[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: openssl Illegal instruction with sid version



On Sat, 2003-01-25 at 15:31, Rob Weir wrote:
> > openssl req -new -x509 -keyout demoCA/private/cakey.pem -out
> > demoCA/cacert.pem -days 3650
> > Illegal instruction
> > 
> > So then I tried running:
> > openssl version
> > Illegal instruction
> > 
> > So I figure that there is a problem with the sid version of openssl on
> > my system.
> > 
> > I have both libc6 and libssl0.9.7 from unstable.
> > 
> > Does anyone have any idea what I should check next?
> 
> Ooooh, ooooh, I know this one!  Are you using a VIA C3 or such chip?
> They apparently don't support an assembly code instruction called CMOV,
> but the dynamic loader, OpenSSL and GCC (incorrectly) assume it does, so
> it generates code the CPU can't execute.  It was the subject of a rather
> long discussion on d-d, and still doesn't seem to have been properly
> resolved.  If you're interested, it's bug #164766.  I think you'll have
> to downgrade to the last version before the bug was filed (sorry, I'm
> offline ATM, so I can't find the version number for you) until it's
> properly fixed. 
> 
You hit the nail on the head!  I was just setting things up on an VIA
epia 5000 mini-itx board.  It does indeed have a C3 chip.

Hmmm.  I have just spent the last day re-installing everything as a
Woody system and compiled slapd so I could get TLS support.  Everything
works OK now with the Woody version of openssl.  I guess it doesn't have
the bug.

I am glad to hear it was a bug and not just me stuffing up (again).  I
think I will stick with the Woody versions of everything now I have them
working.

Thanks for the reply.  I didn't see any mention of the C3 related
problem with my Google searches.  Although I wasn't specifically
searching for that.

Regards.
Mark.



Reply to: