Re: [PATCH] x86: kprobes change kprobe_handler flow

[Masami Hiramatsu <mhiramat@redhat.com>]

Hi Abhishek,

Masami Hiramatsu wrote:
>> Hmm, I can not agree, because it is possible to insert a kprobe
>> into kprobe's instruction buffer. If it should be a bug, we must
>> check it when registering the kprobe.
> 
> I discussed it with other maintainers and knew that current kprobes
> does not allow user to insert a kprobe to another kprobe's instruction
> buffer, because register_kprobe ensures the insertion address is text.
> Now I changed my mind. I think that case (p && kprobe_running() &&
> kcb->kprobe_status==KPROBE_HIT_SS) is BUG(), even if (*p->ainsn.insn ==
> BREAKPOINT_INSTRUCTION).

I could understand what the original code did at last.
If a kprobe is inserted on a breakpoint which other debugger inserts,
it single step inline instead of out-of-line.(this is done in prepare_singlestep)
In this case, (p && kprobe_running() && kcb->kprobe_status==KPROBE_HIT_SS)
is true and we need pass the control to the debugger.
And if (*p->ainsn.insn != BREAKPOINT_INSTRUCTION) (or (p != kprobe_running())) in
that case, there may be some bugs.

Now I think your original suggestion is correct.
Please fix it in another patch.

Thank you very much,

-- 
Masami Hiramatsu

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

e-mail: mhiramat@redhat.com, masami.hiramatsu.pt@hitachi.com