In message <16141.43130.657025.952793@napali.hpl.hp.com> you write:
> You mean there would be three primitives:
>
> (1) get value from a per-CPU variable
> (2) set value of a per-CPU variable
> (3) get the (canonical) address of a per-CPU variable
>
> ?
Almost, #3 would actually be "get lvalue", as now.
The only problem is that a quick audit of 2.5.75 reveals 16 places in
generic code where set/get primitives would work without jumping
through hoops, out of 43.
New idea:
Provide cpu_local_inc(var)/cpu_local_dec(var) and __cpu_local_inc(var)
and __cpu_local_dec(var). Generic implementation:
/* This is local.h */
typedef atomic_t local_t;
#define local_inc(l) atomic_inc(l)
#define local_dec(l) atomic_dec(l)
/* l is a per-cpu local_t. Increment it atomically. */
#define cpu_local_inc(l) local_inc(&__get_cpu_var(l))
#define cpu_local_dec(l) local_dec(&__get_cpu_var(l))
/* Increment non-atomically (must have preempt disabled) */
#define __cpu_local_inc(l) \
atomic_set(&__get_cpu_var(l), atomic_read(&__get_cpu_var(l)) + 1)
#define __cpu_local_dec(l) \
atomic_set(&__get_cpu_var(l), atomic_read(&__get_cpu_var(l)) - 1)
This catches one common case for ia64 to use the magic pinned area,
and also allows x86 to use its incl/decl instructions, which both
networking stats and module refcounts have wanted for a while.
Thoughts?
Rusty.
-- Anyone who quotes me in their sig is an idiot. -- Rusty Russell. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Jul 15 2003 - 22:00:38 EST