Forw: Re: [17/19] ftrace: dynamic enabling/disabling of function calls

[Masami Hiramatsu]

Frank Ch. Eigler wrote:
> The following blurb from LKML suggests a possible way to implement
> batched kprobe (un)registration too.

I think the current issue on the batched kprobe feature is just its
interface. Kprobes itself can be disabled/enabled immediately, but for
unregistering from the RCU list, we need to wait the scheduler on each CPU.
I had implemented it, but the discussion about interface had not ended.
I'm not so sure that stop_machine can solve the RCU list issue.
Anyway, I'll rework it again.

Thanks,

-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com