From 134129520beaf3339482c557361ea0bde709cf36 Mon Sep 17 00:00:00 2001 From: Alexander Aring Date: Tue, 19 Nov 2024 15:56:44 -0500 Subject: dlm: fix removal of rsb struct that is master and dir record An rsb struct was not being removed in the case where it was both the master and the dir record. This case (master and dir node) was missed in the condition for doing add_scan() from deactivate_rsb(). Fixing this triggers a related WARN_ON that needs to be fixed, and requires adjusting where two del_scan() calls are made. Fixes: c217adfc8caa ("dlm: fix add_scan and del_scan usage") Signed-off-by: Alexander Aring Signed-off-by: David Teigland --- fs/dlm/lock.c | 46 ++++++++++++++++++++++++++++++---------------- 1 file changed, 30 insertions(+), 16 deletions(-) diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c index fc1d710166e9..c8ff88f1cdcf 100644 --- a/fs/dlm/lock.c +++ b/fs/dlm/lock.c @@ -824,9 +824,12 @@ static int find_rsb_dir(struct dlm_ls *ls, const void *name, int len, r->res_first_lkid = 0; } - /* A dir record will not be on the scan list. */ - if (r->res_dir_nodeid != our_nodeid) - del_scan(ls, r); + /* we always deactivate scan timer for the rsb, when + * we move it out of the inactive state as rsb state + * can be changed and scan timers are only for inactive + * rsbs. + */ + del_scan(ls, r); list_move(&r->res_slow_list, &ls->ls_slow_active); rsb_clear_flag(r, RSB_INACTIVE); kref_init(&r->res_ref); /* ref is now used in active state */ @@ -989,10 +992,10 @@ static int find_rsb_nodir(struct dlm_ls *ls, const void *name, int len, r->res_nodeid = 0; } + del_scan(ls, r); list_move(&r->res_slow_list, &ls->ls_slow_active); rsb_clear_flag(r, RSB_INACTIVE); kref_init(&r->res_ref); - del_scan(ls, r); write_unlock_bh(&ls->ls_rsbtbl_lock); goto out; @@ -1337,9 +1340,13 @@ static int _dlm_master_lookup(struct dlm_ls *ls, int from_nodeid, const char *na __dlm_master_lookup(ls, r, our_nodeid, from_nodeid, true, flags, r_nodeid, result); - /* A dir record rsb should never be on scan list. */ - /* Try to fix this with del_scan? */ - WARN_ON(!list_empty(&r->res_scan_list)); + /* A dir record rsb should never be on scan list. + * Except when we are the dir and master node. + * This function should only be called by the dir + * node. + */ + WARN_ON(!list_empty(&r->res_scan_list) && + r->res_master_nodeid != our_nodeid); write_unlock_bh(&ls->ls_rsbtbl_lock); @@ -1430,16 +1437,23 @@ static void deactivate_rsb(struct kref *kref) list_move(&r->res_slow_list, &ls->ls_slow_inactive); /* - * When the rsb becomes unused: - * - If it's not a dir record for a remote master rsb, - * then it is put on the scan list to be freed. - * - If it's a dir record for a remote master rsb, - * then it is kept in the inactive state until - * receive_remove() from the master node. + * When the rsb becomes unused, there are two possibilities: + * 1. Leave the inactive rsb in place (don't remove it). + * 2. Add it to the scan list to be removed. + * + * 1 is done when the rsb is acting as the dir record + * for a remotely mastered rsb. The rsb must be left + * in place as an inactive rsb to act as the dir record. + * + * 2 is done when a) the rsb is not the master and not the + * dir record, b) when the rsb is both the master and the + * dir record, c) when the rsb is master but not dir record. + * + * (If no directory is used, the rsb can always be removed.) */ - if (!dlm_no_directory(ls) && - (r->res_master_nodeid != our_nodeid) && - (dlm_dir_nodeid(r) != our_nodeid)) + if (dlm_no_directory(ls) || + (r->res_master_nodeid == our_nodeid || + dlm_dir_nodeid(r) != our_nodeid)) add_scan(ls, r); if (r->res_lvbptr) { -- cgit v1.2.3 From 57cdd1a1cf1464199678f9338049b63fb5d5b41c Mon Sep 17 00:00:00 2001 From: Alexander Aring Date: Mon, 2 Dec 2024 10:26:37 -0500 Subject: dlm: fix srcu_read_lock() return type to int The return type of srcu_read_lock() is int and not bool. Whereas we using the ret variable only to evaluate a bool type of dlm_lowcomms_con_has_addr() to check if an address is already being set. Fixes: 6f0b0b5d7ae7 ("fs: dlm: remove dlm_node_addrs lookup list") Signed-off-by: Alexander Aring Signed-off-by: David Teigland --- fs/dlm/lowcomms.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c index df40c3fd1070..d28141829c05 100644 --- a/fs/dlm/lowcomms.c +++ b/fs/dlm/lowcomms.c @@ -462,7 +462,8 @@ static bool dlm_lowcomms_con_has_addr(const struct connection *con, int dlm_lowcomms_addr(int nodeid, struct sockaddr_storage *addr) { struct connection *con; - bool ret, idx; + bool ret; + int idx; idx = srcu_read_lock(&connections_srcu); con = nodeid2con(nodeid, GFP_NOFS); -- cgit v1.2.3 From 6784ed98fde5b7538fff6b329b686b119ca23d8b Mon Sep 17 00:00:00 2001 From: Alexander Aring Date: Mon, 2 Dec 2024 10:26:41 -0500 Subject: dlm: return -ENOENT if no comm was found Currently if no comm can be found dlm_comm_seq() returns -EEXIST which means entry already exists for a lookup it makes no sense to return -EEXIST. We change it to -ENOENT. There is no user that will evaluate the return value on a specific value so this should be fine. Signed-off-by: Alexander Aring Signed-off-by: David Teigland --- fs/dlm/config.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/dlm/config.c b/fs/dlm/config.c index b2f21aa00719..cf9ba6fd7a28 100644 --- a/fs/dlm/config.c +++ b/fs/dlm/config.c @@ -935,7 +935,7 @@ int dlm_comm_seq(int nodeid, uint32_t *seq, bool locked) mutex_unlock(&clusters_root.subsys.su_mutex); } if (!cm) - return -EEXIST; + return -ENOENT; *seq = cm->seq; put_comm(cm); -- cgit v1.2.3