Re: [RFC PATCH 0/5] Make eBPF programs output data to perf event

From: Alexei Starovoitov
Date: Wed Jul 01 2015 - 22:48:43 EST


On 7/1/15 4:58 AM, Peter Zijlstra wrote:

But why create a separate trace buffer, it should go into the regular
perf buffer.

+1

I think
+static char __percpu *perf_extra_trace_buf[PERF_NR_CONTEXTS];
is redundant.
It adds quite a bit of unnecessary complexity to the whole patch set.

Also the call to bpf_output_sample() is not effective unless program
returns 1. It's a confusing user interface.

Also you cannot ever do:
BPF_FUNC_probe_read,
+ BPF_FUNC_output_sample,
BPF_FUNC_ktime_get_ns,
new functions must be added to the end.

Why not just do:
perf_trace_buf_prepare() + perf_trace_buf_submit() from the helper?
No changes to current code.
No need to call __get_data_size() and other overhead.
The helper can be called multiple times from the same program.
imo much cleaner.

Also how about calling this helper:
bpf_trace_buf_submit(void *stack_ptr, int size) ?
bpf_output_sample, I think, is odd name. It's not a sample.
May be other name?

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/