Re: [Patch 1/2] CPU hotplug compatible alloc_percpu

[Martin Peschke <mp3@de.ibm.com>]

Andrew Morton wrote:
> On Mon, 24 Jul 2006 19:16:54 +0200
> Martin Peschke <mp3@de.ibm.com> wrote:
> 
>> This patch splits alloc_percpu() up into two phases. Likewise for
>> free_percpu(). This allows clients to limit initial allocations to
>> online cpu's, and to populate or depopulate per-cpu data at run time as
>> needed:
>>
>>   struct my_struct *obj;
>>
>>   /* initial allocation for online cpu's */
>>   obj = percpu_alloc(sizeof(struct my_struct), GFP_KERNEL);
>>
>>   ...
>>
>>   /* populate per-cpu data for cpu coming online */
>>   ptr = percpu_populate(obj, sizeof(struct my_struct), GFP_KERNEL, cpu);
>>
>>   ...
>>
>>   /* access per-cpu object */
>>   ptr = percpu_ptr(obj, smp_processor_id());
>>
>>   ...
>>
>>   /* depopulate per-cpu data for cpu going offline */
>>   percpu_depopulate(obj, cpu);
>>
>>   ...
>>
>>   /* final removal */
>>   percpu_free(obj);
> 
> That looks pretty thorough.
> 
> The one little nit I'd have is that the code passes cpumasks by value.  See
> the tricks in <linux/cpumask.h> which pretend to take the caller's cpumask
> by value but which instead pass it via const reference to the callee.
> 
> CONFIG_NR_CPUS=1024 leads to a 128-byte cpumask_t.  It's worth doing.

Oops. I will send a patch.