* SLUB: Define functions for cpu slab handling instead of using PageActive
@ 2007-05-15 3:00 Christoph Lameter
2007-05-15 4:54 ` Andrew Morton
0 siblings, 1 reply; 3+ messages in thread
From: Christoph Lameter @ 2007-05-15 3:00 UTC (permalink / raw)
To: akpm; +Cc: linux-mm
Use inline functions to access the per cpu bit. Intoduce the notion of
"freezing" a slab to make things more understandable.
Signed-off-by: Christoph Lameter <clameter@sgi.com>
---
mm/slub.c | 57 ++++++++++++++++++++++++++++++++++++++-------------------
1 file changed, 38 insertions(+), 19 deletions(-)
Index: slub/mm/slub.c
===================================================================
--- slub.orig/mm/slub.c 2007-05-14 19:46:56.000000000 -0700
+++ slub/mm/slub.c 2007-05-14 19:56:14.000000000 -0700
@@ -78,10 +78,18 @@
*
* Overloading of page flags that are otherwise used for LRU management.
*
- * PageActive The slab is used as a cpu cache. Allocations
- * may be performed from the slab. The slab is not
- * on any slab list and cannot be moved onto one.
- * The cpu slab may be equipped with an additioanl
+ * PageActive The slab is frozen and excempt from list processing.
+ * This means that the slab is dedicated to a purpose
+ * such as satisfying allocations for a specific
+ * processor. Objects may be freed in the slab while
+ * it is frozen but slab_free will then skip the usual
+ * list operations. It is up to the processor holding
+ * the slab to integrate the slab into the slab lists
+ * when the slab is no longer needed.
+ *
+ * One use of this flag is to mark slabs that are
+ * used for allocations. Then such a slab becomes a cpu
+ * slab. The cpu slab may be equipped with an additional
* lockless_freelist that allows lockless access to
* free objects in addition to the regular freelist
* that requires the slab lock.
@@ -91,6 +99,21 @@
* the fast path and disables lockless freelists.
*/
+static inline int SlabFrozen(struct page *page)
+{
+ return PageActive(page);
+}
+
+static inline void SetSlabFrozen(struct page *page)
+{
+ SetPageActive(page);
+}
+
+static inline void ClearSlabFrozen(struct page *page)
+{
+ __ClearPageActive(page);
+}
+
static inline int SlabDebug(struct page *page)
{
#ifdef CONFIG_SLUB_DEBUG
@@ -1142,11 +1165,12 @@ static void remove_partial(struct kmem_c
*
* Must hold list_lock.
*/
-static int lock_and_del_slab(struct kmem_cache_node *n, struct page *page)
+static inline int lock_and_freeze_slab(struct kmem_cache_node *n, struct page *page)
{
if (slab_trylock(page)) {
list_del(&page->lru);
n->nr_partial--;
+ SetSlabFrozen(page);
return 1;
}
return 0;
@@ -1170,7 +1194,7 @@ static struct page *get_partial_node(str
spin_lock(&n->list_lock);
list_for_each_entry(page, &n->partial, lru)
- if (lock_and_del_slab(n, page))
+ if (lock_and_freeze_slab(n, page))
goto out;
page = NULL;
out:
@@ -1249,10 +1273,11 @@ static struct page *get_partial(struct k
*
* On exit the slab lock will have been dropped.
*/
-static void putback_slab(struct kmem_cache *s, struct page *page)
+static void unfreeze_slab(struct kmem_cache *s, struct page *page)
{
struct kmem_cache_node *n = get_node(s, page_to_nid(page));
+ ClearSlabFrozen(page);
if (page->inuse) {
if (page->freelist)
@@ -1303,9 +1328,7 @@ static void deactivate_slab(struct kmem_
page->inuse--;
}
s->cpu_slab[cpu] = NULL;
- ClearPageActive(page);
-
- putback_slab(s, page);
+ unfreeze_slab(s, page);
}
static void flush_slab(struct kmem_cache *s, struct page *page, int cpu)
@@ -1396,9 +1419,7 @@ another_slab:
new_slab:
page = get_partial(s, gfpflags, node);
if (page) {
-have_slab:
s->cpu_slab[cpu] = page;
- SetPageActive(page);
goto load_freelist;
}
@@ -1428,7 +1449,9 @@ have_slab:
flush_slab(s, s->cpu_slab[cpu], cpu);
}
slab_lock(page);
- goto have_slab;
+ SetSlabFrozen(page);
+ s->cpu_slab[cpu] = page;
+ goto load_freelist;
}
return NULL;
debug:
@@ -1515,11 +1538,7 @@ checks_ok:
page->freelist = object;
page->inuse--;
- if (unlikely(PageActive(page)))
- /*
- * Cpu slabs are never on partial lists and are
- * never freed.
- */
+ if (unlikely(SlabFrozen(page)))
goto out_unlock;
if (unlikely(!page->inuse))
@@ -1551,7 +1570,7 @@ slab_empty:
debug:
if (!free_object_checks(s, page, x))
goto out_unlock;
- if (!PageActive(page) && !page->freelist)
+ if (!SlabFrozen(page) && !page->freelist)
remove_full(s, page);
if (s->flags & SLAB_STORE_USER)
set_track(s, x, TRACK_FREE, addr);
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: SLUB: Define functions for cpu slab handling instead of using PageActive
2007-05-15 3:00 SLUB: Define functions for cpu slab handling instead of using PageActive Christoph Lameter
@ 2007-05-15 4:54 ` Andrew Morton
2007-05-15 17:07 ` Christoph Lameter
0 siblings, 1 reply; 3+ messages in thread
From: Andrew Morton @ 2007-05-15 4:54 UTC (permalink / raw)
To: Christoph Lameter; +Cc: linux-mm
On Mon, 14 May 2007 20:00:07 -0700 (PDT) Christoph Lameter <clameter@sgi.com> wrote:
> Use inline functions to access the per cpu bit. Intoduce the notion of
> "freezing" a slab to make things more understandable.
>
> ...
>
> +static inline void ClearSlabFrozen(struct page *page)
> +{
> + __ClearPageActive(page);
> +}
Non-atomic.
> - ClearPageActive(page);
Atomic.
A substitution like this can lead to quite revoltingly subtle bugs and needs
lots of justfication.
I'll switch this back to the atomic version. If you're really sure about
this micro-optimisation then let's do it as a standalone patch. One which
adds a comment explaining why it is safe, and under which circumstances it
will become unsafe, etc.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: SLUB: Define functions for cpu slab handling instead of using PageActive
2007-05-15 4:54 ` Andrew Morton
@ 2007-05-15 17:07 ` Christoph Lameter
0 siblings, 0 replies; 3+ messages in thread
From: Christoph Lameter @ 2007-05-15 17:07 UTC (permalink / raw)
To: Andrew Morton; +Cc: linux-mm
On Mon, 14 May 2007, Andrew Morton wrote:
> On Mon, 14 May 2007 20:00:07 -0700 (PDT) Christoph Lameter <clameter@sgi.com> wrote:
>
> > Use inline functions to access the per cpu bit. Intoduce the notion of
> > "freezing" a slab to make things more understandable.
> >
> > ...
> >
> > +static inline void ClearSlabFrozen(struct page *page)
> > +{
> > + __ClearPageActive(page);
> > +}
>
> Non-atomic.
>
> > - ClearPageActive(page);
>
> Atomic.
>
> A substitution like this can lead to quite revoltingly subtle bugs and needs
> lots of justfication.
>
> I'll switch this back to the atomic version. If you're really sure about
> this micro-optimisation then let's do it as a standalone patch. One which
> adds a comment explaining why it is safe, and under which circumstances it
> will become unsafe, etc.
There is no need for atomics here since the flag is only modified with the
slab lock taken but there is no __SetPageActive.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2007-05-15 17:07 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-05-15 3:00 SLUB: Define functions for cpu slab handling instead of using PageActive Christoph Lameter
2007-05-15 4:54 ` Andrew Morton
2007-05-15 17:07 ` Christoph Lameter
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox