Re: [tip:x86/mm] x86, mm, kprobes: fault.c, simplify notify_page_fault()

[Ingo Molnar <mingo@elte.hu>]

* Masami Hiramatsu <mhiramat@redhat.com> wrote:

> Ingo Molnar wrote:
> > Author:     Ingo Molnar <mingo@elte.hu>
> > AuthorDate: Fri, 20 Feb 2009 22:42:57 +0100
> > Commit:     Ingo Molnar <mingo@elte.hu>
> > CommitDate: Sat, 21 Feb 2009 00:09:42 +0100
> > 
> > x86, mm, kprobes: fault.c, simplify notify_page_fault()
> > 
> > Impact: cleanup
> > 
> > Remove an #ifdef from notify_page_fault(). The function still
> > compiles to nothing in the !CONFIG_KPROBES case.
> > 
> > Introduce kprobes_built_in() and kprobe_fault_handler() helpers
> > to allow this - they returns 0 if !CONFIG_KPROBES.
> > 
> > No code changed:
> > 
> >    text	   data	    bss	    dec	    hex	filename
> >    4618	     32	     24	   4674	   1242	fault.o.before
> >    4618	     32	     24	   4674	   1242	fault.o.after
> 
> It seems good for me. Thank you for cleanup!
> 
> Acked-by: Masami Hiramatsu <mhiramat@redhat.com>

another very small thing, while we are discussing kprobes:

I always found that the __kprobes annotation is very confusingly 
euphemistic: what those annotations really mean is not 
'kprobes', but 'no kprobes'.

So how about renaming __kprobes to __nokprobes, similar to how 
we have the notrace attribute?

We have about 350 __kprobes annotations in the kernel, so 
renaming it now would not be practical - but any objections 
against me sending Linus a rename patch somewhere late in the 
next merge window that just does this rename?

[ likewise, i'll rename notrace to __notrace to make it visually 
  less intrusive to the return value type. There's a lot less 
  such annotations in the kernel. ]

	Ingo