Patch applied, thanks.
On Fri, May 7, 2010 at 4:28 AM, Corey Ashford
<cjash...@linux.vnet.ibm.com> wrote:
> perf_get_group_nevents was returning the wrong number of events
> when the group was the last in the list - it always returned 1
> instead of the true number.
>
> Also, when reading up event groups, only a "max size" (sz) was
> being passed to the read() function, which resulted in getting
> fewer bytes than requested and a resulting warning when groups with
> fewer events follow groups with more events.
>
> Signed-off-by: Corey Ashford <cjash...@linux.vnet.ibm.com>
> ---
> perf_examples/perf_util.c | 2 +-
> perf_examples/task.c | 6 +++---
> 2 files changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/perf_examples/perf_util.c b/perf_examples/perf_util.c
> index 9c99b15..b84f0e2 100644
> --- a/perf_examples/perf_util.c
> +++ b/perf_examples/perf_util.c
> @@ -161,7 +161,7 @@ perf_get_group_nevents(perf_event_desc_t *fds, int num,
> int idx)
> return i - leader;
> }
> }
> - return 1;
> + return i - leader;
> }
>
> int
> diff --git a/perf_examples/task.c b/perf_examples/task.c
> index c66e3ed..75de973 100644
> --- a/perf_examples/task.c
> +++ b/perf_examples/task.c
> @@ -99,14 +99,14 @@ read_groups(perf_event_desc_t *fds, int num)
> if (!values)
> err(1, "cannot allocate memory for values\n");
>
> - ret = read(fds[evt].fd, values, sz);
> - if (ret != sz) { /* unsigned */
> + ret = read(fds[evt].fd, values, new_sz);
> + if (ret != new_sz) { /* unsigned */
> if (ret == -1)
> err(1, "cannot read values event %s",
> fds[evt].name);
>
> /* likely pinned and could not be loaded */
> warnx("could not read event %d, tried to read %d
> bytes, but got %d",
> - evt, (int)sz, ret);
> + evt, (int)new_sz, ret);
> }
>
> /*
> --
> 1.7.0.4
>
>
------------------------------------------------------------------------------
_______________________________________________
perfmon2-devel mailing list
perfmon2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
