Some bits for zone reclaim exists in 2.6.15 but they are not usable.
This patch fixes them up, removes unused code and makes zone reclaim usable.
Zone reclaim allows the reclaiming of pages from a zone if the number of free
pages falls below the watermarks even if other zones still have enough pages
available. Zone reclaim is of particular importance for NUMA machines. It can
be more beneficial to reclaim a page than taking the performance penalties
that come with allocating a page on a remote zone.
Zone reclaim is enabled if the maximum distance to another node is higher
than RECLAIM_DISTANCE, which may be defined by an arch. By default
RECLAIM_DISTANCE is 20. 20 is the distance to another node in the
same component (enclosure or motherboard) on IA64. The meaning of the
NUMA distance information seems to vary by arch.
If zone reclaim is not successful then no further reclaim attempts
will occur for a certain time period (ZONE_RECLAIM_INTERVAL).
This patch was discussed before. See
http://marc.theaimsgroup.com/?l=linux-kernel&m=113519961504207&w=2
http://marc.theaimsgroup.com/?l=linux-kernel&m=113408418232531&w=2
http://marc.theaimsgroup.com/?l=linux-kernel&m=113389027420032&w=2
http://marc.theaimsgroup.com/?l=linux-kernel&m=113380938612205&w=2
Signed-off-by: Christoph Lameter <[email protected]>
Index: linux-2.6.15/mm/page_alloc.c
===================================================================
--- linux-2.6.15.orig/mm/page_alloc.c 2006-01-02 19:21:10.000000000 -0800
+++ linux-2.6.15/mm/page_alloc.c 2006-01-03 14:04:59.000000000 -0800
@@ -842,7 +842,9 @@ get_page_from_freelist(gfp_t gfp_mask, u
mark = (*z)->pages_high;
if (!zone_watermark_ok(*z, order, mark,
classzone_idx, alloc_flags))
- continue;
+ if (!zone_reclaim_mode ||
+ !zone_reclaim(*z, gfp_mask, order))
+ continue;
}
page = buffered_rmqueue(*z, order, gfp_mask);
@@ -1559,13 +1561,22 @@ static void __init build_zonelists(pg_da
prev_node = local_node;
nodes_clear(used_mask);
while ((node = find_next_best_node(local_node, &used_mask)) >= 0) {
+ int distance = node_distance(local_node, node);
+
+ /*
+ * If another node is sufficiently far away then it is better
+ * to reclaim pages in a zone before going off node.
+ */
+ if (distance > RECLAIM_DISTANCE)
+ zone_reclaim_mode = 1;
+
/*
* We don't want to pressure a particular node.
* So adding penalty to the first node in same
* distance group to make it round-robin.
*/
- if (node_distance(local_node, node) !=
- node_distance(local_node, prev_node))
+
+ if (distance != node_distance(local_node, prev_node))
node_load[node] += load;
prev_node = node;
load--;
Index: linux-2.6.15/include/linux/swap.h
===================================================================
--- linux-2.6.15.orig/include/linux/swap.h 2006-01-02 19:21:10.000000000 -0800
+++ linux-2.6.15/include/linux/swap.h 2006-01-03 14:07:39.000000000 -0800
@@ -172,7 +172,17 @@ extern void swap_setup(void);
/* linux/mm/vmscan.c */
extern int try_to_free_pages(struct zone **, gfp_t);
+#ifdef CONFIG_NUMA
+extern int zone_reclaim_mode;
extern int zone_reclaim(struct zone *, gfp_t, unsigned int);
+#else
+#define zone_reclaim_mode 0
+static inline int zone_reclaim(struct zone *z, gfp_t mask,
+ unsigned int order)
+{
+ return 0;
+}
+#endif
extern int shrink_all_memory(int);
extern int vm_swappiness;
Index: linux-2.6.15/include/linux/topology.h
===================================================================
--- linux-2.6.15.orig/include/linux/topology.h 2006-01-02 19:21:10.000000000 -0800
+++ linux-2.6.15/include/linux/topology.h 2006-01-03 14:04:59.000000000 -0800
@@ -56,6 +56,9 @@
#define REMOTE_DISTANCE 20
#define node_distance(from,to) ((from) == (to) ? LOCAL_DISTANCE : REMOTE_DISTANCE)
#endif
+#ifndef RECLAIM_DISTANCE
+#define RECLAIM_DISTANCE 20
+#endif
#ifndef PENALTY_FOR_NODE_WITH_CPUS
#define PENALTY_FOR_NODE_WITH_CPUS (1)
#endif
Index: linux-2.6.15/mm/vmscan.c
===================================================================
--- linux-2.6.15.orig/mm/vmscan.c 2006-01-02 19:21:10.000000000 -0800
+++ linux-2.6.15/mm/vmscan.c 2006-01-03 14:36:21.000000000 -0800
@@ -1354,6 +1354,19 @@ static int __init kswapd_init(void)
module_init(kswapd_init)
+#ifdef CONFIG_NUMA
+/*
+ * Zone reclaim mode
+ *
+ * If non-zero call zone_reclaim when the number of free pages falls below
+ * the watermarks.
+ */
+int zone_reclaim_mode __read_mostly;
+
+/*
+ * Mininum time between zone reclaim scans
+ */
+#define ZONE_RECLAIM_INTERVAL HZ/2
/*
* Try to free up some pages from this zone through reclaim.
@@ -1362,13 +1375,18 @@ int zone_reclaim(struct zone *zone, gfp_
{
struct scan_control sc;
int nr_pages = 1 << order;
- int total_reclaimed = 0;
+ struct task_struct *p = current;
+ struct reclaim_state reclaim_state;
- /* The reclaim may sleep, so don't do it if sleep isn't allowed */
- if (!(gfp_mask & __GFP_WAIT))
- return 0;
- if (zone->all_unreclaimable)
- return 0;
+ if (!(gfp_mask & __GFP_WAIT) ||
+ zone->zone_pgdat->node_id != numa_node_id() ||
+ zone->all_unreclaimable ||
+ atomic_read(&zone->reclaim_in_progress) > 0)
+ return 0;
+
+ if (time_before(jiffies,
+ zone->last_unsuccessful_zone_reclaim + ZONE_RECLAIM_INTERVAL))
+ return 0;
sc.gfp_mask = gfp_mask;
sc.may_writepage = 0;
@@ -1385,44 +1403,17 @@ int zone_reclaim(struct zone *zone, gfp_
else
sc.swap_cluster_max = SWAP_CLUSTER_MAX;
- /* Don't reclaim the zone if there are other reclaimers active */
- if (atomic_read(&zone->reclaim_in_progress) > 0)
- goto out;
-
+ cond_resched();
+ p->flags |= PF_MEMALLOC;
+ reclaim_state.reclaimed_slab = 0;
+ p->reclaim_state = &reclaim_state;
shrink_zone(zone, &sc);
- total_reclaimed = sc.nr_reclaimed;
-
- out:
- return total_reclaimed;
-}
-
-asmlinkage long sys_set_zone_reclaim(unsigned int node, unsigned int zone,
- unsigned int state)
-{
- struct zone *z;
- int i;
-
- if (!capable(CAP_SYS_ADMIN))
- return -EACCES;
-
- if (node >= MAX_NUMNODES || !node_online(node))
- return -EINVAL;
-
- /* This will break if we ever add more zones */
- if (!(zone & (1<<ZONE_DMA|1<<ZONE_NORMAL|1<<ZONE_HIGHMEM)))
- return -EINVAL;
-
- for (i = 0; i < MAX_NR_ZONES; i++) {
- if (!(zone & 1<<i))
- continue;
-
- z = &NODE_DATA(node)->node_zones[i];
+ p->reclaim_state = NULL;
+ current->flags &= ~PF_MEMALLOC;
- if (state)
- z->reclaim_pages = 1;
- else
- z->reclaim_pages = 0;
- }
+ if (sc.nr_reclaimed == 0)
+ zone->last_unsuccessful_zone_reclaim = jiffies;
- return 0;
+ return sc.nr_reclaimed > nr_pages;
}
+#endif
Index: linux-2.6.15/include/linux/mmzone.h
===================================================================
--- linux-2.6.15.orig/include/linux/mmzone.h 2006-01-02 19:21:10.000000000 -0800
+++ linux-2.6.15/include/linux/mmzone.h 2006-01-03 14:19:48.000000000 -0800
@@ -150,14 +150,11 @@ struct zone {
unsigned long pages_scanned; /* since last reclaim */
int all_unreclaimable; /* All pages pinned */
- /*
- * Does the allocator try to reclaim pages from the zone as soon
- * as it fails a watermark_ok() in __alloc_pages?
- */
- int reclaim_pages;
/* A count of how many reclaimers are scanning this zone */
atomic_t reclaim_in_progress;
+ unsigned long last_unsuccessful_zone_reclaim;
+
/*
* prev_priority holds the scanning priority for this zone. It is
* defined as the scanning priority at which we achieved our reclaim
Index: linux-2.6.15/include/asm-i386/unistd.h
===================================================================
--- linux-2.6.15.orig/include/asm-i386/unistd.h 2006-01-02 19:21:10.000000000 -0800
+++ linux-2.6.15/include/asm-i386/unistd.h 2006-01-03 14:19:13.000000000 -0800
@@ -256,7 +256,7 @@
#define __NR_io_submit 248
#define __NR_io_cancel 249
#define __NR_fadvise64 250
-#define __NR_set_zone_reclaim 251
+/* 251 is available for reuse (was briefly sys_set_zone_reclaim) */
#define __NR_exit_group 252
#define __NR_lookup_dcookie 253
#define __NR_epoll_create 254
Index: linux-2.6.15/include/asm-ia64/unistd.h
===================================================================
--- linux-2.6.15.orig/include/asm-ia64/unistd.h 2006-01-02 19:21:10.000000000 -0800
+++ linux-2.6.15/include/asm-ia64/unistd.h 2006-01-03 14:19:13.000000000 -0800
@@ -265,7 +265,7 @@
#define __NR_keyctl 1273
#define __NR_ioprio_set 1274
#define __NR_ioprio_get 1275
-#define __NR_set_zone_reclaim 1276
+/* 1276 is available for reuse (was briefly sys_set_zone_reclaim) */
#define __NR_inotify_init 1277
#define __NR_inotify_add_watch 1278
#define __NR_inotify_rm_watch 1279
Christoph Lameter <[email protected]> wrote:
>
> Some bits for zone reclaim exists in 2.6.15 but they are not usable.
> This patch fixes them up, removes unused code and makes zone reclaim usable.
>
You know that there are over 100 mm/ patches in -mm. If Linus applies this
patch, it will cause extensive wreckage. And this patch doesn't vaguely
apply to the mm/ patches which I have queued. So it's basically useless.
Please try to play along.
yes, it's awkward that there's such a large backlog in that area. We just
need to be patient and extra careful.
On Tue, 3 Jan 2006, Andrew Morton wrote:
> Christoph Lameter <[email protected]> wrote:
> >
> > Some bits for zone reclaim exists in 2.6.15 but they are not usable.
> > This patch fixes them up, removes unused code and makes zone reclaim usable.
> >
>
> You know that there are over 100 mm/ patches in -mm. If Linus applies this
> patch, it will cause extensive wreckage. And this patch doesn't vaguely
> apply to the mm/ patches which I have queued. So it's basically useless.
>
> Please try to play along.
Well, this is one case where there is crap in Linus tree that needs to be
fixed. Its an urgent issue. And the existing patches in mm do not fix
this issue but remove the code altogether. When I asked you to remove
these patches, you got mad at me for some reason.
> yes, it's awkward that there's such a large backlog in that area. We just
> need to be patient and extra careful.
So you are saying we need to remove this feature and then add it back in
later?
This means another 2 and 3 release cycles with this severe unfixed
problem?
Christoph Lameter <[email protected]> wrote:
>
> On Tue, 3 Jan 2006, Andrew Morton wrote:
>
> > Christoph Lameter <[email protected]> wrote:
> > >
> > > Some bits for zone reclaim exists in 2.6.15 but they are not usable.
> > > This patch fixes them up, removes unused code and makes zone reclaim usable.
> > >
> >
> > You know that there are over 100 mm/ patches in -mm. If Linus applies this
> > patch, it will cause extensive wreckage. And this patch doesn't vaguely
> > apply to the mm/ patches which I have queued. So it's basically useless.
> >
> > Please try to play along.
>
> Well, this is one case where there is crap in Linus tree that needs to be
> fixed. Its an urgent issue. And the existing patches in mm do not fix
> this issue but remove the code altogether. When I asked you to remove
> these patches, you got mad at me for some reason.
The code in Linus's tree is dead and is unused.
> > yes, it's awkward that there's such a large backlog in that area. We just
> > need to be patient and extra careful.
>
> So you are saying we need to remove this feature and then add it back in
> later?
Yes. It's cleaner to remove all the dead stuff and to then reintroduce the
new feature as a clean, standalone thing. So we can see its effects ore
clearly.
Also, what happens if for some reason we later want to revert your patch?
Such a reversion would reintroduce all the old dead stuff!
Patches should each do a single thing.
> This means another 2 and 3 release cycles with this severe unfixed
> problem?
There's plenty of time to get this into 2.6.16.
Your changelog didn't describe this as a "severe" problem. Things have
been like this for quite some time, haven't they?
And we shouldn't be merging new, untested things ahead of work which has
had considerable amounts of testing.
On Wednesday 04 January 2006 01:43, Andrew Morton wrote:
> Your changelog didn't describe this as a "severe" problem. Things have
> been like this for quite some time, haven't they?
I get at least regular complaints on x86-64 about it. It's not really a FAQ
yet, but at least an issue multiple people run into.
-Andi
On Tue, 3 Jan 2006, Andrew Morton wrote:
> There's plenty of time to get this into 2.6.16.
Ok. Thanks.