On Tue, 14 Aug 2007, Luck, Tony wrote:
> I re-tried the macros ... the three warnings from mm/slub.c all result in
> broken code ... and quite rightly too, they all come from code that does:
>
> atomic_read(&n->nr_slabs)
>
> But the nr_slabs field is an atomic_long_t, so we shouldn't be using
> atomic_read(). I didn't spot these last time around because I was using
> slab, not slub for the previous build.
Hmmmm... Strange that this did not cause failures before on any other
platforms?
Fix atomic_read's in slub
Signed-off-by: Christoph Lameter <[EMAIL PROTECTED]>
diff --git a/mm/slub.c b/mm/slub.c
index 69d02e3..0c106d7 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -3112,7 +3112,7 @@ static int list_locations(struct kmem_cache *s, char *buf,
unsigned long flags;
struct page *page;
- if (!atomic_read(&n->nr_slabs))
+ if (!atomic_long_read(&n->nr_slabs))
continue;
spin_lock_irqsave(&n->list_lock, flags);
@@ -3247,7 +3247,7 @@ static unsigned long slab_objects(struct kmem_cache *s,
}
if (flags & SO_FULL) {
- int full_slabs = atomic_read(&n->nr_slabs)
+ int full_slabs = atomic_long_read(&n->nr_slabs)
- per_cpu[node]
- n->nr_partial;
@@ -3283,7 +3283,7 @@ static int any_slab_objects(struct kmem_cache *s)
for_each_node(node) {
struct kmem_cache_node *n = get_node(s, node);
- if (n->nr_partial || atomic_read(&n->nr_slabs))
+ if (n->nr_partial || atomic_long_read(&n->nr_slabs))
return 1;
}
return 0;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
