Re: [PATCH 06/13] cpuset: cleanup cpuset[_can]_attach()

From: Rusty Russell
Date: Wed Jan 02 2013 - 00:12:30 EST

Tejun Heo <tj@xxxxxxxxxx> writes:

> (cc'ing Rusty, hi!)
> Hello, Li.
> On Wed, Dec 26, 2012 at 06:20:11PM +0800, Li Zefan wrote:
>> On 2012/11/29 5:34, Tejun Heo wrote:
>> > cpuset_can_attach() prepare global variables cpus_attach and
>> > cpuset_attach_nodemask_{to|from} which are used by cpuset_attach().
>> > There is no reason to prepare in cpuset_can_attach(). The same
>> > information can be accessed from cpuset_attach().
>> >
>> > Move the prepartion logic from cpuset_can_attach() to cpuset_attach()
>> > and make the global variables static ones inside cpuset_attach().
>> >
>> > While at it, convert cpus_attach to cpumask_t from cpumask_var_t.
>> > There's no reason to mess with dynamic allocation on a static buffer.
>> >
>> But Rusty had been deprecating the use of cpumask_t. I don't know why
>> the final deprecation hasn't been completed yet.
> Hmmm? cpumask_t can't be used for stack but other than that I don't
> see how it would be deprecated completely. Rusty, can you please
> chime in?

The long-never-quite-complete-plan was for struct cpumask to be
undefined when CONFIG_CPUMASK_OFFSTACK=y. That means noone can declare
them, or pass them on the stack, since they'll get a compiler error.

Now, there are some cases where it really is a reason to use a static
bitmap, and 1/2 a K of wasted space be damned. There's a
deliberately-ugly way of doing that: declare a bitmap and use
to_cpumask(). Of course, if we ever really want to remove NR_CPUS and
make it completely generic, we have to kill all these too, but noone is
serious about that.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at