libceph: respect RADOS_BACKOFF backoffs
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
diff --git a/include/linux/ceph/ceph_fs.h b/include/linux/ceph/ceph_fs.h
index ad078eb..edf5b04 100644
--- a/include/linux/ceph/ceph_fs.h
+++ b/include/linux/ceph/ceph_fs.h
@@ -147,6 +147,7 @@ struct ceph_dir_layout {
#define CEPH_MSG_OSD_OP 42
#define CEPH_MSG_OSD_OPREPLY 43
#define CEPH_MSG_WATCH_NOTIFY 44
+#define CEPH_MSG_OSD_BACKOFF 61
/* watch-notify operations */
diff --git a/include/linux/ceph/osd_client.h b/include/linux/ceph/osd_client.h
index 62c672b..c6d96a5 100644
--- a/include/linux/ceph/osd_client.h
+++ b/include/linux/ceph/osd_client.h
@@ -1,6 +1,7 @@
#ifndef _FS_CEPH_OSD_CLIENT_H
#define _FS_CEPH_OSD_CLIENT_H
+#include <linux/bitrev.h>
#include <linux/completion.h>
#include <linux/kref.h>
#include <linux/mempool.h>
@@ -36,6 +37,8 @@ struct ceph_osd {
struct ceph_connection o_con;
struct rb_root o_requests;
struct rb_root o_linger_requests;
+ struct rb_root o_backoff_mappings;
+ struct rb_root o_backoffs_by_id;
struct list_head o_osd_lru;
struct ceph_auth_handshake o_auth;
unsigned long lru_ttl;
@@ -275,6 +278,48 @@ struct ceph_watch_item {
struct ceph_entity_addr addr;
};
+struct ceph_spg_mapping {
+ struct rb_node node;
+ struct ceph_spg spgid;
+
+ struct rb_root backoffs;
+};
+
+struct ceph_hobject_id {
+ void *key;
+ size_t key_len;
+ void *oid;
+ size_t oid_len;
+ u64 snapid;
+ u32 hash;
+ u8 is_max;
+ void *nspace;
+ size_t nspace_len;
+ s64 pool;
+
+ /* cache */
+ u32 hash_reverse_bits;
+};
+
+static inline void ceph_hoid_build_hash_cache(struct ceph_hobject_id *hoid)
+{
+ hoid->hash_reverse_bits = bitrev32(hoid->hash);
+}
+
+/*
+ * PG-wide backoff: [begin, end)
+ * per-object backoff: begin == end
+ */
+struct ceph_osd_backoff {
+ struct rb_node spg_node;
+ struct rb_node id_node;
+
+ struct ceph_spg spgid;
+ u64 id;
+ struct ceph_hobject_id *begin;
+ struct ceph_hobject_id *end;
+};
+
#define CEPH_LINGER_ID_START 0xffff000000000000ULL
struct ceph_osd_client {
diff --git a/include/linux/ceph/osdmap.h b/include/linux/ceph/osdmap.h
index 060d059..fe6d189 100644
--- a/include/linux/ceph/osdmap.h
+++ b/include/linux/ceph/osdmap.h
@@ -32,6 +32,7 @@ struct ceph_spg {
};
int ceph_pg_compare(const struct ceph_pg *lhs, const struct ceph_pg *rhs);
+int ceph_spg_compare(const struct ceph_spg *lhs, const struct ceph_spg *rhs);
#define CEPH_POOL_FLAG_HASHPSPOOL (1ULL << 0) /* hash pg seed and pool id
together */
diff --git a/include/linux/ceph/rados.h b/include/linux/ceph/rados.h
index 5d00187..385db08 100644
--- a/include/linux/ceph/rados.h
+++ b/include/linux/ceph/rados.h
@@ -439,6 +439,12 @@ enum {
const char *ceph_osd_watch_op_name(int o);
+enum {
+ CEPH_OSD_BACKOFF_OP_BLOCK = 1,
+ CEPH_OSD_BACKOFF_OP_ACK_BLOCK = 2,
+ CEPH_OSD_BACKOFF_OP_UNBLOCK = 3,
+};
+
/*
* an individual object operation. each may be accompanied by some data
* payload