Current-Users archive

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

Re: Build failure for port-alpha



On Tue, Jan 08, 2013 at 01:57:10PM +0100, Havard Eidnes wrote:
> > On Mon, Jan 07, 2013 at 03:52:14PM +0100, Havard Eidnes wrote:
> >> However, alpha's cpu.h includes <sys/cpu_data.h> under _KMEMUSER,
> >> and it looks like "struct cpu_data ci_data" needs to be part of
> >> what's visible of "struct cpu_info" under _KMEMUSER (I tried to
> >> move the #ifndef but that lead to more problems).
> >
> > If the cpu_data.h is needed by kmem grovelers, it should be installed to
> > userland.
> >
> > (The change exposing this now is probably dsl's adding of _KMEMUSER to a
> > few binaries that realy need it but missed it before.)
> >
> > If alpha cpuinfo should be rearranged to not require this file instead - 
> > I don't know.
> 
> The thing is...
> 
> sys/sys/Makefile has this log message:
> 
> revision 1.68
> date: 2004/10/10 08:57:30;  author: yamt;  state: Exp;  lines: +2 -2
> don't install cpu_data.h or cc_microtime.h.
> they're not intended to be exposed to userland.
> 
> I'm reluctant to reverse that decision if it's just alpha which
> needs cpu_data.h to be installed.
> 
> I have tried various other types of re-arrangement of the alpha
> cpu.h file, none of them successful -- it either falls over while
> trying to build libkvm, vmstat or pstat.
> 
> I'm therefore asking for input as to the right course of action.

I got pstat to compile by dropping in an #ifdef _KERNEL around the include.
But that doesn't explain why anything that included sys/param.h builds
(and almost everything does!).

I did see amd64/i386 trying to include a missing file, I didn't
completely track that one down - I suspect something was defining an
include guard.

        David

-- 
David Laight: david%l8s.co.uk@localhost


Home | Main Index | Thread Index | Old Index