libceph: switch to calc_target(), part 1
authorIlya Dryomov <idryomov@gmail.com>
Thu, 28 Apr 2016 14:07:23 +0000 (16:07 +0200)
committerIlya Dryomov <idryomov@gmail.com>
Wed, 25 May 2016 22:36:26 +0000 (00:36 +0200)
Replace __calc_request_pg() and most of __map_request() with
calc_target() and start using req->r_t.

ceph_osdc_build_request() however still encodes base_oid, because it's
called before calc_target() is and target_oid is empty at that point in
time; a printf in osdc_show() also shows base_oid.  This is fixed in
"libceph: switch to calc_target(), part 2".

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
include/linux/ceph/osd_client.h
net/ceph/debugfs.c
net/ceph/osd_client.c

index 48806ee4488d012b7daba36201da30b2ede1f5a4..03bf9d9e1517bc92fd9bba910168ecf5ff5e25e2 100644 (file)
@@ -150,12 +150,13 @@ struct ceph_osd_request {
        struct list_head r_linger_item;
        struct list_head r_linger_osd_item;
        struct ceph_osd *r_osd;
-       struct ceph_pg   r_pgid;
-       int              r_pg_osds[CEPH_PG_MAX_SIZE];
-       int              r_num_pg_osds;
+
+       struct ceph_osd_request_target r_t;
+#define r_base_oid     r_t.base_oid
+#define r_base_oloc    r_t.base_oloc
+#define r_flags                r_t.flags
 
        struct ceph_msg  *r_request, *r_reply;
-       int               r_flags;     /* any additional flags for the osd */
        u32               r_sent;      /* >0 if r_request is sending/sent */
 
        /* request osd ops array  */
@@ -167,7 +168,6 @@ struct ceph_osd_request {
        __le64           *r_request_pool;
        void             *r_request_pgid;
        __le32           *r_request_attempts;
-       bool              r_paused;
        struct ceph_eversion *r_request_reassert_version;
 
        int               r_result;
@@ -186,11 +186,6 @@ struct ceph_osd_request {
        struct inode *r_inode;                /* for use by callbacks */
        void *r_priv;                         /* ditto */
 
-       struct ceph_object_locator r_base_oloc;
-       struct ceph_object_id r_base_oid;
-       struct ceph_object_locator r_target_oloc;
-       struct ceph_object_id r_target_oid;
-
        u64               r_snapid;
        unsigned long     r_stamp;            /* send OR check time */
 
index 7f1cc22c3e8ba25af7aac80dc026f2c9ce117067..0c11ab5f8c30b2cada3b53bcd5baea936fac5370 100644 (file)
@@ -161,7 +161,7 @@ static int osdc_show(struct seq_file *s, void *pp)
 
                seq_printf(s, "%lld\tosd%d\t%lld.%x\t", req->r_tid,
                           req->r_osd ? req->r_osd->o_osd : -1,
-                          req->r_pgid.pool, req->r_pgid.seed);
+                          req->r_t.pgid.pool, req->r_t.pgid.seed);
 
                seq_printf(s, "%*pE", req->r_base_oid.name_len,
                           req->r_base_oid.name);
index cff3a7e29233db8e038df72f52e0afbfff09d576..013101598c4109ef1bed5daccbf738f832d9b8f8 100644 (file)
@@ -350,8 +350,7 @@ static void ceph_osdc_release_request(struct kref *kref)
        for (which = 0; which < req->r_num_ops; which++)
                osd_req_op_data_release(req, which);
 
-       ceph_oid_destroy(&req->r_base_oid);
-       ceph_oid_destroy(&req->r_target_oid);
+       target_destroy(&req->r_t);
        ceph_put_snap_context(req->r_snapc);
 
        if (req->r_mempool)
@@ -420,10 +419,7 @@ struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
        INIT_LIST_HEAD(&req->r_req_lru_item);
        INIT_LIST_HEAD(&req->r_osd_item);
 
-       ceph_oid_init(&req->r_base_oid);
-       req->r_base_oloc.pool = -1;
-       ceph_oid_init(&req->r_target_oid);
-       req->r_target_oloc.pool = -1;
+       target_init(&req->r_t);
 
        dout("%s req %p\n", __func__, req);
        return req;
@@ -1308,16 +1304,6 @@ static bool __pool_full(struct ceph_pg_pool_info *pi)
  *
  * Caller should hold map_sem for read.
  */
-static bool __req_should_be_paused(struct ceph_osd_client *osdc,
-                                  struct ceph_osd_request *req)
-{
-       bool pauserd = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD);
-       bool pausewr = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSEWR) ||
-               ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL);
-       return (req->r_flags & CEPH_OSD_FLAG_READ && pauserd) ||
-               (req->r_flags & CEPH_OSD_FLAG_WRITE && pausewr);
-}
-
 static bool target_should_be_paused(struct ceph_osd_client *osdc,
                                    const struct ceph_osd_request_target *t,
                                    struct ceph_pg_pool_info *pi)
@@ -1332,45 +1318,6 @@ static bool target_should_be_paused(struct ceph_osd_client *osdc,
               (t->flags & CEPH_OSD_FLAG_WRITE && pausewr);
 }
 
-/*
- * Calculate mapping of a request to a PG.  Takes tiering into account.
- */
-static int __calc_request_pg(struct ceph_osdmap *osdmap,
-                            struct ceph_osd_request *req,
-                            struct ceph_pg *pg_out)
-{
-       bool need_check_tiering;
-
-       need_check_tiering = false;
-       if (req->r_target_oloc.pool == -1) {
-               req->r_target_oloc = req->r_base_oloc; /* struct */
-               need_check_tiering = true;
-       }
-       if (ceph_oid_empty(&req->r_target_oid)) {
-               ceph_oid_copy(&req->r_target_oid, &req->r_base_oid);
-               need_check_tiering = true;
-       }
-
-       if (need_check_tiering &&
-           (req->r_flags & CEPH_OSD_FLAG_IGNORE_OVERLAY) == 0) {
-               struct ceph_pg_pool_info *pi;
-
-               pi = ceph_pg_pool_by_id(osdmap, req->r_target_oloc.pool);
-               if (pi) {
-                       if ((req->r_flags & CEPH_OSD_FLAG_READ) &&
-                           pi->read_tier >= 0)
-                               req->r_target_oloc.pool = pi->read_tier;
-                       if ((req->r_flags & CEPH_OSD_FLAG_WRITE) &&
-                           pi->write_tier >= 0)
-                               req->r_target_oloc.pool = pi->write_tier;
-               }
-               /* !pi is caught in ceph_oloc_oid_to_pg() */
-       }
-
-       return ceph_object_locator_to_pg(osdmap, &req->r_target_oid,
-                                        &req->r_target_oloc, pg_out);
-}
-
 enum calc_target_result {
        CALC_TARGET_NO_ACTION = 0,
        CALC_TARGET_NEED_RESEND,
@@ -1510,46 +1457,26 @@ static void __enqueue_request(struct ceph_osd_request *req)
 static int __map_request(struct ceph_osd_client *osdc,
                         struct ceph_osd_request *req, int force_resend)
 {
-       struct ceph_pg pgid;
-       struct ceph_osds up, acting;
+       enum calc_target_result ct_res;
        int err;
-       bool was_paused;
 
        dout("map_request %p tid %lld\n", req, req->r_tid);
 
-       err = __calc_request_pg(osdc->osdmap, req, &pgid);
-       if (err) {
+       ct_res = calc_target(osdc, &req->r_t, NULL, force_resend);
+       switch (ct_res) {
+       case CALC_TARGET_POOL_DNE:
                list_move(&req->r_req_lru_item, &osdc->req_notarget);
-               return err;
-       }
-       req->r_pgid = pgid;
-
-       ceph_pg_to_up_acting_osds(osdc->osdmap, &pgid, &up, &acting);
-
-       was_paused = req->r_paused;
-       req->r_paused = __req_should_be_paused(osdc, req);
-       if (was_paused && !req->r_paused)
-               force_resend = 1;
-
-       if ((!force_resend &&
-            req->r_osd && req->r_osd->o_osd == acting.primary &&
-            req->r_sent >= req->r_osd->o_incarnation &&
-            req->r_num_pg_osds == acting.size &&
-            memcmp(req->r_pg_osds, acting.osds,
-                   acting.size * sizeof(acting.osds[0])) == 0) ||
-           (req->r_osd == NULL && acting.primary == -1) ||
-           req->r_paused)
+               return -EIO;
+       case CALC_TARGET_NO_ACTION:
                return 0;  /* no change */
+       default:
+               BUG_ON(ct_res != CALC_TARGET_NEED_RESEND);
+       }
 
        dout("map_request tid %llu pgid %lld.%x osd%d (was osd%d)\n",
-            req->r_tid, pgid.pool, pgid.seed, acting.primary,
+            req->r_tid, req->r_t.pgid.pool, req->r_t.pgid.seed, req->r_t.osd,
             req->r_osd ? req->r_osd->o_osd : -1);
 
-       /* record full pg acting set */
-       memcpy(req->r_pg_osds, acting.osds,
-              acting.size * sizeof(acting.osds[0]));
-       req->r_num_pg_osds = acting.size;
-
        if (req->r_osd) {
                __cancel_request(req);
                list_del_init(&req->r_osd_item);
@@ -1557,22 +1484,22 @@ static int __map_request(struct ceph_osd_client *osdc,
                req->r_osd = NULL;
        }
 
-       req->r_osd = lookup_osd(&osdc->osds, acting.primary);
-       if (!req->r_osd && acting.primary >= 0) {
+       req->r_osd = lookup_osd(&osdc->osds, req->r_t.osd);
+       if (!req->r_osd && req->r_t.osd >= 0) {
                err = -ENOMEM;
-               req->r_osd = create_osd(osdc, acting.primary);
+               req->r_osd = create_osd(osdc, req->r_t.osd);
                if (!req->r_osd) {
                        list_move(&req->r_req_lru_item, &osdc->req_notarget);
                        goto out;
                }
 
                dout("map_request osd %p is osd%d\n", req->r_osd,
-                    acting.primary);
+                    req->r_osd->o_osd);
                insert_osd(&osdc->osds, req->r_osd);
 
                ceph_con_open(&req->r_osd->o_con,
-                             CEPH_ENTITY_TYPE_OSD, acting.primary,
-                             &osdc->osdmap->osd_addr[acting.primary]);
+                             CEPH_ENTITY_TYPE_OSD, req->r_osd->o_osd,
+                             &osdc->osdmap->osd_addr[req->r_osd->o_osd]);
        }
 
        __enqueue_request(req);
@@ -1592,15 +1519,15 @@ static void __send_request(struct ceph_osd_client *osdc,
 
        dout("send_request %p tid %llu to osd%d flags %d pg %lld.%x\n",
             req, req->r_tid, req->r_osd->o_osd, req->r_flags,
-            (unsigned long long)req->r_pgid.pool, req->r_pgid.seed);
+            req->r_t.pgid.pool, req->r_t.pgid.seed);
 
        /* fill in message content that changes each time we send it */
        put_unaligned_le32(osdc->osdmap->epoch, req->r_request_osdmap_epoch);
        put_unaligned_le32(req->r_flags, req->r_request_flags);
-       put_unaligned_le64(req->r_target_oloc.pool, req->r_request_pool);
+       put_unaligned_le64(req->r_t.target_oloc.pool, req->r_request_pool);
        p = req->r_request_pgid;
-       ceph_encode_64(&p, req->r_pgid.pool);
-       ceph_encode_32(&p, req->r_pgid.seed);
+       ceph_encode_64(&p, req->r_t.pgid.pool);
+       ceph_encode_32(&p, req->r_t.pgid.seed);
        put_unaligned_le64(1, req->r_request_attempts);  /* FIXME */
        memcpy(req->r_request_reassert_version, &req->r_reassert_version,
               sizeof(req->r_reassert_version));
@@ -1963,7 +1890,7 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg)
 
                __unregister_request(osdc, req);
 
-               ceph_oloc_copy(&req->r_target_oloc, &redir.oloc);
+               ceph_oloc_copy(&req->r_t.target_oloc, &redir.oloc);
 
                /*
                 * Start redirect requests with nofail=true.  If