FreeBSD-SA-06:14.fpu

[Rostislav Krasny]

On Fri, 19 May 2006 08:02:08 +0800
David Xu <davidxu at freebsd.org> wrote:

> On Friday 19 May 2006 07:37, Rostislav Krasny wrote:
> > On Sun, 30 Apr 2006 16:19:29 -0700
> >
> > Colin Percival <cperciva at freebsd.org> wrote:
> > > Rostislav Krasny wrote:
> > > > Other possible solution is making the fpu_clean_state() optional by
> > > > something like following:
> > > >
> > > > #ifdef BUG_FXSAVE
> > > > #define fpu_clean_state() __fpu_clean_state()
> > > > #else
> > > > #define fpu_clean_state() ;
> > > > #endif
> > > >
> > > > ... and including "options BUG_FXSAVE" to GENERIC.
> > >
> > > Yes, this is probably the right solution.  My priority was to fix the
> > > bug; optimizing performance comes second.
> >
> > Ok. Is this solution going to be done some day? I could try to make a
> > patch but I'm not familiar with the build infrastructure internals.
> >
> > P.S. what is a better option name: "options BUG_FXSAVE" or "options
> > AMD_FXSAVE"?
> 
> Patch is welcome,

Ok. What should I do before adding the above #ifdef's? Is it enough to
change following files only?

src/sys/conf/options
src/sys/i386/conf/GENERIC
src/sys/i386/conf/NOTES
src/sys/i386/conf/PAE
src/sys/i386/conf/SMP
src/sys/amd64/conf/GENERIC
src/sys/amd64/conf/NOTES
src/sys/amd64/conf/SMP

> but I would call it BUG_FXSAVE like Linux's select()
> changed timeout value which only added incompatibility rather than
> advantage.

I don't know about Linux's select() but according to what I've seen on
their "7466f9e72dac13452d871a3fb72fc7bd9c93c864" commit they use
X86_FEATURE_FXSAVE_LEAK. What do you think about FXSAVE_LEAK instead of
BUG_FXSAVE?