Hi.

On Sun, 2007-05-27 at 22:10 +0100, Julian Sikorski wrote:
quoted text
> Pavel Machek pisze:
> > Hi!
> >=20
> >> As promised I took another look at the patch and at what Randy had
> >> prepared to fix the IA64 compilation error. I did some more work on it=
,
quoted text
> >> and believe that the following is the tidiest correct solution I can
> >> come up with. It differs from the version that caused the compilation
> >> error primarily in that:
> >>
> >> * the #include <asm/resume-trace.h> is inside the #ifdef=20
> >>   CONFIG_PM_TRACE.
> >> * now-unnecessary protection for multiple #includes and ifdef testing =
of
quoted text
> >>   CONFIG_PM_TRACE in the asm code were removed.
> >> * do-nothing definitions for !PM_TRACE restored to
> >>   include/linux/resume-trace.h.
> >>
> >> We're therefore depending upon kernel/power/Kconfig having the right
> >> depends condition. As far as I can see, IA64 doesn't define CONFIG_X86=
.
quoted text
> >> Is that correct, or do we need to have (X86 && !IA64)?
> >=20
> > ia64? did you mean x86-64?
> >=20
> > Otherwise looks ok to me.
> >=20
> IIRC enabling pm_trace on x86_64 was breaking compilation on ia64, so I
> think Nigel meant the latter.

Yes, it was breaking ia64, so I meant what I said - I can test x86_64
easily. I was concerned about ensuring the condition was right for ia64.

Regards,

Nigel