> # floating point instructions in an object file.
> #
>
> OBJDUMP=/opt/fsp/i586-pc-linux-gnu/local/bin/powerpc-linux-objdump
>
> function usage()
> {
>     echo "Here's a useful little one-liner that lets you find out if there"
>     echo "are any floating point instructions in an object file."
>     echo
>     echo "Usage: $(basename $0) ppc-objfile"
> }
>
> if [ -z $1 ]; then
>         usage
>         exit
> else
>         $OBJDUMP -S $1 | egrep \
>         'mtfsf|lfd|lfs|stfd|stflwx|stfs|:[ \t]*(fc|fd|fe|ff|ec|ed|ee|ef) ';
> fi
>
> exit 0
>
> I found this script which finds out if there are any floating point
> instructions in you binaries/libraries. I copied it somewhere and
> hope that the one you originally wrote it is ok with seeing this
> published ;-)
>
> Frank
>
> Oliver Amft <oam@oamx.net> schrieb am 20.03.02:
> > I am trying to boot Linux (linuxppc_2_4_devel 2.4.19-pre1) from a
> > Walnut,
> > 405GP Rev D board - All works fine till INIT starts (see screen
> > printout below). The root fs (nfs boot) is an older version of the
> > HardHat filesystem. bash crashes somehow, but unforunatly it prints no
> > report. An old HHL 2.4.0-test2 kernel still works with this root fs.
> >
> > INIT: version 2.77 booting
> > INIT: Entering runlevel: 3
> > INIT: Id "sh" respawning too fast: disabled for 5 minutes
> > INIT: no more processes left in this runlevel
> > INIT: Id "sh" respawning too fast: disabled for 5 minutes
> > INIT: Id "sh" respawning too fast: disabled for 5 minutes
> > INIT: Id "sh" respawning too fast: disabled for 5 minutes
> > INIT: Id "sh" respawning too fast: disabled for 5 minutes
> >
> >
> > Probably I am missing some trivial thing - any ideas welcome!
> >
> > Oliver
> >
> >
>
> --
> Frank Haverkamp
> f.haverkamp@web.de
> ______________________________________________________________________________
> Die Nummer, die sich jeder merkt: Ihre 01212 Wunschrufnummer von WEB.DE!
> Jetzt sichern: http://freemail.web.de/?mc=990001
>

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/