RMW of r5c write back cache uses an extra page to store old data for
prexor. handle_stripe_dirtying() allocates this page by calling
alloc_page(). However, alloc_page() may fail.
To handle alloc_page() failures, this patch adds a small mempool
in r5l_log. When alloc_page fails, the stripe is added to a waiting
list. Then, these stripes get pages from the mempool (from work queue).
Signed-off-by: Song Liu <songliubraving@xxxxxx>
---
drivers/md/raid5-cache.c | 100 ++++++++++++++++++++++++++++++++++++++++++++++-
drivers/md/raid5.c | 34 +++++++++++-----
drivers/md/raid5.h | 6 +++
3 files changed, 130 insertions(+), 10 deletions(-)
diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c
index 8cb79fc..ce7f114 100644
--- a/drivers/md/raid5-cache.c
+++ b/drivers/md/raid5-cache.c
@@ -51,6 +51,8 @@
*/
#define R5L_POOL_SIZE 4
+#define R5C_EXTRA_PAGE_POOL_SIZE 2
+
/*
* r5c journal modes of the array: write-back or write-through.
* write-through mode has identical behavior as existing log only
@@ -162,6 +164,16 @@ struct r5l_log {
/* to submit async io_units, to fulfill ordering of flush */
struct work_struct deferred_io_work;
+
+ /* to handle alloc_page failures in handle_stripe_dirtying() */
+ /* mempool for up to R5C_EXTRA_PAGE_POOL_SIZE stripes */
+ mempool_t *extra_page_pool;
+ /* list of stripes waiting to use the page pool */
+ struct list_head extra_page_pool_list;
+ /* lock for extra_page_pool_list */
+ spinlock_t extra_page_pool_lock;
+ /* work that allocates pages from extra_page_pool */
+ struct work_struct extra_page_pool_work;
};
/*
@@ -252,6 +264,69 @@ static void __r5l_set_io_unit_state(struct r5l_io_unit *io,
io->state = state;
}
+void r5c_add_to_extra_page_pool_list(struct stripe_head *sh,
+ struct stripe_head_state *s)
+{
+ struct r5conf *conf = sh->raid_conf;
+ struct r5l_log *log = conf->log;
+ int i;
+ struct page *p;
+
+ BUG_ON(!log);
+
+ /* free any extra orig_page from alloc_page() */
+ for (i = sh->disks; i--; )
+ if (sh->dev[i].page != sh->dev[i].orig_page) {
+ p = sh->dev[i].orig_page;
+ sh->dev[i].orig_page = sh->dev[i].page;
+ put_page(p);
+ }
+
+ WARN_ON(!list_empty(&sh->log_list));
+
+ atomic_inc(&sh->count);
+ set_bit(STRIPE_R5C_EXTRA_PAGE, &sh->state);
+
+ spin_lock(&log->extra_page_pool_lock);
+ list_add_tail(&sh->log_list, &log->extra_page_pool_list);
+ spin_unlock(&log->extra_page_pool_lock);
+
+ s->waiting_extra_page = 1;
+ schedule_work(&log->extra_page_pool_work);
+}
+
+static void r5c_run_extra_page_pool_list(struct work_struct *work)
+{
+ struct r5l_log *log = container_of(work, struct r5l_log,
+ extra_page_pool_work);
+ struct stripe_head *sh;
+ struct r5dev *dev;
+ int i;
+
+ while (1) {
+ spin_lock(&log->extra_page_pool_lock);
+ if (list_empty(&log->extra_page_pool_list)) {
+ spin_unlock(&log->extra_page_pool_lock);
+ break;
+ }
+ sh = list_first_entry(&log->extra_page_pool_list,
+ struct stripe_head, log_list);
+ list_del_init(&sh->log_list);
+ spin_unlock(&log->extra_page_pool_lock);
+
+ for (i = sh->disks; i--; ) {
+ dev = &sh->dev[i];
+ BUG_ON(dev->page != dev->orig_page);
+
+ if (test_bit(R5_InJournal, &dev->flags))
+ dev->orig_page = mempool_alloc(
+ log->extra_page_pool, GFP_NOIO);
+ }
+ set_bit(STRIPE_HANDLE, &sh->state);
+ raid5_release_stripe(sh);
+ }
+}
+
static void
r5c_return_dev_pending_writes(struct r5conf *conf, struct r5dev *dev,
struct bio_list *return_bi)
@@ -2334,14 +2409,25 @@ int r5c_try_caching_write(struct r5conf *conf,
*/
void r5c_release_extra_page(struct stripe_head *sh)
{
+ struct r5conf *conf = sh->raid_conf;
+ struct r5l_log *log = conf->log;
int i;
+ bool using_extra_page_pool;;
+
+ BUG_ON(!log);
+
+ using_extra_page_pool = test_and_clear_bit(
+ STRIPE_R5C_EXTRA_PAGE, &sh->state);
for (i = sh->disks; i--; )
if (sh->dev[i].page != sh->dev[i].orig_page) {
struct page *p = sh->dev[i].orig_page;
sh->dev[i].orig_page = sh->dev[i].page;
- put_page(p);
+ if (using_extra_page_pool)
+ mempool_free(p, log->extra_page_pool);
+ else
+ put_page(p);
}
}
@@ -2581,6 +2667,15 @@ int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev)
if (!log->meta_pool)
goto out_mempool;
+ log->extra_page_pool = mempool_create_page_pool(
+ R5C_EXTRA_PAGE_POOL_SIZE *
+ (conf->raid_disks - conf->max_degraded), 0);
+ if (!log->extra_page_pool)
+ goto extra_page_pool;
+ INIT_LIST_HEAD(&log->extra_page_pool_list);
+ spin_lock_init(&log->extra_page_pool_lock);
+ INIT_WORK(&log->extra_page_pool_work, r5c_run_extra_page_pool_list);
+
log->reclaim_thread = md_register_thread(r5l_reclaim_thread,
log->rdev->mddev, "reclaim");
if (!log->reclaim_thread)
@@ -2611,6 +2706,8 @@ int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev)
error:
md_unregister_thread(&log->reclaim_thread);
reclaim_thread:
+ mempool_destroy(log->extra_page_pool);
+extra_page_pool:
mempool_destroy(log->meta_pool);
out_mempool:
bioset_free(log->bs);
@@ -2626,6 +2723,7 @@ int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev)
void r5l_exit_log(struct r5l_log *log)
{
md_unregister_thread(&log->reclaim_thread);
+ mempool_destroy(log->extra_page_pool);
mempool_destroy(log->meta_pool);
bioset_free(log->bs);
mempool_destroy(log->io_pool);
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index dbab8c7..5870ca9 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -876,6 +876,8 @@ static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
if (!test_bit(STRIPE_R5C_CACHING, &sh->state)) {
/* writing out phase */
+ if (s->waiting_extra_page)
+ return;
if (r5l_write_stripe(conf->log, sh) == 0)
return;
} else { /* caching phase */
@@ -2007,6 +2009,7 @@ static struct stripe_head *alloc_stripe(struct kmem_cache *sc, gfp_t gfp,
INIT_LIST_HEAD(&sh->batch_list);
INIT_LIST_HEAD(&sh->lru);
INIT_LIST_HEAD(&sh->r5c);
+ INIT_LIST_HEAD(&sh->log_list);
atomic_set(&sh->count, 1);
sh->log_start = MaxSector;
for (i = 0; i < disks; i++) {
@@ -2897,6 +2900,7 @@ schedule_reconstruction(struct stripe_head *sh, struct stripe_head_state *s,
s->locked++;
}
}
+
/* if we are not expanding this is a proper write request, and
* there will be bios with new data to be drained into the
* stripe cache
@@ -3580,10 +3584,10 @@ static void handle_stripe_clean_event(struct r5conf *conf,
break_stripe_batch_list(head_sh, STRIPE_EXPAND_SYNC_FLAGS);
}
-static void handle_stripe_dirtying(struct r5conf *conf,
- struct stripe_head *sh,
- struct stripe_head_state *s,
- int disks)
+static int handle_stripe_dirtying(struct r5conf *conf,
+ struct stripe_head *sh,
+ struct stripe_head_state *s,
+ int disks)
{
int rmw = 0, rcw = 0, i;
sector_t recovery_cp = conf->mddev->recovery_cp;
@@ -3649,12 +3653,19 @@ static void 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 */
- dev->orig_page = alloc_page(GFP_NOIO);
+ struct page *p;
- /* will handle failure in a later patch*/
- BUG_ON(!dev->orig_page);
+ p = alloc_page(GFP_NOIO);
+ if (!p) {
+ r5c_add_to_extra_page_pool_list(sh, s);
+ return -EAGAIN;
+ }
+ dev->orig_page = p;
}
+ }
+ for (i = disks; i--; ) {
+ struct r5dev *dev = &sh->dev[i];
if ((dev->towrite ||
i == sh->pd_idx || i == sh->qd_idx ||
test_bit(R5_InJournal, &dev->flags)) &&
@@ -3730,6 +3741,7 @@ static void handle_stripe_dirtying(struct r5conf *conf,
(s->locked == 0 && (rcw == 0 || rmw == 0) &&
!test_bit(STRIPE_BIT_DELAY, &sh->state)))
schedule_reconstruction(sh, s, rcw == 0, 0);
+ return 0;
}
static void handle_parity_checks5(struct r5conf *conf, struct stripe_head *sh,
@@ -4545,8 +4557,12 @@ static void handle_stripe(struct stripe_head *sh)
if (ret == -EAGAIN ||
/* stripe under reclaim: !caching && injournal */
(!test_bit(STRIPE_R5C_CACHING, &sh->state) &&
- s.injournal > 0))
- handle_stripe_dirtying(conf, sh, &s, disks);
+ s.injournal > 0)) {
+ ret = handle_stripe_dirtying(conf, sh, &s,
+ disks);
+ if (ret == -EAGAIN)
+ goto finish;
+ }
}
}
diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
index d13fe45..e0efd46 100644
--- a/drivers/md/raid5.h
+++ b/drivers/md/raid5.h
@@ -276,6 +276,7 @@ struct stripe_head_state {
struct md_rdev *blocked_rdev;
int handle_bad_blocks;
int log_failed;
+ int waiting_extra_page;
};
/* Flags for struct r5dev.flags */
@@ -377,6 +378,9 @@ enum {
* in conf->r5c_full_stripe_list)
*/
STRIPE_R5C_PREFLUSH, /* need to flush journal device */
+ STRIPE_R5C_EXTRA_PAGE, /* extra orig_page of this stripe is allocated
+ * from r5l_log->extra_page_pool
+ */
};
#define STRIPE_EXPAND_SYNC_FLAGS \
@@ -774,5 +778,7 @@ extern void r5c_make_stripe_write_out(struct stripe_head *sh);
extern void r5c_flush_cache(struct r5conf *conf, int num);
extern void r5c_check_stripe_cache_usage(struct r5conf *conf);
extern void r5c_check_cached_full_stripe(struct r5conf *conf);
+extern void r5c_add_to_extra_page_pool_list(struct stripe_head *sh,
+ struct stripe_head_state *s);
extern struct md_sysfs_entry r5c_journal_mode;
#endif
--
2.9.3
--
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
