tip/x86/nmi nmi-32/64.c merge v3

This series tries to merge down nmi_32/64.c to nmi.c. It is done by small steps in hope it would help to bisect problems (which will appear anyway ;) Please review *carefully* since it is NMI what I've touched (and I'm not nmi specialist). So I really apreciate _ANY_ comments on this. The series is produced over today 'tip' tree branch 'x86/nmi'.

- Cyrill -

Cyrill,

very good job. I did a short review and found no obvious problems yet. Will have a closer look when I apply it.

Thanks,
tglx

[Thomas Gleixner - Sun, May 25, 2008 at 06:44:28PM +0200]

| On Sat, 24 May 2008, Cyrill Gorcunov wrote:
| 
| > This series tries to merge down nmi_32/64.c to nmi.c. It is done by small
| > steps in hope it would help to bisect problems (which will appear anyway ;)
| > Please review *carefully* since it is NMI what I've touched (and I'm not
| > nmi specialist). So I really apreciate _ANY_ comments on this. The series
| > is produced over today 'tip' tree branch 'x86/nmi'.
| 
| Cyrill,
| 
| very good job. I did a short review and found no obvious problems
| yet. Will have a closer look when I apply it.
| 
| Thanks,
| 	tglx
| 
| 
| 

thanks Thomas, unfortunelly I think the problems will arise soon, I've just noticed the commit

commit 57b2b8947a5f68e59d5f50ab749efdc79d9ba040

Author: Mike Travis <[email protected]>
Date:   Mon May 12 21:21:12 2008 +0200

x86: fix remove cpu_pda table patch

in tip/testing tree

so we will need some workaround for check_nmi_watchdog() 'cause of changing

...
- prev_nmi_count = kmalloc(NR_CPUS * sizeof(int), GFP_KERNEL);
+ prev_nmi_count = kmalloc(nr_cpu_ids * sizeof(int), GFP_KERNEL);
...
- for (cpu = 0; cpu < NR_CPUS; cpu++)
+ for (cpu = 0; cpu < nr_cpu_ids; cpu++)
...

didn't take a precise look on this, but will take as only be able to ;)

- Cyrill -