From dcbbd97ccb9c6f4dad39875c1404d2643eaf110b Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 5 Jun 2017 14:44:59 +0200 Subject: libceph: remove ceph_sanitize_features() workaround Reflects ceph.git commit ff1959282826ae6acd7134e1b1ede74ffd1cc04a. Signed-off-by: Ilya Dryomov --- include/linux/ceph/ceph_features.h | 21 --------------------- 1 file changed, 21 deletions(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/ceph_features.h b/include/linux/ceph/ceph_features.h index fd8b2953c78f..4962708841b5 100644 --- a/include/linux/ceph/ceph_features.h +++ b/include/linux/ceph/ceph_features.h @@ -77,29 +77,8 @@ #define CEPH_FEATURE_NEW_OSDOPREPLY_ENCODING (1ULL<<58) /* New, v7 encoding */ #define CEPH_FEATURE_FS_FILE_LAYOUT_V2 (1ULL<<58) /* file_layout_t */ -/* - * The introduction of CEPH_FEATURE_OSD_SNAPMAPPER caused the feature - * vector to evaluate to 64 bit ~0. To cope, we designate 1ULL << 63 - * to mean 33 bit ~0, and introduce a helper below to do the - * translation. - * - * This was introduced by ceph.git commit - * 9ea02b84104045c2ffd7e7f4e7af512953855ecd v0.58-657-g9ea02b8 - * and fixed by ceph.git commit - * 4255b5c2fb54ae40c53284b3ab700fdfc7e61748 v0.65-263-g4255b5c - */ #define CEPH_FEATURE_RESERVED (1ULL<<63) -static inline u64 ceph_sanitize_features(u64 features) -{ - if (features & CEPH_FEATURE_RESERVED) { - /* everything through OSD_SNAPMAPPER */ - return 0x1ffffffffull; - } else { - return features; - } -} - /* * Features supported. */ -- cgit v1.2.3 From f179d3ba8cb9073c2d96315b79ff7bc658a1feee Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 5 Jun 2017 14:44:59 +0200 Subject: libceph: new features macros Signed-off-by: Ilya Dryomov --- include/linux/ceph/ceph_features.h | 242 +++++++++++++++++++++++++------------ 1 file changed, 167 insertions(+), 75 deletions(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/ceph_features.h b/include/linux/ceph/ceph_features.h index 4962708841b5..7fce9ed44af0 100644 --- a/include/linux/ceph/ceph_features.h +++ b/include/linux/ceph/ceph_features.h @@ -2,82 +2,174 @@ #define __CEPH_FEATURES /* - * feature bits + * Each time we reclaim bits for reuse we need to specify another bit + * that, if present, indicates we have the new incarnation of that + * feature. Base case is 1 (first use). */ -#define CEPH_FEATURE_UID (1ULL<<0) -#define CEPH_FEATURE_NOSRCADDR (1ULL<<1) -#define CEPH_FEATURE_MONCLOCKCHECK (1ULL<<2) -#define CEPH_FEATURE_FLOCK (1ULL<<3) -#define CEPH_FEATURE_SUBSCRIBE2 (1ULL<<4) -#define CEPH_FEATURE_MONNAMES (1ULL<<5) -#define CEPH_FEATURE_RECONNECT_SEQ (1ULL<<6) -#define CEPH_FEATURE_DIRLAYOUTHASH (1ULL<<7) -#define CEPH_FEATURE_OBJECTLOCATOR (1ULL<<8) -#define CEPH_FEATURE_PGID64 (1ULL<<9) -#define CEPH_FEATURE_INCSUBOSDMAP (1ULL<<10) -#define CEPH_FEATURE_PGPOOL3 (1ULL<<11) -#define CEPH_FEATURE_OSDREPLYMUX (1ULL<<12) -#define CEPH_FEATURE_OSDENC (1ULL<<13) -#define CEPH_FEATURE_OMAP (1ULL<<14) -#define CEPH_FEATURE_MONENC (1ULL<<15) -#define CEPH_FEATURE_QUERY_T (1ULL<<16) -#define CEPH_FEATURE_INDEP_PG_MAP (1ULL<<17) -#define CEPH_FEATURE_CRUSH_TUNABLES (1ULL<<18) -#define CEPH_FEATURE_CHUNKY_SCRUB (1ULL<<19) -#define CEPH_FEATURE_MON_NULLROUTE (1ULL<<20) -#define CEPH_FEATURE_MON_GV (1ULL<<21) -#define CEPH_FEATURE_BACKFILL_RESERVATION (1ULL<<22) -#define CEPH_FEATURE_MSG_AUTH (1ULL<<23) -#define CEPH_FEATURE_RECOVERY_RESERVATION (1ULL<<24) -#define CEPH_FEATURE_CRUSH_TUNABLES2 (1ULL<<25) -#define CEPH_FEATURE_CREATEPOOLID (1ULL<<26) -#define CEPH_FEATURE_REPLY_CREATE_INODE (1ULL<<27) -#define CEPH_FEATURE_OSD_HBMSGS (1ULL<<28) -#define CEPH_FEATURE_MDSENC (1ULL<<29) -#define CEPH_FEATURE_OSDHASHPSPOOL (1ULL<<30) -#define CEPH_FEATURE_MON_SINGLE_PAXOS (1ULL<<31) -#define CEPH_FEATURE_OSD_SNAPMAPPER (1ULL<<32) -#define CEPH_FEATURE_MON_SCRUB (1ULL<<33) -#define CEPH_FEATURE_OSD_PACKED_RECOVERY (1ULL<<34) -#define CEPH_FEATURE_OSD_CACHEPOOL (1ULL<<35) -#define CEPH_FEATURE_CRUSH_V2 (1ULL<<36) /* new indep; SET_* steps */ -#define CEPH_FEATURE_EXPORT_PEER (1ULL<<37) -#define CEPH_FEATURE_OSD_ERASURE_CODES (1ULL<<38) -#define CEPH_FEATURE_OSD_TMAP2OMAP (1ULL<<38) /* overlap with EC */ -/* The process supports new-style OSDMap encoding. Monitors also use - this bit to determine if peers support NAK messages. */ -#define CEPH_FEATURE_OSDMAP_ENC (1ULL<<39) -#define CEPH_FEATURE_MDS_INLINE_DATA (1ULL<<40) -#define CEPH_FEATURE_CRUSH_TUNABLES3 (1ULL<<41) -#define CEPH_FEATURE_OSD_PRIMARY_AFFINITY (1ULL<<41) /* overlap w/ tunables3 */ -#define CEPH_FEATURE_MSGR_KEEPALIVE2 (1ULL<<42) -#define CEPH_FEATURE_OSD_POOLRESEND (1ULL<<43) -#define CEPH_FEATURE_ERASURE_CODE_PLUGINS_V2 (1ULL<<44) -#define CEPH_FEATURE_OSD_SET_ALLOC_HINT (1ULL<<45) -#define CEPH_FEATURE_OSD_FADVISE_FLAGS (1ULL<<46) -#define CEPH_FEATURE_OSD_REPOP (1ULL<<46) /* overlap with fadvise */ -#define CEPH_FEATURE_OSD_OBJECT_DIGEST (1ULL<<46) /* overlap with fadvise */ -#define CEPH_FEATURE_OSD_TRANSACTION_MAY_LAYOUT (1ULL<<46) /* overlap w/ fadvise */ -#define CEPH_FEATURE_MDS_QUOTA (1ULL<<47) -#define CEPH_FEATURE_CRUSH_V4 (1ULL<<48) /* straw2 buckets */ -#define CEPH_FEATURE_OSD_MIN_SIZE_RECOVERY (1ULL<<49) -// duplicated since it was introduced at the same time as MIN_SIZE_RECOVERY -#define CEPH_FEATURE_OSD_PROXY_FEATURES (1ULL<<49) /* overlap w/ above */ -#define CEPH_FEATURE_MON_METADATA (1ULL<<50) -#define CEPH_FEATURE_OSD_BITWISE_HOBJ_SORT (1ULL<<51) /* can sort objs bitwise */ -#define CEPH_FEATURE_OSD_PROXY_WRITE_FEATURES (1ULL<<52) -#define CEPH_FEATURE_ERASURE_CODE_PLUGINS_V3 (1ULL<<53) -#define CEPH_FEATURE_OSD_HITSET_GMT (1ULL<<54) -#define CEPH_FEATURE_HAMMER_0_94_4 (1ULL<<55) -#define CEPH_FEATURE_NEW_OSDOP_ENCODING (1ULL<<56) /* New, v7 encoding */ -#define CEPH_FEATURE_MON_STATEFUL_SUB (1ULL<<57) /* stateful mon subscription */ -#define CEPH_FEATURE_MON_ROUTE_OSDMAP (1ULL<<57) /* peon sends osdmaps */ -#define CEPH_FEATURE_CRUSH_TUNABLES5 (1ULL<<58) /* chooseleaf stable mode */ -// duplicated since it was introduced at the same time as CEPH_FEATURE_CRUSH_TUNABLES5 -#define CEPH_FEATURE_NEW_OSDOPREPLY_ENCODING (1ULL<<58) /* New, v7 encoding */ -#define CEPH_FEATURE_FS_FILE_LAYOUT_V2 (1ULL<<58) /* file_layout_t */ - -#define CEPH_FEATURE_RESERVED (1ULL<<63) +#define CEPH_FEATURE_INCARNATION_1 (0ull) +#define CEPH_FEATURE_INCARNATION_2 (1ull<<57) // CEPH_FEATURE_SERVER_JEWEL + +#define DEFINE_CEPH_FEATURE(bit, incarnation, name) \ + const static uint64_t CEPH_FEATURE_##name = (1ULL< Date: Mon, 5 Jun 2017 14:45:00 +0200 Subject: libceph: advertise support for OSD_POOLRESEND The code has been in place since commit 63244fa123a7 ("libceph: introduce ceph_osd_request_target, calc_target()"), and, with the ceph_{oloc,oid}_copy() issue fixed in the previous commit, is now in working order. Signed-off-by: Ilya Dryomov --- include/linux/ceph/ceph_features.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/ceph_features.h b/include/linux/ceph/ceph_features.h index 7fce9ed44af0..89c68af48539 100644 --- a/include/linux/ceph/ceph_features.h +++ b/include/linux/ceph/ceph_features.h @@ -197,6 +197,7 @@ DEFINE_CEPH_FEATURE_DEPRECATED(63, 1, RESERVED_BROKEN, LUMINOUS) // client-facin CEPH_FEATURE_CRUSH_TUNABLES3 | \ CEPH_FEATURE_OSD_PRIMARY_AFFINITY | \ CEPH_FEATURE_MSGR_KEEPALIVE2 | \ + CEPH_FEATURE_OSD_POOLRESEND | \ CEPH_FEATURE_CRUSH_V4 | \ CEPH_FEATURE_CRUSH_TUNABLES5 | \ CEPH_FEATURE_NEW_OSDOPREPLY_ENCODING) -- cgit v1.2.3 From 220abf5aa7ba5f544f1b589bde33761c60bbf9a0 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 5 Jun 2017 14:45:00 +0200 Subject: libceph: support SERVER_JEWEL feature bits Only MON_STATEFUL_SUB, really. MON_ROUTE_OSDMAP and OSDSUBOP_NO_SNAPCONTEXT are irrelevant. Signed-off-by: Ilya Dryomov --- include/linux/ceph/ceph_features.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/ceph_features.h b/include/linux/ceph/ceph_features.h index 89c68af48539..78a58770e6e9 100644 --- a/include/linux/ceph/ceph_features.h +++ b/include/linux/ceph/ceph_features.h @@ -199,6 +199,8 @@ DEFINE_CEPH_FEATURE_DEPRECATED(63, 1, RESERVED_BROKEN, LUMINOUS) // client-facin CEPH_FEATURE_MSGR_KEEPALIVE2 | \ CEPH_FEATURE_OSD_POOLRESEND | \ CEPH_FEATURE_CRUSH_V4 | \ + CEPH_FEATURE_SERVER_JEWEL | \ + CEPH_FEATURE_MON_STATEFUL_SUB | \ CEPH_FEATURE_CRUSH_TUNABLES5 | \ CEPH_FEATURE_NEW_OSDOPREPLY_ENCODING) -- cgit v1.2.3 From dc93e0e2831de2f80817b89aae4864b88332fcce Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 5 Jun 2017 14:45:00 +0200 Subject: libceph: fold [l]req->last_force_resend into ceph_osd_request_target Signed-off-by: Ilya Dryomov --- include/linux/ceph/osd_client.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osd_client.h b/include/linux/ceph/osd_client.h index 85650b415e73..ef630ebd1169 100644 --- a/include/linux/ceph/osd_client.h +++ b/include/linux/ceph/osd_client.h @@ -148,6 +148,8 @@ struct ceph_osd_request_target { unsigned int flags; /* CEPH_OSD_FLAG_* */ bool paused; + u32 last_force_resend; + int osd; }; @@ -193,7 +195,6 @@ struct ceph_osd_request { unsigned long r_stamp; /* jiffies, send or check time */ unsigned long r_start_stamp; /* jiffies */ int r_attempts; - u32 r_last_force_resend; u32 r_map_dne_bound; struct ceph_osd_req_op r_ops[]; @@ -221,7 +222,6 @@ struct ceph_osd_linger_request { struct list_head pending_lworks; struct ceph_osd_request_target t; - u32 last_force_resend; u32 map_dne_bound; struct timespec mtime; -- cgit v1.2.3 From dc98ff7230e5ccf11c621dff0d590e83574a7184 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 15 Jun 2017 16:30:53 +0200 Subject: libceph: introduce ceph_spg, ceph_pg_to_primary_shard() Store both raw pgid and actual spgid in ceph_osd_request_target. Signed-off-by: Ilya Dryomov --- include/linux/ceph/osd_client.h | 3 ++- include/linux/ceph/osdmap.h | 10 ++++++++++ 2 files changed, 12 insertions(+), 1 deletion(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osd_client.h b/include/linux/ceph/osd_client.h index ef630ebd1169..6114f7b02135 100644 --- a/include/linux/ceph/osd_client.h +++ b/include/linux/ceph/osd_client.h @@ -136,7 +136,8 @@ struct ceph_osd_request_target { struct ceph_object_id target_oid; struct ceph_object_locator target_oloc; - struct ceph_pg pgid; + struct ceph_pg pgid; /* last raw pg we mapped to */ + struct ceph_spg spgid; /* last actual spg we mapped to */ u32 pg_num; u32 pg_num_mask; struct ceph_osds acting; diff --git a/include/linux/ceph/osdmap.h b/include/linux/ceph/osdmap.h index 938656f70807..7ae5b416b4b6 100644 --- a/include/linux/ceph/osdmap.h +++ b/include/linux/ceph/osdmap.h @@ -24,6 +24,13 @@ struct ceph_pg { uint32_t seed; }; +#define CEPH_SPG_NOSHARD -1 + +struct ceph_spg { + struct ceph_pg pgid; + s8 shard; +}; + int ceph_pg_compare(const struct ceph_pg *lhs, const struct ceph_pg *rhs); #define CEPH_POOL_FLAG_HASHPSPOOL (1ULL << 0) /* hash pg seed and pool id @@ -271,6 +278,9 @@ void ceph_pg_to_up_acting_osds(struct ceph_osdmap *osdmap, const struct ceph_pg *raw_pgid, struct ceph_osds *up, struct ceph_osds *acting); +bool ceph_pg_to_primary_shard(struct ceph_osdmap *osdmap, + const struct ceph_pg *raw_pgid, + struct ceph_spg *spgid); int ceph_pg_to_acting_primary(struct ceph_osdmap *osdmap, const struct ceph_pg *raw_pgid); -- cgit v1.2.3 From 98ad5ebd1505eb903ae8bc27e94c1ab0d1c3e651 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 15 Jun 2017 16:30:54 +0200 Subject: libceph: ceph_connection_operations::reencode_message() method Give upper layers a chance to reencode the message after the connection is negotiated and ->peer_features is set. OSD client will use this to support both luminous and pre-luminous OSDs (in a single cluster): the former need MOSDOp v8; the latter will continue to be sent MOSDOp v4. Signed-off-by: Ilya Dryomov --- include/linux/ceph/messenger.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/messenger.h b/include/linux/ceph/messenger.h index c5c4c713e00f..fbd94d9fa5dd 100644 --- a/include/linux/ceph/messenger.h +++ b/include/linux/ceph/messenger.h @@ -44,6 +44,8 @@ struct ceph_connection_operations { struct ceph_msg_header *hdr, int *skip); + void (*reencode_message) (struct ceph_msg *msg); + int (*sign_message) (struct ceph_msg *msg); int (*check_message_signature) (struct ceph_msg *msg); }; -- cgit v1.2.3 From 8cb441c0545dfd4dafeedc1e2d7157e1072413ac Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 15 Jun 2017 16:30:54 +0200 Subject: libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov --- include/linux/ceph/osd_client.h | 17 +++++++++++++++++ include/linux/ceph/osdmap.h | 4 +++- 2 files changed, 20 insertions(+), 1 deletion(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osd_client.h b/include/linux/ceph/osd_client.h index 6114f7b02135..bca2718ac253 100644 --- a/include/linux/ceph/osd_client.h +++ b/include/linux/ceph/osd_client.h @@ -205,6 +205,23 @@ struct ceph_request_redirect { struct ceph_object_locator oloc; }; +/* + * osd request identifier + * + * caller name + incarnation# + tid to unique identify this request + */ +struct ceph_osd_reqid { + struct ceph_entity_name name; + __le64 tid; + __le32 inc; +} __packed; + +struct ceph_blkin_trace_info { + __le64 trace_id; + __le64 span_id; + __le64 parent_span_id; +} __packed; + typedef void (*rados_watchcb2_t)(void *arg, u64 notify_id, u64 cookie, u64 notifier_id, void *data, size_t data_len); typedef void (*rados_watcherrcb_t)(void *arg, u64 cookie, int err); diff --git a/include/linux/ceph/osdmap.h b/include/linux/ceph/osdmap.h index 7ae5b416b4b6..66447fc7f334 100644 --- a/include/linux/ceph/osdmap.h +++ b/include/linux/ceph/osdmap.h @@ -205,11 +205,13 @@ static inline struct ceph_entity_addr *ceph_osd_addr(struct ceph_osdmap *map, return &map->osd_addr[osd]; } +#define CEPH_PGID_ENCODING_LEN (1 + 8 + 4 + 4) + static inline int ceph_decode_pgid(void **p, void *end, struct ceph_pg *pgid) { __u8 version; - if (!ceph_has_room(p, end, 1 + 8 + 4 + 4)) { + if (!ceph_has_room(p, end, CEPH_PGID_ENCODING_LEN)) { pr_warn("incomplete pg encoding\n"); return -EINVAL; } -- cgit v1.2.3 From 7de030d6b10a56e991312a978ace6be3c090097c Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 15 Jun 2017 16:30:54 +0200 Subject: libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov --- include/linux/ceph/osdmap.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osdmap.h b/include/linux/ceph/osdmap.h index 66447fc7f334..63fb073a3355 100644 --- a/include/linux/ceph/osdmap.h +++ b/include/linux/ceph/osdmap.h @@ -249,6 +249,8 @@ static inline void ceph_osds_init(struct ceph_osds *set) void ceph_osds_copy(struct ceph_osds *dest, const struct ceph_osds *src); +bool ceph_pg_is_split(const struct ceph_pg *pgid, u32 old_pg_num, + u32 new_pg_num); bool ceph_is_new_interval(const struct ceph_osds *old_acting, const struct ceph_osds *new_acting, const struct ceph_osds *old_up, -- cgit v1.2.3 From 04c7d789e269c2b82bbd08106049a5a979cdb3fd Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 15 Jun 2017 16:30:55 +0200 Subject: libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov --- include/linux/ceph/osd_client.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osd_client.h b/include/linux/ceph/osd_client.h index bca2718ac253..62c672bcbb31 100644 --- a/include/linux/ceph/osd_client.h +++ b/include/linux/ceph/osd_client.h @@ -149,6 +149,7 @@ struct ceph_osd_request_target { unsigned int flags; /* CEPH_OSD_FLAG_* */ bool paused; + u32 epoch; u32 last_force_resend; int osd; -- cgit v1.2.3 From df28152d53b449a72258000f592472215fc9371e Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 15 Jun 2017 16:30:56 +0200 Subject: libceph: avoid unnecessary pi lookups in calc_target() Signed-off-by: Ilya Dryomov --- include/linux/ceph/osdmap.h | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osdmap.h b/include/linux/ceph/osdmap.h index 63fb073a3355..060d059acbf8 100644 --- a/include/linux/ceph/osdmap.h +++ b/include/linux/ceph/osdmap.h @@ -273,16 +273,22 @@ extern int ceph_calc_file_object_mapping(struct ceph_file_layout *layout, u64 off, u64 len, u64 *bno, u64 *oxoff, u64 *oxlen); +int __ceph_object_locator_to_pg(struct ceph_pg_pool_info *pi, + const struct ceph_object_id *oid, + const struct ceph_object_locator *oloc, + struct ceph_pg *raw_pgid); int ceph_object_locator_to_pg(struct ceph_osdmap *osdmap, - struct ceph_object_id *oid, - struct ceph_object_locator *oloc, + const struct ceph_object_id *oid, + const struct ceph_object_locator *oloc, struct ceph_pg *raw_pgid); void ceph_pg_to_up_acting_osds(struct ceph_osdmap *osdmap, + struct ceph_pg_pool_info *pi, const struct ceph_pg *raw_pgid, struct ceph_osds *up, struct ceph_osds *acting); bool ceph_pg_to_primary_shard(struct ceph_osdmap *osdmap, + struct ceph_pg_pool_info *pi, const struct ceph_pg *raw_pgid, struct ceph_spg *spgid); int ceph_pg_to_acting_primary(struct ceph_osdmap *osdmap, -- cgit v1.2.3 From 76f827a7b1faaaebc53f89d184e95ea3a0b8dd71 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 19 Jun 2017 12:18:05 +0200 Subject: libceph: make DEFINE_RB_* helpers more general Initially for ceph_pg_mapping, ceph_spg_mapping and ceph_hobject_id, compared with ceph_pg_compare(), ceph_spg_compare() and hoid_compare() respectively. Signed-off-by: Ilya Dryomov --- include/linux/ceph/libceph.h | 49 +++++++++++++++++++++++++++++++++----------- 1 file changed, 37 insertions(+), 12 deletions(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/libceph.h b/include/linux/ceph/libceph.h index 3229ae6c7846..8a79587e1317 100644 --- a/include/linux/ceph/libceph.h +++ b/include/linux/ceph/libceph.h @@ -184,10 +184,11 @@ static inline int calc_pages_for(u64 off, u64 len) (off >> PAGE_SHIFT); } -/* - * These are not meant to be generic - an integer key is assumed. - */ -#define DEFINE_RB_INSDEL_FUNCS(name, type, keyfld, nodefld) \ +#define RB_BYVAL(a) (a) +#define RB_BYPTR(a) (&(a)) +#define RB_CMP3WAY(a, b) ((a) < (b) ? -1 : (a) > (b)) + +#define DEFINE_RB_INSDEL_FUNCS2(name, type, keyfld, cmpexp, keyexp, nodefld) \ static void insert_##name(struct rb_root *root, type *t) \ { \ struct rb_node **n = &root->rb_node; \ @@ -197,11 +198,13 @@ static void insert_##name(struct rb_root *root, type *t) \ \ while (*n) { \ type *cur = rb_entry(*n, type, nodefld); \ + int cmp; \ \ parent = *n; \ - if (t->keyfld < cur->keyfld) \ + cmp = cmpexp(keyexp(t->keyfld), keyexp(cur->keyfld)); \ + if (cmp < 0) \ n = &(*n)->rb_left; \ - else if (t->keyfld > cur->keyfld) \ + else if (cmp > 0) \ n = &(*n)->rb_right; \ else \ BUG(); \ @@ -217,19 +220,24 @@ static void erase_##name(struct rb_root *root, type *t) \ RB_CLEAR_NODE(&t->nodefld); \ } -#define DEFINE_RB_LOOKUP_FUNC(name, type, keyfld, nodefld) \ -extern type __lookup_##name##_key; \ -static type *lookup_##name(struct rb_root *root, \ - typeof(__lookup_##name##_key.keyfld) key) \ +/* + * @lookup_param_type is a parameter and not constructed from (@type, + * @keyfld) with typeof() because adding const is too unwieldy. + */ +#define DEFINE_RB_LOOKUP_FUNC2(name, type, keyfld, cmpexp, keyexp, \ + lookup_param_type, nodefld) \ +static type *lookup_##name(struct rb_root *root, lookup_param_type key) \ { \ struct rb_node *n = root->rb_node; \ \ while (n) { \ type *cur = rb_entry(n, type, nodefld); \ + int cmp; \ \ - if (key < cur->keyfld) \ + cmp = cmpexp(key, keyexp(cur->keyfld)); \ + if (cmp < 0) \ n = n->rb_left; \ - else if (key > cur->keyfld) \ + else if (cmp > 0) \ n = n->rb_right; \ else \ return cur; \ @@ -238,6 +246,23 @@ static type *lookup_##name(struct rb_root *root, \ return NULL; \ } +#define DEFINE_RB_FUNCS2(name, type, keyfld, cmpexp, keyexp, \ + lookup_param_type, nodefld) \ +DEFINE_RB_INSDEL_FUNCS2(name, type, keyfld, cmpexp, keyexp, nodefld) \ +DEFINE_RB_LOOKUP_FUNC2(name, type, keyfld, cmpexp, keyexp, \ + lookup_param_type, nodefld) + +/* + * Shorthands for integer keys. + */ +#define DEFINE_RB_INSDEL_FUNCS(name, type, keyfld, nodefld) \ +DEFINE_RB_INSDEL_FUNCS2(name, type, keyfld, RB_CMP3WAY, RB_BYVAL, nodefld) + +#define DEFINE_RB_LOOKUP_FUNC(name, type, keyfld, nodefld) \ +extern type __lookup_##name##_key; \ +DEFINE_RB_LOOKUP_FUNC2(name, type, keyfld, RB_CMP3WAY, RB_BYVAL, \ + typeof(__lookup_##name##_key.keyfld), nodefld) + #define DEFINE_RB_FUNCS(name, type, keyfld, nodefld) \ DEFINE_RB_INSDEL_FUNCS(name, type, keyfld, nodefld) \ DEFINE_RB_LOOKUP_FUNC(name, type, keyfld, nodefld) -- cgit v1.2.3 From a02a946dfe9633d7e0202359836f6b5217a62824 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 19 Jun 2017 12:18:05 +0200 Subject: libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov --- include/linux/ceph/ceph_fs.h | 1 + include/linux/ceph/osd_client.h | 45 +++++++++++++++++++++++++++++++++++++++++ include/linux/ceph/osdmap.h | 1 + include/linux/ceph/rados.h | 6 ++++++ 4 files changed, 53 insertions(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/ceph_fs.h b/include/linux/ceph/ceph_fs.h index ad078ebe25d6..edf5b04b918a 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 62c672bcbb31..c6d96a5f46fd 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 #include #include #include @@ -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 060d059acbf8..fe6d189bdd30 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 5d0018782d50..385db08bb8b2 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 -- cgit v1.2.3 From 278b1d709c6acc6f7d138fed775c76695b068e43 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Wed, 21 Jun 2017 17:27:17 +0200 Subject: libceph: ceph_decode_skip_* helpers Some of these won't be as efficient as they could be (e.g. ceph_decode_skip_set(... 32 ...) could advance by len * sizeof(u32) once instead of advancing by sizeof(u32) len times), but that's fine and not worth a bunch of extra macro code. Replace skip_name_map() with ceph_decode_skip_map as an example. Signed-off-by: Ilya Dryomov --- include/linux/ceph/decode.h | 60 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 60 insertions(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/decode.h b/include/linux/ceph/decode.h index f990f2cc907a..14af9b70d301 100644 --- a/include/linux/ceph/decode.h +++ b/include/linux/ceph/decode.h @@ -132,6 +132,66 @@ bad: return ERR_PTR(-ERANGE); } +/* + * skip helpers + */ +#define ceph_decode_skip_n(p, end, n, bad) \ + do { \ + ceph_decode_need(p, end, n, bad); \ + *p += n; \ + } while (0) + +#define ceph_decode_skip_64(p, end, bad) \ +ceph_decode_skip_n(p, end, sizeof(u64), bad) + +#define ceph_decode_skip_32(p, end, bad) \ +ceph_decode_skip_n(p, end, sizeof(u32), bad) + +#define ceph_decode_skip_16(p, end, bad) \ +ceph_decode_skip_n(p, end, sizeof(u16), bad) + +#define ceph_decode_skip_8(p, end, bad) \ +ceph_decode_skip_n(p, end, sizeof(u8), bad) + +#define ceph_decode_skip_string(p, end, bad) \ + do { \ + u32 len; \ + \ + ceph_decode_32_safe(p, end, len, bad); \ + ceph_decode_skip_n(p, end, len, bad); \ + } while (0) + +#define ceph_decode_skip_set(p, end, type, bad) \ + do { \ + u32 len; \ + \ + ceph_decode_32_safe(p, end, len, bad); \ + while (len--) \ + ceph_decode_skip_##type(p, end, bad); \ + } while (0) + +#define ceph_decode_skip_map(p, end, ktype, vtype, bad) \ + do { \ + u32 len; \ + \ + ceph_decode_32_safe(p, end, len, bad); \ + while (len--) { \ + ceph_decode_skip_##ktype(p, end, bad); \ + ceph_decode_skip_##vtype(p, end, bad); \ + } \ + } while (0) + +#define ceph_decode_skip_map_of_map(p, end, ktype1, ktype2, vtype2, bad) \ + do { \ + u32 len; \ + \ + ceph_decode_32_safe(p, end, len, bad); \ + while (len--) { \ + ceph_decode_skip_##ktype1(p, end, bad); \ + ceph_decode_skip_map(p, end, ktype2, vtype2, bad); \ + } \ + } while (0) + /* * struct ceph_timespec <-> struct timespec */ -- cgit v1.2.3 From 6f428df47dae2c8ea31fd4c0c74a12a8a5ac2d1d Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Wed, 21 Jun 2017 17:27:18 +0200 Subject: libceph: pg_upmap[_items] infrastructure pg_temp and pg_upmap encodings are the same (PG -> array of osds), except for the incremental remove: it's an empty mapping in new_pg_temp for pg_temp and a separate old_pg_upmap set for pg_upmap. (This isn't to allow for empty pg_upmap mappings -- apparently, pg_temp just wasn't looked at as an example for pg_upmap encoding.) Reuse __decode_pg_temp() for decoding pg_upmap and new_pg_upmap. __decode_pg_temp() stores into pg_temp union member, but since pg_upmap union member is identical, reading through pg_upmap later is OK. Signed-off-by: Ilya Dryomov --- include/linux/ceph/osdmap.h | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osdmap.h b/include/linux/ceph/osdmap.h index fe6d189bdd30..c612cff81f5c 100644 --- a/include/linux/ceph/osdmap.h +++ b/include/linux/ceph/osdmap.h @@ -143,10 +143,14 @@ struct ceph_pg_mapping { struct { int len; int osds[]; - } pg_temp; + } pg_temp, pg_upmap; struct { int osd; } primary_temp; + struct { + int len; + int from_to[][2]; + } pg_upmap_items; }; }; @@ -165,6 +169,10 @@ struct ceph_osdmap { struct rb_root pg_temp; struct rb_root primary_temp; + /* remap (post-CRUSH, pre-up) */ + struct rb_root pg_upmap; /* PG := raw set */ + struct rb_root pg_upmap_items; /* from -> to within raw set */ + u32 *osd_primary_affinity; struct rb_root pg_pools; -- cgit v1.2.3 From 0bb05da2ec57163b7a25efef001ed8f52b18b070 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 22 Jun 2017 19:44:06 +0200 Subject: libceph: osd_state is 32 bits wide in luminous Signed-off-by: Ilya Dryomov --- include/linux/ceph/osdmap.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/osdmap.h b/include/linux/ceph/osdmap.h index c612cff81f5c..a0996cb9faed 100644 --- a/include/linux/ceph/osdmap.h +++ b/include/linux/ceph/osdmap.h @@ -162,7 +162,7 @@ struct ceph_osdmap { u32 flags; /* CEPH_OSDMAP_* */ u32 max_osd; /* size of osd_state, _offload, _addr arrays */ - u8 *osd_state; /* CEPH_OSD_* */ + u32 *osd_state; /* CEPH_OSD_* */ u32 *osd_weight; /* 0 = failed, 0x10000 = 100% normal */ struct ceph_entity_addr *osd_addr; @@ -203,7 +203,7 @@ static inline bool ceph_osd_is_down(struct ceph_osdmap *map, int osd) return !ceph_osd_is_up(map, osd); } -extern char *ceph_osdmap_state_str(char *str, int len, int state); +char *ceph_osdmap_state_str(char *str, int len, u32 state); extern u32 ceph_get_primary_affinity(struct ceph_osdmap *map, int osd); static inline struct ceph_entity_addr *ceph_osd_addr(struct ceph_osdmap *map, -- cgit v1.2.3 From 33e9c8dbfbcef8e4cda8e43a445e692ab7e0d8c0 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Mon, 26 Jun 2017 12:05:55 +0200 Subject: libceph: advertise support for NEW_OSDOP_ENCODING and SERVER_LUMINOUS All four SERVER_LUMINOUS feature bits are implemented, switch it on! NEW_OSDOP_ENCODING doesn't mean much for the client (it signifies support for MOSDOp v6) but needs to be enabled in order to get the latest (currently v25) pg_pool_t. Signed-off-by: Ilya Dryomov Acked-by: Sage Weil --- include/linux/ceph/ceph_features.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/ceph_features.h b/include/linux/ceph/ceph_features.h index 78a58770e6e9..f0f6c537b64c 100644 --- a/include/linux/ceph/ceph_features.h +++ b/include/linux/ceph/ceph_features.h @@ -184,6 +184,11 @@ DEFINE_CEPH_FEATURE_DEPRECATED(63, 1, RESERVED_BROKEN, LUMINOUS) // client-facin CEPH_FEATURE_PGPOOL3 | \ CEPH_FEATURE_OSDENC | \ CEPH_FEATURE_CRUSH_TUNABLES | \ + CEPH_FEATURE_SERVER_LUMINOUS | \ + CEPH_FEATURE_RESEND_ON_SPLIT | \ + CEPH_FEATURE_RADOS_BACKOFF | \ + CEPH_FEATURE_OSDMAP_PG_UPMAP | \ + CEPH_FEATURE_CRUSH_CHOOSE_ARGS | \ CEPH_FEATURE_MSG_AUTH | \ CEPH_FEATURE_CRUSH_TUNABLES2 | \ CEPH_FEATURE_REPLY_CREATE_INODE | \ @@ -199,6 +204,7 @@ DEFINE_CEPH_FEATURE_DEPRECATED(63, 1, RESERVED_BROKEN, LUMINOUS) // client-facin CEPH_FEATURE_MSGR_KEEPALIVE2 | \ CEPH_FEATURE_OSD_POOLRESEND | \ CEPH_FEATURE_CRUSH_V4 | \ + CEPH_FEATURE_NEW_OSDOP_ENCODING | \ CEPH_FEATURE_SERVER_JEWEL | \ CEPH_FEATURE_MON_STATEFUL_SUB | \ CEPH_FEATURE_CRUSH_TUNABLES5 | \ -- cgit v1.2.3 From e67ae2b7b23b283e657865b498b151e6a17b919d Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Mon, 10 Jul 2017 13:17:26 +0200 Subject: libceph: fix old style declaration warnings The new macros don't follow the usual style for declarations, which we get a warning for with 'make W=1': In file included from fs/ceph/mds_client.c:16:0: include/linux/ceph/ceph_features.h:74:1: error: 'static' is not at beginning of declaration [-Werror=old-style-declaration] This moves the 'static' keyword to the front of the declaration. Fixes: f179d3ba8cb9 ("libceph: new features macros") Signed-off-by: Arnd Bergmann Signed-off-by: Ilya Dryomov --- include/linux/ceph/ceph_features.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'include/linux/ceph') diff --git a/include/linux/ceph/ceph_features.h b/include/linux/ceph/ceph_features.h index f0f6c537b64c..040dd105c3e7 100644 --- a/include/linux/ceph/ceph_features.h +++ b/include/linux/ceph/ceph_features.h @@ -10,14 +10,14 @@ #define CEPH_FEATURE_INCARNATION_2 (1ull<<57) // CEPH_FEATURE_SERVER_JEWEL #define DEFINE_CEPH_FEATURE(bit, incarnation, name) \ - const static uint64_t CEPH_FEATURE_##name = (1ULL<