Re: [PATCH] Reflect cpupool in numa node affinity

[Juergen Gross <juergen.gross@ts.fujitsu.com>]

On 01/23/2012 11:27 AM, Jan Beulich wrote:
>>>> On 23.01.12 at 10:51, Juergen Gross<juergen.gross@ts.fujitsu.com>  wrote:
> @@ -365,15 +366,21 @@ void domain_update_node_affinity(struct
>> void domain_update_node_affinity(struct domain *d)
>> {
>>      cpumask_t cpumask;
>> +    cpumask_t online_affinity;
> If at all possible, please don't introduce new automatic cpumask_t
> variables. Allocating them will of course mean that the function can
> fail, and that callers need to deal with the failure. (Probably a prior
> patch should then first convert the 'cpumask' variable.)

In this case I don't think it is very complicated.
Not doing anything in domain_update_node_affinity() will just produce a
lower performance. So doing a return in case of an allocation failure should
be fine.

>> +    cpumask_t *online;
> const.

Okay.

>>      nodemask_t nodemask = NODE_MASK_NONE;
>>      struct vcpu *v;
>>      unsigned int node;
>>
>> +    online = (d->cpupool == NULL) ?&cpu_online_map : d->cpupool->cpu_valid;
> This construct (together with its brother using 'cpupool_free_cpus')
> meanwhile enjoys quite a number of instances - could it get abstracted
> into a pair of inline functions or macros?

Okay.


Juergen

-- 
Juergen Gross                 Principal Developer Operating Systems
PDG ES&S SWE OS6                       Telephone: +49 (0) 89 3222 2967
Fujitsu Technology Solutions              e-mail: juergen.gross@ts.fujitsu.com
Domagkstr. 28                           Internet: ts.fujitsu.com
D-80807 Muenchen                 Company details: ts.fujitsu.com/imprint.html