Re: Conflict between tip/tracing/core and tip/perf/core

[Frederic Weisbecker <fweisbec@gmail.com>]

On Mon, May 17, 2010 at 02:50:42PM -0400, Steven Rostedt wrote:
> Hi,
> 
> Ingo asked me to resolve a conflict between tip/tracing/core and
> perf/core, and I came up with the below solution.
> 
> The conflict stems from the shrinking of TRACE_EVENT(), which affects
> both ftrace and perf (saves size on both too). It conflicts with:
> 
> 6cc8a7c1d8560c042f486b23318a6291569ab96b
> Author: Frederic Weisbecker <fweisbec@gmail.com>
> Date:   Fri Mar 19 01:23:53 2010 +0100
> perf: Fetch hot regs from the template caller
> 
> 
> The shrinking code removed the per event caller to the template
> (TRACE_CLASS). This was done because the shrinking code allows the trace
> event to be passed to the tracepoint probe, and removed the need to have
> a separate function for every event because the class can now have the
> event passed to it.
> 
> The conflicting code added the regs to the per event probe, which no
> longer exists.
> 
> Masami,
> 
> It also conflicted with the kprobe code, which is also in the fix up.
> 
> 
> Here's my conflict resolution:
> 
> is everyone fine with it?
> 
> -- Steve



Yep, no problem with it,

Acked-by: Frederic Weisbecker <fweisbec@gmail.com>


Nothing related to this conflict resolution, but we have these
per cpu regs that are racy against NMI, I'll need to solve that
soon.

Thanks.