Re: [PATCH bpf-next] bpf: allow bpf_perf_event_read_value in all BPF programs

On 2/15/20 12:41 AM, Song Liu wrote:

bpf_perf_event_read_value() is NMI safe. Enable it for all BPF programs.
This can be used in fentry/fexit to profile BPF program and individual
kernel function with hardware counters.

Cc: Yonghong Song <redacted>
Signed-off-by: Song Liu <redacted>

And ironically, we had that already for perf_event_read. Applied, thanks!