Re: [PATCH 0/3] ring-buffer: less locking and only disable preemption

[Mathieu Desnoyers <compudj@krystal.dyndns.org>]

* Steven Rostedt (rostedt@goodmis.org) wrote:
> 
> On Sat, 4 Oct 2008, Mathieu Desnoyers wrote:
> > 
> > Or use this code, based on a temporary breakpoint, to do the code
> > patching (part of the -lttng tree). It does not require stop_machine at
> > all and is nmi safe.
> > 
> 
> When this is supported for all archs, and can be done at all functions 
> then I could use it.
> 

How about incrementally using this piece of infrastructure when
available on a given architecture ? This way we keep a sub-optimal
fall-back for archs which does not support NMI-safe code patching and
incrementally get the optimal behavior. Otherwise, we would require any
new architecture to implement that up-front, which I doubt is a good
idea.

> I may just have the arch specific code use it, but we'll see.
> 
> Also, how good is it at patching 20,000 call sites?
> 

Can be done really fast using a hash table, see my previous mail.

Mathieu

> -- Steve
> 

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68