Re: patched oskit10 for gcc-2.95.2

• To: Kalle Olavi Niemitalo <tosi@ees2.oulu.fi>
• Subject: Re: patched oskit10 for gcc-2.95.2
• From: Roland McGrath <roland@cs.utah.edu>
• Date: Wed, 10 May 2000 14:32:15 -0400 (EDT)
• Cc: oskit-users@fast.cs.utah.edu
• In-Reply-To: Kalle Olavi Niemitalo's message of , 10 May 2000 21:23:22 +0300 <87vh0mp8md.fsf@PC486.Niemitalo.local>
• Sender: owner-oskit-users@fast.cs.utah.edu

> >  - memtest asks for some parameter and then crashes.
> >    I haven't examined this yet.
> 
> Oh.  It tries to execute "rdtsc" which this 486 processor doesn't
> support.  Is there alternative code for time keeping?

The oskit code that uses rdtsc is the only thing there intended for
performance-level timing.  I think the off-chip clocks (PIT and RTC) are
the only other option.  Beware that this is probably not the only place in
the oskit that assumes a Pentium or later chip; 486 support is certainly
not a great burden and should be there, but I don't think the Flux group
has any 486 machines that have ever been used for testing (and it's
understandably not going to be a priority to support hardware worth far
less than a day's hacker time).  The earliest chip on which I have used
oskit (and thus oskit-mach) is an early Pentium (100MHz with F00F bug),
and I just gave away my 486 on Sunday. :-)

• Re: patched oskit10 for gcc-2.95.2
• From: Kalle Olavi Niemitalo <tosi@ees2.oulu.fi>

• Prev: Re: patched oskit10 for gcc-2.95.2
• Next: how to program vesa driver of oskit
• Index(es):
  • Main
  • Thread