rdma: pre-register iobuf_pool with rdma devices.

        Back port pf : http://review.gluster.org/9506

registring buffers with rdma device is a time consuming
operation. So performing registration in code path will
decrease the performance.
Using a pre registered memory will give a bettor performance,
ie, register iobuf_pool during rdma initialization. For
dynamically created arena, we can register with all the
device.

Change-Id: Ic79183e2efd014c43faf5911fdb6d5cfbcee64ca
BUG: 1202212
Signed-off-by: Mohammed Rafi KC <rkavunga@redhat.com>
Reviewed-on: http://review.gluster.org/9506
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Raghavendra G <rgowdapp@redhat.com>
Tested-by: Raghavendra G <rgowdapp@redhat.com>
Reviewed-on: http://review.gluster.org/9889
Reviewed-by: Raghavendra Bhat <raghavendra@redhat.com>

3 files changed, 44 insertions, 7 deletions

diff --git a/libglusterfs/src/iobuf.c b/libglusterfs/src/iobuf.c
index 82ffe2dd8fd..f8f1860889b 100644
--- a/libglusterfs/src/iobuf.c
+++ b/libglusterfs/src/iobuf.c
@@ -50,6 +50,7 @@ gf_iobuf_get_arena_index (size_t page_size)
         return i;
 }
 
+
 size_t
 gf_iobuf_get_pagesize (size_t page_size)
 {
@@ -138,10 +139,15 @@ out:
 
 
 void
-__iobuf_arena_destroy (struct iobuf_arena *iobuf_arena)
+__iobuf_arena_destroy (struct iobuf_pool *iobuf_pool,
+                       struct iobuf_arena *iobuf_arena)
 {
         GF_VALIDATE_OR_GOTO ("iobuf", iobuf_arena, out);
 
+        if (iobuf_pool->rdma_deregistration)
+                iobuf_pool->rdma_deregistration (iobuf_pool->mr_list,
+                                                 iobuf_arena);
+
         __iobuf_arena_destroy_iobufs (iobuf_arena);
 
         if (iobuf_arena->mem_base
@@ -169,6 +175,7 @@ __iobuf_arena_alloc (struct iobuf_pool *iobuf_pool, size_t page_size,
                 goto err;
 
         INIT_LIST_HEAD (&iobuf_arena->list);
+        INIT_LIST_HEAD (&iobuf_arena->all_list);
         INIT_LIST_HEAD (&iobuf_arena->active.list);
         INIT_LIST_HEAD (&iobuf_arena->passive.list);
         iobuf_arena->iobuf_pool = iobuf_pool;
@@ -188,6 +195,13 @@ __iobuf_arena_alloc (struct iobuf_pool *iobuf_pool, size_t page_size,
                 goto err;
         }
 
+        if (iobuf_pool->rdma_registration) {
+                iobuf_pool->rdma_registration (iobuf_pool->device,
+                                               iobuf_arena);
+        }
+
+        list_add_tail (&iobuf_arena->all_list, &iobuf_pool->all_arenas);
+
         __iobuf_arena_init_iobufs (iobuf_arena);
         if (!iobuf_arena->iobufs) {
                 gf_log (THIS->name, GF_LOG_ERROR, "init failed");
@@ -199,7 +213,7 @@ __iobuf_arena_alloc (struct iobuf_pool *iobuf_pool, size_t page_size,
         return iobuf_arena;
 
 err:
-        __iobuf_arena_destroy (iobuf_arena);
+        __iobuf_arena_destroy (iobuf_pool, iobuf_arena);
 
 out:
         return NULL;
@@ -258,8 +272,8 @@ __iobuf_pool_add_arena (struct iobuf_pool *iobuf_pool, size_t page_size,
                 gf_log (THIS->name, GF_LOG_WARNING, "arena not found");
                 return NULL;
         }
+        list_add (&iobuf_arena->list, &iobuf_pool->arenas[index]);
 
-        list_add_tail (&iobuf_arena->list, &iobuf_pool->arenas[index]);
 
         return iobuf_arena;
 }
@@ -299,7 +313,8 @@ iobuf_pool_destroy (struct iobuf_pool *iobuf_pool)
                                           &iobuf_pool->arenas[i], list) {
                         list_del_init (&iobuf_arena->list);
                         iobuf_pool->arena_cnt--;
-                        __iobuf_arena_destroy (iobuf_arena);
+
+                        __iobuf_arena_destroy (iobuf_pool, iobuf_arena);
                 }
 
         }
@@ -347,7 +362,7 @@ iobuf_pool_new (void)
                                 gf_common_mt_iobuf_pool);
         if (!iobuf_pool)
                 goto out;
-
+        INIT_LIST_HEAD (&iobuf_pool->all_arenas);
         pthread_mutex_init (&iobuf_pool->mutex, NULL);
         for (i = 0; i <= IOBUF_ARENA_MAX_INDEX; i++) {
                 INIT_LIST_HEAD (&iobuf_pool->arenas[i]);
@@ -357,6 +372,16 @@ iobuf_pool_new (void)
 
         iobuf_pool->default_page_size  = 128 * GF_UNIT_KB;
 
+        iobuf_pool->rdma_registration = NULL;
+        iobuf_pool->rdma_deregistration = NULL;
+
+        for (i = 0; i < GF_RDMA_DEVICE_COUNT; i++) {
+
+                iobuf_pool->device[i] = NULL;
+                iobuf_pool->mr_list[i] = NULL;
+
+        }
+
         arena_size = 0;
         for (i = 0; i < IOBUF_ARENA_MAX_INDEX; i++) {
                 page_size = gf_iobuf_init_config[i].pagesize;
@@ -393,9 +418,10 @@ __iobuf_arena_prune (struct iobuf_pool *iobuf_pool,
 
         /* All cases matched, destroy */
         list_del_init (&iobuf_arena->list);
+        list_del_init (&iobuf_arena->all_list);
         iobuf_pool->arena_cnt--;
 
-        __iobuf_arena_destroy (iobuf_arena);
+        __iobuf_arena_destroy (iobuf_pool, iobuf_arena);
 
 out:
         return;
diff --git a/libglusterfs/src/iobuf.h b/libglusterfs/src/iobuf.h
index 4e07910d722..7e5cfe37a28 100644
--- a/libglusterfs/src/iobuf.h
+++ b/libglusterfs/src/iobuf.h
@@ -19,6 +19,8 @@
 
 #define GF_VARIABLE_IOBUF_COUNT 32
 
+#define GF_RDMA_DEVICE_COUNT 8
+
 /* Lets try to define the new anonymous mapping
  * flag, in case the system is still using the
  * now deprecated MAP_ANON flag.
@@ -81,6 +83,7 @@ struct iobuf_arena {
                 };
         };
 
+        struct list_head    all_list;
         size_t              page_size;  /* size of all iobufs in this arena */
         size_t              arena_size; /* this is equal to
                                            (iobuf_pool->arena_size / page_size)
@@ -110,6 +113,7 @@ struct iobuf_pool {
         size_t              default_page_size; /* default size of iobuf */
 
         int                 arena_cnt;
+        struct list_head    all_arenas;
         struct list_head    arenas[GF_VARIABLE_IOBUF_COUNT];
         /* array of arenas. Each element of the array is a list of arenas
            holding iobufs of particular page_size */
@@ -121,7 +125,13 @@ struct iobuf_pool {
         /* array of of arenas which can be purged */
 
         uint64_t            request_misses; /* mostly the requests for higher
-                                               value of iobufs */
+                                              value of iobufs */
+        int                 rdma_device_count;
+        struct list_head    *mr_list[GF_RDMA_DEVICE_COUNT];
+        void                *device[GF_RDMA_DEVICE_COUNT];
+        int (*rdma_registration)(void **, void*);
+        int (*rdma_deregistration)(struct list_head**, struct iobuf_arena *);
+
 };
 
 
diff --git a/libglusterfs/src/mem-types.h b/libglusterfs/src/mem-types.h
index 4f566f9ec57..4359488c5f9 100644
--- a/libglusterfs/src/mem-types.h
+++ b/libglusterfs/src/mem-types.h
@@ -126,6 +126,7 @@ enum gf_common_mem_types_ {
 	gf_common_mt_strfd_data_t         = 110,
         gf_common_mt_regex_t              = 111,
         gf_common_mt_wr                   = 112,
+        gf_common_mt_rdma_arena_mr        = 113,
         gf_common_mt_end
 };
 #endif