386 'ls' gets SIGILL iff /proc is mounted

386 'ls' gets SIGILL iff /proc is mounted

[Eric Buddington]

2.4.2-ac23 nfsroot on a 386SX/20 with 6Mb RAM

On boot to single user, 'ls' and 'ls -l' work fine.

After mounting /proc, 'ls' still works, but 'ls -l' fails
with SIGILL after reading /etc/timezone (so says strace).

Unmounting /proc fixes the problem. Unmounting /dev doesn't.

I also, just now, had a spate of 'permission denied' errors
while trying to ls /dev/ subdirectories, and unexpected stale NFS handles.

The problems are varied enough that I suspect bad hardware, but would
flaky RAM cause such similar failures repeatedly? And is there a way
to test RAM explicitly?

Any tips appreciated, either to me (ebuddington@wesleyan.edu) or to
the list.

-Eric

Re: 386 'ls' gets SIGILL iff /proc is mounted

[Brian Gerst]

Eric Buddington wrote:
> 
> 2.4.2-ac23 nfsroot on a 386SX/20 with 6Mb RAM
>
> On boot to single user, 'ls' and 'ls -l' work fine.
> 
> After mounting /proc, 'ls' still works, but 'ls -l' fails
> with SIGILL after reading /etc/timezone (so says strace).
> 
> Unmounting /proc fixes the problem. Unmounting /dev doesn't.
> 
> I also, just now, had a spate of 'permission denied' errors
> while trying to ls /dev/ subdirectories, and unexpected stale NFS handles.
> 
> The problems are varied enough that I suspect bad hardware, but would
> flaky RAM cause such similar failures repeatedly? And is there a way
> to test RAM explicitly?
> 
> Any tips appreciated, either to me (ebuddington@wesleyan.edu) or to
> the list.

Silly question, but is math emulation enabled?

--

				Brian Gerst

Re: 386 'ls' gets SIGILL iff /proc is mounted

[Eric Buddington]

On Tue, Mar 27, 2001 at 09:09:24AM -0500, Brian Gerst wrote:
> > 
> > The problems are varied enough that I suspect bad hardware, but would
> > flaky RAM cause such similar failures repeatedly? And is there a way
> > to test RAM explicitly?
> > 
> > Any tips appreciated, either to me (ebuddington@wesleyan.edu) or to
> > the list.
> 
> Silly question, but is math emulation enabled?

Yes. I should have mentioned that in the original, since it occurred
to me also.

-Eric

Re: 386 'ls' gets SIGILL iff /proc is mounted

[Brian Gerst]

Eric Buddington wrote:
> 
> On Tue, Mar 27, 2001 at 09:09:24AM -0500, Brian Gerst wrote:
> > >
> > > The problems are varied enough that I suspect bad hardware, but would
> > > flaky RAM cause such similar failures repeatedly? And is there a way
> > > to test RAM explicitly?
> > >
> > > Any tips appreciated, either to me (ebuddington@wesleyan.edu) or to
> > > the list.
> >
> > Silly question, but is math emulation enabled?
> 
> Yes. I should have mentioned that in the original, since it occurred
> to me also.
> 
> -Eric

Try running ls under gdb and find out what instruction is causing SIGILL
(illegal opcode).  It is possible that it was compiled to use
instructions available only on later processors, or it could potentially
be a bug in the math emulation code.

--

				Brian Gerst

Re: 386 'ls' gets SIGILL iff /proc is mounted

[Eric Buddington]

On Wed, Mar 28, 2001 at 10:09:11AM -0800, Ulrich Drepper wrote:
> Eric Buddington <eric@sparrow.nad.adelphia.net> writes:
> 
> > OK. Context again (since I clipped preceding notes): 386SX/20 nfsroot,
> > getting SIGILL on lots of processes, math emulation is enabled, ls and
> > glibc were compiled with '-march=i386 -mcpu=i386' to be sure.
> 
> This isn't enough.  You compiled glibc on a i586 or later without
> telling configure to use the i386 configuration.  Run configure with
> the usual options and
> 
>    i386-linux-gnu
> 
> at the end of the command line and then use the options above.

This seems to have solved the problem. Thanks!

-Eric