On 06/02/2023, Yonghong Song wrote:
On 2/6/23 5:35 AM, Jon Doron wrote:
From: Jon Doron <jond@xxxxxx>
Add option to set when the perf buffer should wake up, by default the
perf buffer becomes signaled for every event that is being pushed to it.
In case of a high throughput of events it will be more efficient to wake
up only once you have X events ready to be read.
So your application can wakeup once and drain the entire perf buffer.
Signed-off-by: Jon Doron <jond@xxxxxx>
LGTM with one possible change below.
Acked-by: Yonghong Song <yhs@xxxxxx>
---
tools/lib/bpf/libbpf.c | 9 +++++++--
tools/lib/bpf/libbpf.h | 3 ++-
2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index eed5cec6f510..cd0bce5482b2 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -11710,17 +11710,22 @@ struct perf_buffer *perf_buffer__new(int map_fd, size_t page_cnt,
const size_t attr_sz = sizeof(struct perf_event_attr);
struct perf_buffer_params p = {};
struct perf_event_attr attr;
+ __u32 sample_period;
if (!OPTS_VALID(opts, perf_buffer_opts))
return libbpf_err_ptr(-EINVAL);
+ sample_period = OPTS_GET(opts, sample_period, 1);
+ if (!sample_period)
+ sample_period = 1;
+
memset(&attr, 0, attr_sz);
attr.size = attr_sz;
attr.config = PERF_COUNT_SW_BPF_OUTPUT;
attr.type = PERF_TYPE_SOFTWARE;
attr.sample_type = PERF_SAMPLE_RAW;
- attr.sample_period = 1;
- attr.wakeup_events = 1;
+ attr.sample_period = sample_period;
+ attr.wakeup_events = sample_period;
p.attr = &attr;
p.sample_cb = sample_cb;
diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
index 8777ff21ea1d..5d3b75a5acde 100644
--- a/tools/lib/bpf/libbpf.h
+++ b/tools/lib/bpf/libbpf.h
@@ -1246,8 +1246,9 @@ typedef void (*perf_buffer_lost_fn)(void *ctx, int cpu, __u64 cnt);
/* common use perf buffer options */
struct perf_buffer_opts {
size_t sz;
+ __u32 sample_period;
};
The data structure now may be 16 bytes for 64bit system and we have
4 byte padding at the end which could be arbitrary value. The libbpf
convention is to add "size_t :0;" at the end of structure to zero
out tail padding during declaration.
Done
-#define perf_buffer_opts__last_field sz
+#define perf_buffer_opts__last_field sample_period
/**
* @brief **perf_buffer__new()** creates BPF perfbuf manager for a specified
