Re: [PATCH 0/4] perf record: Cleanups and mmap-based output

[Jiri Olsa <jolsa@redhat.com>]

On Wed, Nov 06, 2013 at 11:41:33AM -0700, David Ahern wrote:
> I know Jiri is working on cleanups of the output file, but had this
> sitting around for a couple of weeks now. Might as well push it out
> for the next baseline. The cleanups of perf-record can be taken
> independently.
> 
> Ingo: I took a look at leveraging the copy_user_nocache and was not able
>       to get it to work. I won't have time to come back to it for a while.
>       Given that the mmap output already improves perf-trace a lot I would
>       like to get the option into 3.13 and come back to the optimization
>       later.
> 
> David Ahern (4):
>   perf record: Refactor feature handling into a separate function
>   perf record: Remove advance_output function
>   perf record: Remove post_processing_offset variable
>   perf record: mmap output file - v3
> 
>  tools/perf/Documentation/perf-record.txt |   5 ++
>  tools/perf/builtin-record.c              | 137 ++++++++++++++++++++++++++-----
>  2 files changed, 123 insertions(+), 19 deletions(-)
> 

hi,
got this on F19:

  CC       builtin-record.o
builtin-record.c: In function ‘do_mmap_output’:
builtin-record.c:118:13: error: ignoring return value of ‘ftruncate’, declared with attribute warn_unused_result [-Werror=unused-result]
    ftruncate(file->fd, offset);
             ^
cc1: all warnings being treated as errors


jirka