Re: [PATCH] cpumask 5/10 rewrite cpumask.h - single bitmap basedimplementation

[Nick Piggin]

Paul Jackson wrote:
> > I don't see what you gain from having the cpumask type but having
> > to get at its internals with the bitop functions.
>
>
> The essential gain, in my view, of cpumask, is that it encapsulates
> the value NR_CPUS.  cpumasks are bitmaps of length NR_CPUS.
>
> Yes, there is an open issue of whether cpumasks are worth it.
> I think enough code has taken to them that they are.

Yes, I'm all for the full cpumask abstraction.

> The getting at internals (via cpus_addr(), I'm guessing you mean)
> was a workaround for some code that messed with cpumasks and simple
> unsigned longs as if they were interoperable.  "cpus_addr" should
> be marked deprecated, and its use coded out.  Its remaining uses
> are in arch-specific areas where I lack the expertise and testing
> environment to accomplish such.
>
> I needed some legacy mechanism such as this, in order to avoid
> having such existing uses bring the entire cpumask overhaul to
> a screeching halt.

No, by getting at the internals, I mean the internals of the
type itself. Its implementation, if you will. (Well I guess
that also *includes* users getting the address and derefing it
as an unsigned long).

But no, I was talking about something more general. Rusty wrote:

>>+#define cpus_addr(src) ((src).bits)
>
>
> We've discussed this before when talking about whether it'd be easier to
> just make people use raw bitop functions directly, so I know we have
> philosophical differences here.
>
> So, opinion alert: if I were doing this, I'd probably live without this
> macro; in my mind it crosses the "too much abstraction" line.  I did
> momentarily wonder what this macro did when I saw it used in the
> succeeding patches.

Now in my opinion, it is either all or nothing. I could be wrong,
but I don't think there is any point with a nice cpumask type if
you are just going to get inside it and do bitmap operations on it.

In summary, I think your patches are nice :)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/