Re: [PATCH -tip v2 3/3] [BUGFIX] kprobes: Prohibit probing on func_ptr_is_kernel_text

[Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>]

(2013/11/06 15:07), Ingo Molnar wrote:
> 
> * Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> wrote:
> 
>>>> [...] I hope to build the list when the kernel build time if 
>>>> possible... Would you have any idea to classify some annotated(but no 
>>>> side-effect) functions?
>>>
>>> The macro magic I can think of would need to change the syntax of the 
>>> function definition - for example that is how the SYSCALL_DEFINE*() 
>>> macros work.
>>
>> Would you mean something like the below macro? :)
>>
>> NOKPROBE_SYMBOL(int, func_ptr_is_kernel_text)(void *ptr)
> 
> I think this is rather ugly and harder to maintain. The whole _point_ of 
> such annotations is to make them 'easy on the eyes', to make it easy to 
> skip a 'noinline', 'noprobe' or 'notrace' tag.
> 
> Using something like NOKPROBE_SYMBOL() makes the whole construct ugly and 
> attention seeking.

Hmm, by the way how about Steven's idea? A macro like EXPORT_SYMBOL?
At least for kprobes_blacklist, which is defined/maintained in kprobes.c
for some symbols(*), that is useful for updating it because we can
put it near the function definition.

* These symbols can not moves to other section because it already
in a different section.

Of course, still this is not a big problem since there are a few symbols
in the kprobe_blacklist.

> So until compilers get smarter (or there's some compiler trick I haven't 
> noticed) lets stay with the separate section - it's not the end of the 
> world, the (effective) 'noinline' aspect of noprobes changes code 
> generation anyway.

I see. :)

So, would you pull this series ? Or I need any update?

Thank you,

-- 
Masami HIRAMATSU
IT Management Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@hitachi.com