Re: Re: [PATCH -tip/x86/jumplabel] x86: call out into int3 handler directly instead of using notifier

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

(2013/07/23 6:00), Jiri Kosina wrote:
> On Mon, 22 Jul 2013, H. Peter Anvin wrote:
> 
>>  What is the baseline for this patch?
> 
> Hi!
> 
> it's x86/jumplabel branch, as a followup to commit 
> fd4363fff3d96 ("x86: Introduce int3 (breakpoint)-based instruction 
> patching") sitting there.

AFAICS, tip/master already merged Jiri's works on tip/x86/jumplabel.
Thus both branches need this. (especially, since this is actual
bugfix, it should go into tip/master.)


> This branch however seems to be broken by Ingo's mismerge; this e-mail 
> from earlier today:
> 
> 	https://lkml.org/lkml/2013/7/22/147
> 
> implied that he fixed this ("all is fine"), however x86/jumplabel still 
> contains just 3/3 of Masami's series, and therefore is broken; 3/3 is not 
> enough, it needs also
> 
> 	kprobes/x86: Remove an incorrect comment about int3 in NMI/MCE
> 	kprobes/x86: Use text_poke_bp() instead of text_poke_smp*()
> 
> included or
> 
> 	kprobes/x86: Remove unused text_poke_smp() and text_poke_smp_batch() functions
> 
> dropped to be in consistent state again.

I hope to pull entire of this series in :)

> 
> I pinged Ingo about this:
> 
> 	https://lkml.org/lkml/2013/7/22/281
> 
> but the branch still seems to be in an odd state, containing last patch of 
> Masami's series.
> 
> Thanks in advance for fixing this and applying mi fix as well,	
> 

Thanks!

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