This module parameter allows user to control which numa node
the memory for mainly structures is allocated from.
Signed-off-by: Zhengyuan Liu <liuzhengyuan@xxxxxxxxxx>
---
drivers/md/raid5.c | 54 ++++++++++++++++++++++++++++++++++++------------------
1 file changed, 36 insertions(+), 18 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 547d5fa..b61417f 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -75,6 +75,11 @@ static bool devices_handle_discard_safely = false;
module_param(devices_handle_discard_safely, bool, 0644);
MODULE_PARM_DESC(devices_handle_discard_safely,
"Set to Y if all devices in each array reliably return zeroes on reads from discarded regions");
+
+static int r5_numa_node = NUMA_NO_NODE;
+module_param(r5_numa_node, int, S_IRUGO);
+MODULE_PARM_DESC(r5_numa_node, "NUMA node for raid5 memory allocations");
+
static struct workqueue_struct *raid5_wq;
static inline struct hlist_head *stripe_hash(struct r5conf *conf, sector_t sect)
@@ -484,7 +489,7 @@ static int grow_buffers(struct stripe_head *sh, gfp_t gfp)
for (i = 0; i < num; i++) {
struct page *page;
- if (!(page = alloc_page(gfp))) {
+ if (!(page = alloc_pages_node(r5_numa_node, gfp, 0))) {
return 1;
}
sh->dev[i].page = page;
@@ -2135,7 +2140,7 @@ static struct stripe_head *alloc_stripe(struct kmem_cache *sc, gfp_t gfp,
struct stripe_head *sh;
int i;
- sh = kmem_cache_zalloc(sc, gfp);
+ sh = kmem_cache_alloc_node(sc, gfp | __GFP_ZERO, r5_numa_node);
if (sh) {
spin_lock_init(&sh->stripe_lock);
spin_lock_init(&sh->batch_lock);
@@ -2154,7 +2159,7 @@ static struct stripe_head *alloc_stripe(struct kmem_cache *sc, gfp_t gfp,
}
if (raid5_has_ppl(conf)) {
- sh->ppl_page = alloc_page(gfp);
+ sh->ppl_page = alloc_pages_node(r5_numa_node, gfp, 0);
if (!sh->ppl_page) {
free_stripe(sc, sh);
sh = NULL;
@@ -2383,13 +2388,15 @@ static int resize_stripes(struct r5conf *conf, int newsize)
* is completely stalled, so now is a good time to resize
* conf->disks and the scribble region
*/
- ndisks = kzalloc(newsize * sizeof(struct disk_info), GFP_NOIO);
+ ndisks = kzalloc_node(newsize * sizeof(struct disk_info), GFP_NOIO,
+ r5_numa_node);
if (ndisks) {
for (i = 0; i < conf->pool_size; i++)
ndisks[i] = conf->disks[i];
for (i = conf->pool_size; i < newsize; i++) {
- ndisks[i].extra_page = alloc_page(GFP_NOIO);
+ ndisks[i].extra_page = alloc_pages_node(r5_numa_node,
+ GFP_NOIO, 0);
if (!ndisks[i].extra_page)
err = -ENOMEM;
}
@@ -2418,7 +2425,8 @@ static int resize_stripes(struct r5conf *conf, int newsize)
for (i=conf->raid_disks; i < newsize; i++)
if (nsh->dev[i].page == NULL) {
- struct page *p = alloc_page(GFP_NOIO);
+ struct page *p = alloc_pages_node(r5_numa_node,
+ GFP_NOIO, 0);
nsh->dev[i].page = p;
nsh->dev[i].orig_page = p;
if (!p)
@@ -3921,7 +3929,8 @@ static int handle_stripe_dirtying(struct r5conf *conf,
dev->page == dev->orig_page &&
!test_bit(R5_LOCKED, &sh->dev[sh->pd_idx].flags)) {
/* alloc page for prexor */
- struct page *p = alloc_page(GFP_NOIO);
+ struct page *p = alloc_pages_node(r5_numa_node,
+ GFP_NOIO, 0);
if (p) {
dev->orig_page = p;
@@ -6653,9 +6662,9 @@ static int alloc_thread_groups(struct r5conf *conf, int cnt,
}
*group_cnt = num_possible_nodes();
size = sizeof(struct r5worker) * cnt;
- workers = kzalloc(size * *group_cnt, GFP_NOIO);
- *worker_groups = kzalloc(sizeof(struct r5worker_group) *
- *group_cnt, GFP_NOIO);
+ workers = kzalloc_node(size * *group_cnt, GFP_NOIO, r5_numa_node);
+ *worker_groups = kzalloc_node(sizeof(struct r5worker_group) *
+ *group_cnt, GFP_NOIO, r5_numa_node);
if (!*worker_groups || !workers) {
kfree(workers);
kfree(*worker_groups);
@@ -6720,7 +6729,8 @@ static void free_scratch_buffer(struct r5conf *conf, struct raid5_percpu *percpu
static int alloc_scratch_buffer(struct r5conf *conf, struct raid5_percpu *percpu)
{
if (conf->level == 6 && !percpu->spare_page)
- percpu->spare_page = alloc_page(GFP_KERNEL);
+ percpu->spare_page = alloc_pages_node(r5_numa_node,
+ GFP_KERNEL, 0);
if (!percpu->scribble)
percpu->scribble = scribble_alloc(max(conf->raid_disks,
conf->previous_raid_disks),
@@ -6880,13 +6890,13 @@ static struct r5conf *setup_conf(struct mddev *mddev)
return ERR_PTR(-EINVAL);
}
- conf = kzalloc(sizeof(struct r5conf), GFP_KERNEL);
+ conf = kzalloc_node(sizeof(struct r5conf), GFP_KERNEL, r5_numa_node);
if (conf == NULL)
goto abort;
INIT_LIST_HEAD(&conf->free_list);
INIT_LIST_HEAD(&conf->pending_list);
- conf->pending_data = kzalloc(sizeof(struct r5pending_data) *
- PENDING_IO_MAX, GFP_KERNEL);
+ conf->pending_data = kzalloc_node(sizeof(struct r5pending_data) *
+ PENDING_IO_MAX, GFP_KERNEL, r5_numa_node);
if (!conf->pending_data)
goto abort;
for (i = 0; i < PENDING_IO_MAX; i++)
@@ -6935,14 +6945,15 @@ static struct r5conf *setup_conf(struct mddev *mddev)
conf->previous_raid_disks = mddev->raid_disks - mddev->delta_disks;
max_disks = max(conf->raid_disks, conf->previous_raid_disks);
- conf->disks = kzalloc(max_disks * sizeof(struct disk_info),
- GFP_KERNEL);
+ conf->disks = kzalloc_node(max_disks * sizeof(struct disk_info),
+ GFP_KERNEL, r5_numa_node);
if (!conf->disks)
goto abort;
for (i = 0; i < max_disks; i++) {
- conf->disks[i].extra_page = alloc_page(GFP_KERNEL);
+ conf->disks[i].extra_page = alloc_pages_node(r5_numa_node,
+ GFP_KERNEL, 0);
if (!conf->disks[i].extra_page)
goto abort;
}
@@ -6952,7 +6963,8 @@ static struct r5conf *setup_conf(struct mddev *mddev)
goto abort;
conf->mddev = mddev;
- if ((conf->stripe_hashtbl = kzalloc(PAGE_SIZE, GFP_KERNEL)) == NULL)
+ if ((conf->stripe_hashtbl = kzalloc_node(PAGE_SIZE, GFP_KERNEL, r5_numa_node))
+ == NULL)
goto abort;
/* We init hash_locks[0] separately to that it can be used
@@ -8445,6 +8457,12 @@ static int __init raid5_init(void)
destroy_workqueue(raid5_wq);
return ret;
}
+
+ if (r5_numa_node < NUMA_NO_NODE)
+ r5_numa_node = NUMA_NO_NODE;
+ else if (r5_numa_node > (num_online_nodes() - 1))
+ r5_numa_node = num_online_nodes() - 1;
+
register_md_personality(&raid6_personality);
register_md_personality(&raid5_personality);
register_md_personality(&raid4_personality);
--
2.7.4
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
