From: Philipp Reisner <philipp.reisner@linbit.com>
To: linux-kernel@vger.kernel.org, Jens Axboe <axboe@kernel.dk>
Cc: drbd-dev@lists.linbit.com
Subject: [PATCH 03/18] drbd: add page pool to be used for meta data IO
Date: Thu, 1 Sep 2011 14:48:50 +0200 [thread overview]
Message-ID: <1314881345-6420-4-git-send-email-philipp.reisner@linbit.com> (raw)
In-Reply-To: <1314881345-6420-1-git-send-email-philipp.reisner@linbit.com>
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
drivers/block/drbd/drbd_int.h | 23 ++++++++++++++++++++++-
drivers/block/drbd/drbd_main.c | 9 +++++++++
2 files changed, 31 insertions(+), 1 deletions(-)
diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index e3f542c..4debab7 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -1463,11 +1463,32 @@ extern struct kmem_cache *drbd_al_ext_cache; /* activity log extents */
extern mempool_t *drbd_request_mempool;
extern mempool_t *drbd_ee_mempool;
-extern struct page *drbd_pp_pool; /* drbd's page pool */
+/* drbd's page pool, used to buffer data received from the peer,
+ * or data requested by the peer.
+ *
+ * This does not have an emergency reserve.
+ *
+ * When allocating from this pool, it first takes pages from the pool.
+ * Only if the pool is depleted will try to allocate from the system.
+ *
+ * The assumption is that pages taken from this pool will be processed,
+ * and given back, "quickly", and then can be recycled, so we can avoid
+ * frequent calls to alloc_page(), and still will be able to make progress even
+ * under memory pressure.
+ */
+extern struct page *drbd_pp_pool;
extern spinlock_t drbd_pp_lock;
extern int drbd_pp_vacant;
extern wait_queue_head_t drbd_pp_wait;
+/* We also need a standard (emergency-reserve backed) page pool
+ * for meta data IO (activity log, bitmap).
+ * We can keep it global, as long as it is used as "N pages at a time".
+ * 128 should be plenty, currently we probably can get away with as few as 1.
+ */
+#define DRBD_MIN_POOL_PAGES 128
+extern mempool_t *drbd_md_io_page_pool;
+
extern rwlock_t global_state_lock;
extern int conn_lowest_minor(struct drbd_tconn *tconn);
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 6d761cb..cb1636c 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -129,6 +129,7 @@ struct kmem_cache *drbd_bm_ext_cache; /* bitmap extents */
struct kmem_cache *drbd_al_ext_cache; /* activity log extents */
mempool_t *drbd_request_mempool;
mempool_t *drbd_ee_mempool;
+mempool_t *drbd_md_io_page_pool;
/* I do not use a standard mempool, because:
1) I want to hand out the pre-allocated objects first.
@@ -1952,6 +1953,8 @@ static void drbd_destroy_mempools(void)
/* D_ASSERT(atomic_read(&drbd_pp_vacant)==0); */
+ if (drbd_md_io_page_pool)
+ mempool_destroy(drbd_md_io_page_pool);
if (drbd_ee_mempool)
mempool_destroy(drbd_ee_mempool);
if (drbd_request_mempool)
@@ -1965,6 +1968,7 @@ static void drbd_destroy_mempools(void)
if (drbd_al_ext_cache)
kmem_cache_destroy(drbd_al_ext_cache);
+ drbd_md_io_page_pool = NULL;
drbd_ee_mempool = NULL;
drbd_request_mempool = NULL;
drbd_ee_cache = NULL;
@@ -1988,6 +1992,7 @@ static int drbd_create_mempools(void)
drbd_bm_ext_cache = NULL;
drbd_al_ext_cache = NULL;
drbd_pp_pool = NULL;
+ drbd_md_io_page_pool = NULL;
/* caches */
drbd_request_cache = kmem_cache_create(
@@ -2011,6 +2016,10 @@ static int drbd_create_mempools(void)
goto Enomem;
/* mempools */
+ drbd_md_io_page_pool = mempool_create_page_pool(DRBD_MIN_POOL_PAGES, 0);
+ if (drbd_md_io_page_pool == NULL)
+ goto Enomem;
+
drbd_request_mempool = mempool_create(number,
mempool_alloc_slab, mempool_free_slab, drbd_request_cache);
if (drbd_request_mempool == NULL)
--
1.7.4.1
next prev parent reply other threads:[~2011-09-01 12:49 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-09-01 12:48 [RFC 00/18] drbd: part 4 of adding multiple volume support to drbd Philipp Reisner
2011-09-01 12:48 ` [PATCH 01/18] drbd: default to detach on-io-error Philipp Reisner
2011-09-01 12:48 ` [PATCH 02/18] drbd: only wakeup if something changed in update_peer_seq Philipp Reisner
2011-09-01 12:48 ` Philipp Reisner [this message]
2011-09-01 12:48 ` [PATCH 04/18] drbd: use the newly introduced page pool for bitmap IO Philipp Reisner
2011-09-01 12:48 ` [PATCH 05/18] drbd: introduce a bio_set to allocate housekeeping bios from Philipp Reisner
2011-09-01 12:48 ` [PATCH 06/18] drbd: fix drbd_delete_device: remove vnr from volumes; idr_remove(); synchronize_rcu(); before cleanup Philipp Reisner
2011-09-01 12:48 ` [PATCH 07/18] drbd: get rid of drbd_bcast_ee, it is of no use anymore Philipp Reisner
2011-09-01 12:48 ` [PATCH 08/18] drbd: prepare the transition from connector to genetlink Philipp Reisner
2011-09-01 12:48 ` [PATCH 09/18] drbd: switch configuration interface " Philipp Reisner
2011-09-01 12:48 ` [PATCH 10/18] drbd: allow holes in minor and volume id allocation Philipp Reisner
2011-09-01 12:48 ` [PATCH 11/18] drbd: remove now unused connector related files Philipp Reisner
2011-09-01 12:48 ` [PATCH 12/18] drbd: drbd_adm_get_status needs to show some more detail Philipp Reisner
2011-09-01 12:49 ` [PATCH 13/18] drbd: simplify conn_all_vols_unconf, make it bool Philipp Reisner
2011-09-01 12:49 ` [PATCH 14/18] drbd: Allow a Diskless Secondary volume to be removed Philipp Reisner
2011-09-01 12:49 ` [PATCH 15/18] drbd: new-connection and new-minor succeed, if the object already exists Philipp Reisner
2011-09-01 12:49 ` [PATCH 16/18] drbd: bail out if a config requrest is over-determined, and not matching Philipp Reisner
2011-09-01 12:49 ` [PATCH 17/18] drbd: add forgotten spin_unlock Philipp Reisner
2011-09-01 12:49 ` [PATCH 18/18] drbd: introduce in-kernel "down" command Philipp Reisner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1314881345-6420-4-git-send-email-philipp.reisner@linbit.com \
--to=philipp.reisner@linbit.com \
--cc=axboe@kernel.dk \
--cc=drbd-dev@lists.linbit.com \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®