[v4,0/2] tracing: make tracer_init_tracefs initcall asynchronous

Message

Mark-PK Tsai (蔡沛剛) April 26, 2022, 12:24 p.m. UTC

Move trace_eval_init() to subsys_initcall to make it start
earlier.
And to avoid tracer_init_tracefs being blocked by
trace_event_sem which trace_eval_init() hold [1],
queue tracer_init_tracefs() to eval_map_wq to let
the two works being executed sequentially.

It can speed up the initialization of kernel as result
of making tracer_init_tracefs asynchronous.

On my arm64 platform, it reduce ~20ms of 125ms which total
time do_initcalls spend.

Changes in v2:
- Fix the use of ftrace_eval_maps in INIT_DATA after initmem_free().

  In v1, a kthread is create for tracer_init_tracefs() which
  might cause kernel crash if initmem is free before it finish.
  In v2, the work of tracer_init_tracefs() is separated into 2 parts,
  one is put to subsys_initcall and the other is queue to eval_map_wq
  which can make sure it always finish before initmem is free.

Changes in v3:
- Avoid adding tracer option before __update_tracer_options finish to fix
  the warning report by kernel test robot.

Reported-by: kernel test robot <[e-mail]>
  https://lore.kernel.org/lkml/20220322133339.GA32582@xsang-OptiPlex-9020/

Changes in v4:
- Set `tracer_options_updated = true` before calling __update_tracer_options().
  Otherwise, __update_tracer_options() will skip all the tracer in trace_array.
- Init tracerfs_init_work only when eval_map_wq is successfully created.

Mark-PK Tsai (2):
  tracing: Avoid adding tracer option before update_tracer_options
  tracing: make tracer_init_tracefs initcall asynchronous

 kernel/trace/trace.c | 39 +++++++++++++++++++++++++++++----------
 1 file changed, 29 insertions(+), 10 deletions(-)