From: Kan Liang <kan.li...@intel.com>

Perf collects overhead information by default.
A new option --no-profile-cost-info is introduced to disable the
information collection.

Signed-off-by: Kan Liang <kan.li...@intel.com>
---
 tools/include/uapi/linux/perf_event.h    | 42 +++++++++++++++++++++++++++++++-
 tools/perf/Documentation/perf-record.txt |  6 +++++
 tools/perf/builtin-record.c              |  2 ++
 tools/perf/perf.h                        |  1 +
 tools/perf/util/evsel.c                  |  1 +
 5 files changed, 51 insertions(+), 1 deletion(-)

diff --git a/tools/include/uapi/linux/perf_event.h 
b/tools/include/uapi/linux/perf_event.h
index c66a485..bdf2eec 100644
--- a/tools/include/uapi/linux/perf_event.h
+++ b/tools/include/uapi/linux/perf_event.h
@@ -344,7 +344,8 @@ struct perf_event_attr {
                                use_clockid    :  1, /* use @clockid for time 
fields */
                                context_switch :  1, /* context switch data */
                                write_backward :  1, /* Write ring buffer from 
end to beginning */
-                               __reserved_1   : 36;
+                               overhead       :  1, /* Log overhead 
information */
+                               __reserved_1   : 35;
 
        union {
                __u32           wakeup_events;    /* wakeup every n events */
@@ -862,6 +863,17 @@ enum perf_event_type {
         */
        PERF_RECORD_SWITCH_CPU_WIDE             = 15,
 
+       /*
+        * Records perf overhead
+        * struct {
+        *      struct perf_event_header        header;
+        *      u64                             type;
+        *      struct perf_overhead_entry      entry;
+        *      struct sample_id                sample_id;
+        * };
+        */
+       PERF_RECORD_OVERHEAD                    = 16,
+
        PERF_RECORD_MAX,                        /* non-ABI */
 };
 
@@ -980,4 +992,32 @@ struct perf_branch_entry {
                reserved:44;
 };
 
+/*
+ * The overhead type could be different among architectures.
+ * The common overhead type can be defined from PERF_CORE_OVERHEAD
+ * The arch specific type should be defined from PERF_PMU_OVERHEAD
+ */
+enum perf_record_overhead_type {
+       PERF_CORE_OVERHEAD       = 0,
+       PERF_CORE_MUX_OVERHEAD   = 0,
+       PERF_CORE_SB_OVERHEAD,
+
+       PERF_PMU_OVERHEAD        = 20,
+       PERF_PMU_SAMPLE_OVERHEAD = 20,
+
+       PERF_OVERHEAD_MAX,
+};
+
+/*
+ * single overhead record layout:
+ *
+ *       nr: Times of overhead happens.
+ *           E.g. for NMI, nr == times of NMI handler are called.
+ *     time: Total overhead cost(ns)
+ */
+struct perf_overhead_entry {
+       __u64   nr;
+       __u64   time;
+};
+
 #endif /* _UAPI_LINUX_PERF_EVENT_H */
diff --git a/tools/perf/Documentation/perf-record.txt 
b/tools/perf/Documentation/perf-record.txt
index 27fc361..29a57f9 100644
--- a/tools/perf/Documentation/perf-record.txt
+++ b/tools/perf/Documentation/perf-record.txt
@@ -460,6 +460,12 @@ config terms. For example: 'cycles/overwrite/' and 
'instructions/no-overwrite/'.
 
 Implies --tail-synthesize.
 
+--no-profile-cost-info::
+Perf collects profiling time cost by default.
+The time cost includes sampling overhead, multiplexing overhead, side-band
+events overhead and so on.
+This option disable the profiling time cost information collection.
+
 SEE ALSO
 --------
 linkperf:perf-stat[1], linkperf:perf-list[1]
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 67d2a90..73b033b 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -1512,6 +1512,8 @@ struct option __record_options[] = {
                    "Switch output when receive SIGUSR2"),
        OPT_BOOLEAN(0, "dry-run", &dry_run,
                    "Parse options then exit"),
+       OPT_BOOLEAN(0, "no-profile-cost-info", &record.opts.no_profile_cost,
+                   "Do not collect perf profiling time cost information"),
        OPT_END()
 };
 
diff --git a/tools/perf/perf.h b/tools/perf/perf.h
index 9a0236a..59b16d3 100644
--- a/tools/perf/perf.h
+++ b/tools/perf/perf.h
@@ -55,6 +55,7 @@ struct record_opts {
        bool         all_user;
        bool         tail_synthesize;
        bool         overwrite;
+       bool         no_profile_cost;
        unsigned int freq;
        unsigned int mmap_pages;
        unsigned int auxtrace_mmap_pages;
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index b2365a63..146c191 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -817,6 +817,7 @@ void perf_evsel__config(struct perf_evsel *evsel, struct 
record_opts *opts,
        attr->sample_id_all = perf_missing_features.sample_id_all ? 0 : 1;
        attr->inherit       = !opts->no_inherit;
        attr->write_backward = opts->overwrite ? 1 : 0;
+       attr->overhead = !opts->no_profile_cost;
 
        perf_evsel__set_sample_bit(evsel, IP);
        perf_evsel__set_sample_bit(evsel, TID);
-- 
2.5.5

Reply via email to