2009-07-11 08:02:26

by Jaswinder Singh Rajput

[permalink] [raw]
Subject: [FIX REQUEST] x86: smp.h is totally confused about processor ID


Latest -tip tree shows
http://git.kernel.org/?p=linux/kernel/git/x86/linux-2.6-tip.git;a=blob;f=arch/x86/include/asm/smp.h;h=6a84ed166aec136334a644cc4fbaa676368ae983;hb=HEAD :


27 DECLARE_PER_CPU(int, cpu_number);
28
29 static inline struct cpumask *cpu_sibling_mask(int cpu)
30 {
31 return per_cpu(cpu_sibling_map, cpu);
32 }
33
34 static inline struct cpumask *cpu_core_mask(int cpu)
35 {
36 return per_cpu(cpu_core_map, cpu);
37 }

it should be unsigned int cpu/cpu_number as it is non-negative.


47
48 struct smp_ops {
49 void (*smp_prepare_boot_cpu)(void);
50 void (*smp_prepare_cpus)(unsigned max_cpus);
51 void (*smp_cpus_done)(unsigned max_cpus);
52
53 void (*smp_send_stop)(void);
54 void (*smp_send_reschedule)(int cpu);
55
56 int (*cpu_up)(unsigned cpu);
57 int (*cpu_disable)(void);
58 void (*cpu_die)(unsigned int cpu);
59 void (*play_dead)(void);
60
61 void (*send_call_func_ipi)(const struct cpumask *mask);
62 void (*send_call_func_single_ipi)(int cpu);
63 };

In same structure it using int, unsigned and unsigned int cpu

It should be consistent either unsigned cpu or unsigned int cpu.

I think unsigned int cpu will be more better choice.

Similarly another instances in same file.

Thanks,
--
JSR




2009-07-11 11:06:39

by Rusty Russell

[permalink] [raw]
Subject: Re: [FIX REQUEST] x86: smp.h is totally confused about processor ID

On Sat, 11 Jul 2009 05:30:34 pm Jaswinder Singh Rajput wrote:
> In same structure it using int, unsigned and unsigned int cpu

I prefer unsigned int, but historically it's been very loose. FWIW, I have no
problems with a cleanup either way.

Thanks,
Rusty.