Re: [PATCH] perf session: Try to read pipe data from file

[Ian Rogers]

On Mon, May 4, 2020 at 3:57 PM Jiri Olsa <jolsa@xxxxxxxxxx> wrote:
>
> On Fri, May 01, 2020 at 01:34:47PM +0200, Jiri Olsa wrote:
> > From: Jiri Olsa <jolsa@xxxxxxxxxx>
> >
> > Ian came with the idea of having support to read the pipe
> > data also from file [1]. Currently pipe mode files fails
> > like:
> >
> >   $ perf record -o - sleep 1 > /tmp/perf.pipe.data
> >   $ perf report -i /tmp/perf.pipe.data
> >   incompatible file format (rerun with -v to learn more)
> >
> > This patch adds the support to do that by trying the pipe
> > header first, and if its successfully detected, switching
> > the perf data to pipe mode.
> >
> > [1] https://lore.kernel.org/lkml/20200409185744.255881-1-irogers@xxxxxxxxxx/
> > Original-patch-by: Ian Rogers <irogers@xxxxxxxxxx>
> > Signed-off-by: Jiri Olsa <jolsa@xxxxxxxxxx>
>
> actualy.. I found another issue while trying this on tracepoints:
>
>   # ./perf record -g  -e 'raw_syscalls:sys_enter' -o -  true > data
>   [ perf record: Woken up 1 times to write data ]
>   [ perf record: Captured and wrote 0.000 MB - ]
>   # ./perf script -i ./data
>   perf_event__process_tracing_data: tracing data size mismatch0x1034 [0xc]: failed to process type: 66
>
> it's because some of the pipe synthesize code calls lseek, which
> fails on pipe, but succeeds on normal file (with pipe data)
>
> patch below fixes that for me, but I wonder there are other leftovers
> like this.. I'll check on post it all together

Thanks for testing! I wonder in the 2nd case whether a comment as to
why the seek isn't needed in pipe mode would be useful.

Ian

> jirka
>
>
> ---
> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> index 8ca709f938b8..33e299674121 100644
> --- a/tools/perf/util/header.c
> +++ b/tools/perf/util/header.c
> @@ -3955,13 +3955,8 @@ int perf_event__process_tracing_data(struct perf_session *session,
>  {
>         ssize_t size_read, padding, size = event->tracing_data.size;
>         int fd = perf_data__fd(session->data);
> -       off_t offset = lseek(fd, 0, SEEK_CUR);
>         char buf[BUFSIZ];
>
> -       /* setup for reading amidst mmap */
> -       lseek(fd, offset + sizeof(struct perf_record_header_tracing_data),
> -             SEEK_SET);
> -
>         size_read = trace_report(fd, &session->tevent,
>                                  session->repipe);
>         padding = PERF_ALIGN(size_read, sizeof(u64)) - size_read;
> diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
> index c11d89e0ee55..b75df19feaf1 100644
> --- a/tools/perf/util/session.c
> +++ b/tools/perf/util/session.c
> @@ -1543,7 +1543,8 @@ static s64 perf_session__process_user_event(struct perf_session *session,
>                 return 0;
>         case PERF_RECORD_HEADER_TRACING_DATA:
>                 /* setup for reading amidst mmap */
> -               lseek(fd, file_offset, SEEK_SET);
> +               if (!perf_data__is_pipe(session->data))
> +                       lseek(fd, file_offset, SEEK_SET);
>                 return tool->tracing_data(session, event);
>         case PERF_RECORD_HEADER_BUILD_ID:
>                 return tool->build_id(session, event);
>