Re: Reorganize perf kernel side

[Borislav Petkov <bp@alien8.de>]

On Fri, Dec 04, 2015 at 11:09:07PM +0100, Peter Zijlstra wrote:
> I _will_ blame you for at least a month after every time I mistype a
> pathname because of this ;-)

Oh boy, I probably shouldn't do it after all out of fear of you coming
over ... :-)

> git blame --follow must keep working. That is, git had better be able to
> understand this code movement, loosing history is just a total PITA.

Well, I did

git annotate arch/x86/kernel/cpu/perf_event.c > blame.before

git annotate arch/x86/perf/perf_event.c > blame.after

The diff is:

--- /tmp/blame.before   2015-12-04 23:30:28.502701279 +0100
+++ /tmp/blame.after    2015-12-04 23:29:53.618702295 +0100
@@ -37,7 +37,7 @@ e3f3541c19c89 (Peter Zijlstra 2011-11-21
 d07bdfd322d30  (Peter Zijlstra 2012-07-10 09:42:15 +0200       37)#include <asm/desc.h>
 d07bdfd322d30  (Peter Zijlstra 2012-07-10 09:42:15 +0200       38)#include <asm/ldt.h>
 241771ef016b5  (Ingo Molnar    2008-12-03 10:39:53 +0100       39)
-de0428a7ad485  (Kevin Winchester       2011-08-30 20:41:05 -0300       40)#include "perf_event.h"
+b5e0c1bab8637  (Borislav Petkov        2015-12-04 22:02:51 +0100       40)#include "../kernel/cpu/perf_event.h"
 de0428a7ad485  (Kevin Winchester       2011-08-30 20:41:05 -0300       41)
 de0428a7ad485  (Kevin Winchester       2011-08-30 20:41:05 -0300       42)struct x86_pmu x86_pmu __read_mostly;
 efc9f05df2dd1  (Stephane Eranian       2011-06-06 16:57:03 +0200       43)

and that include file change is only temporary in order to keep the
churn at the lowest level.

Is this what you had in mind?

> Also, a script that can auto-convert patches would be nice.

Sure, I can do that.

Also, I was thinking of doing this in 4-5 patches sets only so that we
can keep everything easily manageable.

For sure, I don't want to do one insane move in one go and cause
needless churn and regressions.

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.