Re: amd64 profiling kernel build failure

[christos%astron.com@localhost (Christos Zoulas)]

In article <20160108173359.A539F1CC026%yaml.nerv.org@localhost>,
Ryo Shimizu  <ryo%nerv.org@localhost> wrote:
>
>Hi all
>
>>Hi,
>>
>>On 2016/01/08 16:00, David Holland wrote:
>>> On Fri, Jan 08, 2016 at 06:50:02AM +0000, David Holland wrote:
>>>  >  > --- a/sys/kern/subr_prof.c
>>>  >  > +++ b/sys/kern/subr_prof.c
>>>  >  > @@ -48,6 +48,10 @@ __KERNEL_RCSID(0, "$NetBSD: subr_prof.c,v
>1.47 2014/07/10 21:13:52 christos Exp
>>>  >  >  #include <sys/malloc.h>
>>>  >  >  #include <sys/gmon.h>
>>>  >  >  
>>>  >  > +#ifdef MULTIPROCESSOR
>>>  >  > +__cpu_simple_lock_t __mcount_lock;
>>>  >  > +#endif
>>>  >  > +
>>>  > 
>>>  > This should be in an MD file. Not sure offhand which one.
>>> 
>>> Also, the i386 profile.h needs the same change as the amd64 one, so
>>> the md file should probably be one in arch/x86/x86.
>
>BTW, as far as I know other MULTIPROCESSOR arch also needs
>__mcount_lock, but none.
>At least kernel profiling doesn't work on arm/MP.
>
>therefore __mcount_lock should be moved to common/lib/libc/gmon/mcount.c
>from machine/profile.h as below.

Go for it I guess.

christos