bpf: permit multiple bpf attachments for a single perf event
authorYonghong Song <yhs@fb.com>
Tue, 24 Oct 2017 06:53:08 +0000 (23:53 -0700)
committerDavid S. Miller <davem@davemloft.net>
Wed, 25 Oct 2017 01:47:47 +0000 (10:47 +0900)
commite87c6bc3852b981e71c757be20771546ce9f76f3
treebad3be630137d8e873f4ad5a1ea77b4aa1853184
parent0b4c6841fee03e096b735074a0c4aab3a8e92986
bpf: permit multiple bpf attachments for a single perf event

This patch enables multiple bpf attachments for a
kprobe/uprobe/tracepoint single trace event.
Each trace_event keeps a list of attached perf events.
When an event happens, all attached bpf programs will
be executed based on the order of attachment.

A global bpf_event_mutex lock is introduced to protect
prog_array attaching and detaching. An alternative will
be introduce a mutex lock in every trace_event_call
structure, but it takes a lot of extra memory.
So a global bpf_event_mutex lock is a good compromise.

The bpf prog detachment involves allocation of memory.
If the allocation fails, a dummy do-nothing program
will replace to-be-detached program in-place.

Signed-off-by: Yonghong Song <yhs@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/linux/bpf.h
include/linux/trace_events.h
include/trace/perf.h
kernel/bpf/core.c
kernel/events/core.c
kernel/trace/bpf_trace.c
kernel/trace/trace_kprobe.c
kernel/trace/trace_syscalls.c
kernel/trace/trace_uprobe.c