Re: [BUG] perf: arch_perf_out_copy_user default

[Will Deacon <will.deacon@arm.com>]

Hi Peter,

On Wed, Oct 30, 2013 at 08:16:22PM +0000, Peter Zijlstra wrote:
> On Wed, Oct 30, 2013 at 08:50:28PM +0100, Frederic Weisbecker wrote:
> > 
> > Would it make sense to rather make copy_from_user_nmi() to use a return value
> > pattern that is closer to those of the existing copy_from_user_*() ?
> 
> Yeah we can do that I suppose; copy_form_user_nmi() actually uses
> __copy_from_user_inatomic() since about a week.
> 
> Something like so I suppose.. please check, I'm in fail mode.
> 
> It looks like DEFINE_OUTPUT_COPY() functions already returned the bytes
> not copied, and all its users appear to indeed expect that.
> 
> ---
>  arch/x86/kernel/cpu/perf_event.c           |  4 ++--
>  arch/x86/kernel/cpu/perf_event_intel_ds.c  |  2 +-
>  arch/x86/kernel/cpu/perf_event_intel_lbr.c |  2 +-
>  arch/x86/lib/usercopy.c                    |  2 +-
>  arch/x86/oprofile/backtrace.c              |  4 ++--
>  kernel/events/internal.h                   | 35 ++++++++++++++++++++++--------
>  6 files changed, 33 insertions(+), 16 deletions(-)

The core changes look good to me, and I've tested that this still produces
callchain information for ARM. It's odd that this worked so well before...

  Acked-by: Will Deacon <will.deacon@arm.com>

Will