On 2/8/21 1:01 PM, Andrii Nakryiko wrote:
On Mon, Feb 8, 2021 at 1:00 PM Andrii Nakryiko
<andrii.nakryiko@xxxxxxxxx> wrote:
On Sat, Feb 6, 2021 at 9:06 AM Alexei Starovoitov
<alexei.starovoitov@xxxxxxxxx> wrote:
From: Alexei Starovoitov <ast@xxxxxxxxxx>
Since sleepable programs are now executing under migrate_disable
the per-cpu maps are safe to use.
Also made me wonder if PERF_EVENT_ARRAY map is usable in sleepable now?
maybe. Probably would need to add explicit preempt_disable to
__bpf_perf_event_output around perf_event_output.
It needs more analysis.
