Re: enhanced spinlock debugging code for intel

[Brent Baccala <baccala@freesoft.org>]

> Brent Baccala wrote:
> >
> > I've had to add some hideous code to get the processor ID:
> >
> >       #define my_processor_id (((int *)current)[13])
> >
> > since sched.h includes spinlock.h, so task_struct isn't defined when
> > this file is parsed, so we can't just dereference current to find the
> > processor ID.  Any better suggestions would be welcome.


I've been thinking more about my own problem here.

I think it could be solved by splitting the spinlock include file in
two:

  spinlockdef.h   - the structure definitions for spinlocks and their
initializers
  spinlock.h      - includes spinlockdef.h and defines the functions to
manipulate spinlocks

This would have to been done in include/linux, as well as all the
include/asm* directories.

Most stuff would include spinlock.h, get both files, and see no change.

sched.h would be changed to include spinlockdef.h, since that's all it
needs.

asm-i386/spinlock.h could then include sched.h and spinlockdef.h without
creating a self-referential loop, so smp_processor_id would work in this
file.

Comments?  a new include file in all the asm dirs?  think Linus would
take it?

-- 
                                        -bwb

                                        Brent Baccala
                                        baccala@freesoft.org

==============================================================================
       For news from freesoft.org, subscribe to announce@freesoft.org:
   
mailto:announce-request@freesoft.org?subject=subscribe&body=subscribe
==============================================================================