2 files changed, 140 insertions, 71 deletions

diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c
index 0160d7d09a1e..89469592076c 100644
--- a/net/ceph/osd_client.c
+++ b/net/ceph/osd_client.c
@@ -1276,9 +1276,9 @@ static bool target_should_be_paused(struct ceph_osd_client *osdc,
 				    const struct ceph_osd_request_target *t,
 				    struct ceph_pg_pool_info *pi)
 {
-	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) ||
+	bool pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD);
+	bool pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) ||
+		       ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
 		       __pool_full(pi);
 
 	WARN_ON(pi->id != t->base_oloc.pool);
@@ -1303,8 +1303,7 @@ static enum calc_target_result calc_target(struct ceph_osd_client *osdc,
 	bool force_resend = false;
 	bool need_check_tiering = false;
 	bool need_resend = false;
-	bool sort_bitwise = ceph_osdmap_flag(osdc->osdmap,
-					     CEPH_OSDMAP_SORTBITWISE);
+	bool sort_bitwise = ceph_osdmap_flag(osdc, CEPH_OSDMAP_SORTBITWISE);
 	enum calc_target_result ct_res;
 	int ret;
 
@@ -1540,9 +1539,9 @@ static void encode_request(struct ceph_osd_request *req, struct ceph_msg *msg)
 	 */
 	msg->hdr.data_off = cpu_to_le16(req->r_data_offset);
 
-	dout("%s req %p oid %*pE oid_len %d front %zu data %u\n", __func__,
-	     req, req->r_t.target_oid.name_len, req->r_t.target_oid.name,
-	     req->r_t.target_oid.name_len, msg->front.iov_len, data_len);
+	dout("%s req %p oid %s oid_len %d front %zu data %u\n", __func__,
+	     req, req->r_t.target_oid.name, req->r_t.target_oid.name_len,
+	     msg->front.iov_len, data_len);
 }
 
 /*
@@ -1590,9 +1589,9 @@ static void maybe_request_map(struct ceph_osd_client *osdc)
 	verify_osdc_locked(osdc);
 	WARN_ON(!osdc->osdmap->epoch);
 
-	if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL) ||
-	    ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD) ||
-	    ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSEWR)) {
+	if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
+	    ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD) ||
+	    ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR)) {
 		dout("%s osdc %p continuous\n", __func__, osdc);
 		continuous = true;
 	} else {
@@ -1629,19 +1628,19 @@ again:
 	}
 
 	if ((req->r_flags & CEPH_OSD_FLAG_WRITE) &&
-	    ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSEWR)) {
+	    ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR)) {
 		dout("req %p pausewr\n", req);
 		req->r_t.paused = true;
 		maybe_request_map(osdc);
 	} else if ((req->r_flags & CEPH_OSD_FLAG_READ) &&
-		   ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD)) {
+		   ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD)) {
 		dout("req %p pauserd\n", req);
 		req->r_t.paused = true;
 		maybe_request_map(osdc);
 	} else if ((req->r_flags & CEPH_OSD_FLAG_WRITE) &&
 		   !(req->r_flags & (CEPH_OSD_FLAG_FULL_TRY |
 				     CEPH_OSD_FLAG_FULL_FORCE)) &&
-		   (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL) ||
+		   (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
 		    pool_full(osdc, req->r_t.base_oloc.pool))) {
 		dout("req %p full/pool_full\n", req);
 		pr_warn_ratelimited("FULL or reached pool quota\n");
@@ -2280,7 +2279,7 @@ static void send_linger_ping(struct ceph_osd_linger_request *lreq)
 	struct ceph_osd_request *req = lreq->ping_req;
 	struct ceph_osd_req_op *op = &req->r_ops[0];
 
-	if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD)) {
+	if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD)) {
 		dout("%s PAUSERD\n", __func__);
 		return;
 	}
@@ -2893,6 +2892,9 @@ static void handle_reply(struct ceph_osd *osd, struct ceph_msg *msg)
 			dout("req %p tid %llu cb\n", req, req->r_tid);
 			__complete_request(req);
 		}
+		if (m.flags & CEPH_OSD_FLAG_ONDISK)
+			complete_all(&req->r_safe_completion);
+		ceph_osdc_put_request(req);
 	} else {
 		if (req->r_unsafe_callback) {
 			dout("req %p tid %llu unsafe-cb\n", req, req->r_tid);
@@ -2901,10 +2903,7 @@ static void handle_reply(struct ceph_osd *osd, struct ceph_msg *msg)
 			WARN_ON(1);
 		}
 	}
-	if (m.flags & CEPH_OSD_FLAG_ONDISK)
-		complete_all(&req->r_safe_completion);
 
-	ceph_osdc_put_request(req);
 	return;
 
 fail_request:
@@ -3050,7 +3049,7 @@ static int handle_one_map(struct ceph_osd_client *osdc,
 	bool skipped_map = false;
 	bool was_full;
 
-	was_full = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL);
+	was_full = ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL);
 	set_pool_was_full(osdc);
 
 	if (incremental)
@@ -3088,7 +3087,7 @@ static int handle_one_map(struct ceph_osd_client *osdc,
 		osdc->osdmap = newmap;
 	}
 
-	was_full &= !ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL);
+	was_full &= !ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL);
 	scan_requests(&osdc->homeless_osd, skipped_map, was_full, true,
 		      need_resend, need_resend_linger);
 
@@ -3174,9 +3173,9 @@ void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg)
 	if (ceph_check_fsid(osdc->client, &fsid) < 0)
 		goto bad;
 
-	was_pauserd = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD);
-	was_pausewr = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSEWR) ||
-		      ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL) ||
+	was_pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD);
+	was_pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) ||
+		      ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
 		      have_pool_full(osdc);
 
 	/* incremental maps */
@@ -3238,9 +3237,9 @@ done:
 	 * we find out when we are no longer full and stop returning
 	 * ENOSPC.
 	 */
-	pauserd = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD);
-	pausewr = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSEWR) ||
-		  ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL) ||
+	pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD);
+	pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) ||
+		  ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
 		  have_pool_full(osdc);
 	if (was_pauserd || was_pausewr || pauserd || pausewr)
 		maybe_request_map(osdc);
diff --git a/net/ceph/osdmap.c b/net/ceph/osdmap.c
index cde52e94732f..7e480bf75bcf 100644
--- a/net/ceph/osdmap.c
+++ b/net/ceph/osdmap.c
@@ -1261,6 +1261,115 @@ struct ceph_osdmap *ceph_osdmap_decode(void **p, void *end)
 }
 
 /*
+ * Encoding order is (new_up_client, new_state, new_weight).  Need to
+ * apply in the (new_weight, new_state, new_up_client) order, because
+ * an incremental map may look like e.g.
+ *
+ *     new_up_client: { osd=6, addr=... } # set osd_state and addr
+ *     new_state: { osd=6, xorstate=EXISTS } # clear osd_state
+ */
+static int decode_new_up_state_weight(void **p, void *end,
+				      struct ceph_osdmap *map)
+{
+	void *new_up_client;
+	void *new_state;
+	void *new_weight_end;
+	u32 len;
+
+	new_up_client = *p;
+	ceph_decode_32_safe(p, end, len, e_inval);
+	len *= sizeof(u32) + sizeof(struct ceph_entity_addr);
+	ceph_decode_need(p, end, len, e_inval);
+	*p += len;
+
+	new_state = *p;
+	ceph_decode_32_safe(p, end, len, e_inval);
+	len *= sizeof(u32) + sizeof(u8);
+	ceph_decode_need(p, end, len, e_inval);
+	*p += len;
+
+	/* new_weight */
+	ceph_decode_32_safe(p, end, len, e_inval);
+	while (len--) {
+		s32 osd;
+		u32 w;
+
+		ceph_decode_need(p, end, 2*sizeof(u32), e_inval);
+		osd = ceph_decode_32(p);
+		w = ceph_decode_32(p);
+		BUG_ON(osd >= map->max_osd);
+		pr_info("osd%d weight 0x%x %s\n", osd, w,
+		     w == CEPH_OSD_IN ? "(in)" :
+		     (w == CEPH_OSD_OUT ? "(out)" : ""));
+		map->osd_weight[osd] = w;
+
+		/*
+		 * If we are marking in, set the EXISTS, and clear the
+		 * AUTOOUT and NEW bits.
+		 */
+		if (w) {
+			map->osd_state[osd] |= CEPH_OSD_EXISTS;
+			map->osd_state[osd] &= ~(CEPH_OSD_AUTOOUT |
+						 CEPH_OSD_NEW);
+		}
+	}
+	new_weight_end = *p;
+
+	/* new_state (up/down) */
+	*p = new_state;
+	len = ceph_decode_32(p);
+	while (len--) {
+		s32 osd;
+		u8 xorstate;
+		int ret;
+
+		osd = ceph_decode_32(p);
+		xorstate = ceph_decode_8(p);
+		if (xorstate == 0)
+			xorstate = CEPH_OSD_UP;
+		BUG_ON(osd >= map->max_osd);
+		if ((map->osd_state[osd] & CEPH_OSD_UP) &&
+		    (xorstate & CEPH_OSD_UP))
+			pr_info("osd%d down\n", osd);
+		if ((map->osd_state[osd] & CEPH_OSD_EXISTS) &&
+		    (xorstate & CEPH_OSD_EXISTS)) {
+			pr_info("osd%d does not exist\n", osd);
+			map->osd_weight[osd] = CEPH_OSD_IN;
+			ret = set_primary_affinity(map, osd,
+						   CEPH_OSD_DEFAULT_PRIMARY_AFFINITY);
+			if (ret)
+				return ret;
+			memset(map->osd_addr + osd, 0, sizeof(*map->osd_addr));
+			map->osd_state[osd] = 0;
+		} else {
+			map->osd_state[osd] ^= xorstate;
+		}
+	}
+
+	/* new_up_client */
+	*p = new_up_client;
+	len = ceph_decode_32(p);
+	while (len--) {
+		s32 osd;
+		struct ceph_entity_addr addr;
+
+		osd = ceph_decode_32(p);
+		ceph_decode_copy(p, &addr, sizeof(addr));
+		ceph_decode_addr(&addr);
+		BUG_ON(osd >= map->max_osd);
+		pr_info("osd%d up\n", osd);
+		map->osd_state[osd] |= CEPH_OSD_EXISTS | CEPH_OSD_UP;
+		map->osd_addr[osd] = addr;
+	}
+
+	*p = new_weight_end;
+	return 0;
+
+e_inval:
+	return -EINVAL;
+}
+
+/*
  * decode and apply an incremental map update.
  */
 struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
@@ -1358,49 +1467,10 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 			__remove_pg_pool(&map->pg_pools, pi);
 	}
 
-	/* new_up */
-	ceph_decode_32_safe(p, end, len, e_inval);
-	while (len--) {
-		u32 osd;
-		struct ceph_entity_addr addr;
-		ceph_decode_32_safe(p, end, osd, e_inval);
-		ceph_decode_copy_safe(p, end, &addr, sizeof(addr), e_inval);
-		ceph_decode_addr(&addr);
-		pr_info("osd%d up\n", osd);
-		BUG_ON(osd >= map->max_osd);
-		map->osd_state[osd] |= CEPH_OSD_UP | CEPH_OSD_EXISTS;
-		map->osd_addr[osd] = addr;
-	}
-
-	/* new_state */
-	ceph_decode_32_safe(p, end, len, e_inval);
-	while (len--) {
-		u32 osd;
-		u8 xorstate;
-		ceph_decode_32_safe(p, end, osd, e_inval);
-		xorstate = **(u8 **)p;
-		(*p)++;  /* clean flag */
-		if (xorstate == 0)
-			xorstate = CEPH_OSD_UP;
-		if (xorstate & CEPH_OSD_UP)
-			pr_info("osd%d down\n", osd);
-		if (osd < map->max_osd)
-			map->osd_state[osd] ^= xorstate;
-	}
-
-	/* new_weight */
-	ceph_decode_32_safe(p, end, len, e_inval);
-	while (len--) {
-		u32 osd, off;
-		ceph_decode_need(p, end, sizeof(u32)*2, e_inval);
-		osd = ceph_decode_32(p);
-		off = ceph_decode_32(p);
-		pr_info("osd%d weight 0x%x %s\n", osd, off,
-		     off == CEPH_OSD_IN ? "(in)" :
-		     (off == CEPH_OSD_OUT ? "(out)" : ""));
-		if (osd < map->max_osd)
-			map->osd_weight[osd] = off;
-	}
+	/* new_up_client, new_state, new_weight */
+	err = decode_new_up_state_weight(p, end, map);
+	if (err)
+		goto bad;
 
 	/* new_pg_temp */
 	err = decode_new_pg_temp(p, end, map);
@@ -1778,8 +1848,8 @@ int ceph_object_locator_to_pg(struct ceph_osdmap *osdmap,
 	raw_pgid->seed = ceph_str_hash(pi->object_hash, oid->name,
 				       oid->name_len);
 
-	dout("%s %*pE -> raw_pgid %llu.%x\n", __func__, oid->name_len,
-	     oid->name, raw_pgid->pool, raw_pgid->seed);
+	dout("%s %s -> raw_pgid %llu.%x\n", __func__, oid->name,
+	     raw_pgid->pool, raw_pgid->seed);
 	return 0;
 }
 EXPORT_SYMBOL(ceph_object_locator_to_pg);