Merge tag 'ceph-for-5.5-rc2' of git://github.com/ceph/ceph-client

Pull ceph fixes from Ilya Dryomov:
 "A fix to avoid a corner case when scheduling cap reclaim in batches
  from Xiubo, a patch to add some observability into cap waiters from
  Jeff and a couple of cleanups"

* tag 'ceph-for-5.5-rc2' of git://github.com/ceph/ceph-client:
  ceph: add more debug info when decoding mdsmap
  ceph: switch to global cap helper
  ceph: trigger the reclaim work once there has enough pending caps
  ceph: show tasks waiting on caps in debugfs caps file
  ceph: convert int fields in ceph_mount_options to unsigned int

7 files changed, 85 insertions, 42 deletions

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index f5a38910a82b..9d09bb53c1ab 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -1011,18 +1011,13 @@ static int __ceph_is_single_caps(struct ceph_inode_info *ci)
 	return rb_first(&ci->i_caps) == rb_last(&ci->i_caps);
 }
 
-static int __ceph_is_any_caps(struct ceph_inode_info *ci)
-{
-	return !RB_EMPTY_ROOT(&ci->i_caps);
-}
-
 int ceph_is_any_caps(struct inode *inode)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int ret;
 
 	spin_lock(&ci->i_ceph_lock);
-	ret = __ceph_is_any_caps(ci);
+	ret = __ceph_is_any_real_caps(ci);
 	spin_unlock(&ci->i_ceph_lock);
 
 	return ret;
@@ -1099,15 +1094,16 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release)
 	if (removed)
 		ceph_put_cap(mdsc, cap);
 
-	/* when reconnect denied, we remove session caps forcibly,
-	 * i_wr_ref can be non-zero. If there are ongoing write,
-	 * keep i_snap_realm.
-	 */
-	if (!__ceph_is_any_caps(ci) && ci->i_wr_ref == 0 && ci->i_snap_realm)
-		drop_inode_snap_realm(ci);
+	if (!__ceph_is_any_real_caps(ci)) {
+		/* when reconnect denied, we remove session caps forcibly,
+		 * i_wr_ref can be non-zero. If there are ongoing write,
+		 * keep i_snap_realm.
+		 */
+		if (ci->i_wr_ref == 0 && ci->i_snap_realm)
+			drop_inode_snap_realm(ci);
 
-	if (!__ceph_is_any_real_caps(ci))
 		__cap_delay_cancel(mdsc, ci);
+	}
 }
 
 struct cap_msg_args {
@@ -2764,7 +2760,19 @@ int ceph_get_caps(struct file *filp, int need, int want,
 		if (ret == -EAGAIN)
 			continue;
 		if (!ret) {
+			struct ceph_mds_client *mdsc = fsc->mdsc;
+			struct cap_wait cw;
 			DEFINE_WAIT_FUNC(wait, woken_wake_function);
+
+			cw.ino = inode->i_ino;
+			cw.tgid = current->tgid;
+			cw.need = need;
+			cw.want = want;
+
+			spin_lock(&mdsc->caps_list_lock);
+			list_add(&cw.list, &mdsc->cap_wait_list);
+			spin_unlock(&mdsc->caps_list_lock);
+
 			add_wait_queue(&ci->i_cap_wq, &wait);
 
 			flags |= NON_BLOCKING;
@@ -2778,6 +2786,11 @@ int ceph_get_caps(struct file *filp, int need, int want,
 			}
 
 			remove_wait_queue(&ci->i_cap_wq, &wait);
+
+			spin_lock(&mdsc->caps_list_lock);
+			list_del(&cw.list);
+			spin_unlock(&mdsc->caps_list_lock);
+
 			if (ret == -EAGAIN)
 				continue;
 		}
@@ -2928,7 +2941,7 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
 				ci->i_head_snapc = NULL;
 			}
 			/* see comment in __ceph_remove_cap() */
-			if (!__ceph_is_any_caps(ci) && ci->i_snap_realm)
+			if (!__ceph_is_any_real_caps(ci) && ci->i_snap_realm)
 				drop_inode_snap_realm(ci);
 		}
 	spin_unlock(&ci->i_ceph_lock);
diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c
index facb387c2735..c281f32b54f7 100644
--- a/fs/ceph/debugfs.c
+++ b/fs/ceph/debugfs.c
@@ -139,6 +139,7 @@ static int caps_show(struct seq_file *s, void *p)
 	struct ceph_fs_client *fsc = s->private;
 	struct ceph_mds_client *mdsc = fsc->mdsc;
 	int total, avail, used, reserved, min, i;
+	struct cap_wait	*cw;
 
 	ceph_reservation_status(fsc, &total, &avail, &used, &reserved, &min);
 	seq_printf(s, "total\t\t%d\n"
@@ -166,6 +167,18 @@ static int caps_show(struct seq_file *s, void *p)
 	}
 	mutex_unlock(&mdsc->mutex);
 
+	seq_printf(s, "\n\nWaiters:\n--------\n");
+	seq_printf(s, "tgid         ino                need             want\n");
+	seq_printf(s, "-----------------------------------------------------\n");
+
+	spin_lock(&mdsc->caps_list_lock);
+	list_for_each_entry(cw, &mdsc->cap_wait_list, list) {
+		seq_printf(s, "%-13d0x%-17lx%-17s%-17s\n", cw->tgid, cw->ino,
+				ceph_cap_string(cw->need),
+				ceph_cap_string(cw->want));
+	}
+	spin_unlock(&mdsc->caps_list_lock);
+
 	return 0;
 }
 
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 068b029cf073..374db1bd57d1 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -2015,7 +2015,7 @@ void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr)
 	if (!nr)
 		return;
 	val = atomic_add_return(nr, &mdsc->cap_reclaim_pending);
-	if (!(val % CEPH_CAPS_PER_RELEASE)) {
+	if ((val % CEPH_CAPS_PER_RELEASE) < nr) {
 		atomic_set(&mdsc->cap_reclaim_pending, 0);
 		ceph_queue_cap_reclaim_work(mdsc);
 	}
@@ -2032,12 +2032,13 @@ int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
 	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
 	struct ceph_mount_options *opt = req->r_mdsc->fsc->mount_options;
 	size_t size = sizeof(struct ceph_mds_reply_dir_entry);
-	int order, num_entries;
+	unsigned int num_entries;
+	int order;
 
 	spin_lock(&ci->i_ceph_lock);
 	num_entries = ci->i_files + ci->i_subdirs;
 	spin_unlock(&ci->i_ceph_lock);
-	num_entries = max(num_entries, 1);
+	num_entries = max(num_entries, 1U);
 	num_entries = min(num_entries, opt->max_readdir);
 
 	order = get_order(size * num_entries);
@@ -4168,6 +4169,7 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc)
 	INIT_DELAYED_WORK(&mdsc->delayed_work, delayed_work);
 	mdsc->last_renew_caps = jiffies;
 	INIT_LIST_HEAD(&mdsc->cap_delay_list);
+	INIT_LIST_HEAD(&mdsc->cap_wait_list);
 	spin_lock_init(&mdsc->cap_delay_lock);
 	INIT_LIST_HEAD(&mdsc->snap_flush_list);
 	spin_lock_init(&mdsc->snap_flush_lock);
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index 5cd131b41d84..14c7e8c49970 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -340,6 +340,14 @@ struct ceph_quotarealm_inode {
 	struct inode *inode;
 };
 
+struct cap_wait {
+	struct list_head	list;
+	unsigned long		ino;
+	pid_t			tgid;
+	int			need;
+	int			want;
+};
+
 /*
  * mds client state
  */
@@ -416,6 +424,7 @@ struct ceph_mds_client {
 	spinlock_t	caps_list_lock;
 	struct		list_head caps_list; /* unused (reserved or
 						unreserved) */
+	struct		list_head cap_wait_list;
 	int		caps_total_count;    /* total caps allocated */
 	int		caps_use_count;      /* in use */
 	int		caps_use_max;	     /* max used caps */
diff --git a/fs/ceph/mdsmap.c b/fs/ceph/mdsmap.c
index aeec1d6e3769..471bac335fae 100644
--- a/fs/ceph/mdsmap.c
+++ b/fs/ceph/mdsmap.c
@@ -158,6 +158,7 @@ struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end)
 		void *pexport_targets = NULL;
 		struct ceph_timespec laggy_since;
 		struct ceph_mds_info *info;
+		bool laggy;
 
 		ceph_decode_need(p, end, sizeof(u64) + 1, bad);
 		global_id = ceph_decode_64(p);
@@ -190,6 +191,7 @@ struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end)
 		if (err)
 			goto corrupt;
 		ceph_decode_copy(p, &laggy_since, sizeof(laggy_since));
+		laggy = laggy_since.tv_sec != 0 || laggy_since.tv_nsec != 0;
 		*p += sizeof(u32);
 		ceph_decode_32_safe(p, end, namelen, bad);
 		*p += namelen;
@@ -207,10 +209,11 @@ struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end)
 			*p = info_end;
 		}
 
-		dout("mdsmap_decode %d/%d %lld mds%d.%d %s %s\n",
+		dout("mdsmap_decode %d/%d %lld mds%d.%d %s %s%s\n",
 		     i+1, n, global_id, mds, inc,
 		     ceph_pr_addr(&addr),
-		     ceph_mds_state_name(state));
+		     ceph_mds_state_name(state),
+		     laggy ? "(laggy)" : "");
 
 		if (mds < 0 || state <= 0)
 			continue;
@@ -230,8 +233,7 @@ struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end)
 		info->global_id = global_id;
 		info->state = state;
 		info->addr = addr;
-		info->laggy = (laggy_since.tv_sec != 0 ||
-			       laggy_since.tv_nsec != 0);
+		info->laggy = laggy;
 		info->num_export_targets = num_export_targets;
 		if (num_export_targets) {
 			info->export_targets = kcalloc(num_export_targets,
@@ -355,6 +357,8 @@ struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end)
 		m->m_damaged = false;
 	}
 bad_ext:
+	dout("mdsmap_decode m_enabled: %d, m_damaged: %d, m_num_laggy: %d\n",
+	     !!m->m_enabled, !!m->m_damaged, m->m_num_laggy);
 	*p = end;
 	dout("mdsmap_decode success epoch %u\n", m->m_epoch);
 	return m;
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
index 9c9a7c68eea3..29a795f975df 100644
--- a/fs/ceph/super.c
+++ b/fs/ceph/super.c
@@ -172,10 +172,10 @@ static const struct fs_parameter_enum ceph_mount_param_enums[] = {
 static const struct fs_parameter_spec ceph_mount_param_specs[] = {
 	fsparam_flag_no ("acl",				Opt_acl),
 	fsparam_flag_no ("asyncreaddir",		Opt_asyncreaddir),
-	fsparam_u32	("caps_max",			Opt_caps_max),
+	fsparam_s32	("caps_max",			Opt_caps_max),
 	fsparam_u32	("caps_wanted_delay_max",	Opt_caps_wanted_delay_max),
 	fsparam_u32	("caps_wanted_delay_min",	Opt_caps_wanted_delay_min),
-	fsparam_s32	("write_congestion_kb",		Opt_congestion_kb),
+	fsparam_u32	("write_congestion_kb",		Opt_congestion_kb),
 	fsparam_flag_no ("copyfrom",			Opt_copyfrom),
 	fsparam_flag_no ("dcache",			Opt_dcache),
 	fsparam_flag_no ("dirstat",			Opt_dirstat),
@@ -187,8 +187,8 @@ static const struct fs_parameter_spec ceph_mount_param_specs[] = {
 	fsparam_flag_no ("quotadf",			Opt_quotadf),
 	fsparam_u32	("rasize",			Opt_rasize),
 	fsparam_flag_no ("rbytes",			Opt_rbytes),
-	fsparam_s32	("readdir_max_bytes",		Opt_readdir_max_bytes),
-	fsparam_s32	("readdir_max_entries",		Opt_readdir_max_entries),
+	fsparam_u32	("readdir_max_bytes",		Opt_readdir_max_bytes),
+	fsparam_u32	("readdir_max_entries",		Opt_readdir_max_entries),
 	fsparam_enum	("recover_session",		Opt_recover_session),
 	fsparam_flag_no ("require_active_mds",		Opt_require_active_mds),
 	fsparam_u32	("rsize",			Opt_rsize),
@@ -328,7 +328,9 @@ static int ceph_parse_mount_param(struct fs_context *fc,
 		fsopt->caps_wanted_delay_max = result.uint_32;
 		break;
 	case Opt_caps_max:
-		fsopt->caps_max = result.uint_32;
+		if (result.int_32 < 0)
+			goto out_of_range;
+		fsopt->caps_max = result.int_32;
 		break;
 	case Opt_readdir_max_entries:
 		if (result.uint_32 < 1)
@@ -547,25 +549,25 @@ static int ceph_show_options(struct seq_file *m, struct dentry *root)
 		seq_show_option(m, "recover_session", "clean");
 
 	if (fsopt->wsize != CEPH_MAX_WRITE_SIZE)
-		seq_printf(m, ",wsize=%d", fsopt->wsize);
+		seq_printf(m, ",wsize=%u", fsopt->wsize);
 	if (fsopt->rsize != CEPH_MAX_READ_SIZE)
-		seq_printf(m, ",rsize=%d", fsopt->rsize);
+		seq_printf(m, ",rsize=%u", fsopt->rsize);
 	if (fsopt->rasize != CEPH_RASIZE_DEFAULT)
-		seq_printf(m, ",rasize=%d", fsopt->rasize);
+		seq_printf(m, ",rasize=%u", fsopt->rasize);
 	if (fsopt->congestion_kb != default_congestion_kb())
-		seq_printf(m, ",write_congestion_kb=%d", fsopt->congestion_kb);
+		seq_printf(m, ",write_congestion_kb=%u", fsopt->congestion_kb);
 	if (fsopt->caps_max)
 		seq_printf(m, ",caps_max=%d", fsopt->caps_max);
 	if (fsopt->caps_wanted_delay_min != CEPH_CAPS_WANTED_DELAY_MIN_DEFAULT)
-		seq_printf(m, ",caps_wanted_delay_min=%d",
+		seq_printf(m, ",caps_wanted_delay_min=%u",
 			 fsopt->caps_wanted_delay_min);
 	if (fsopt->caps_wanted_delay_max != CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT)
-		seq_printf(m, ",caps_wanted_delay_max=%d",
+		seq_printf(m, ",caps_wanted_delay_max=%u",
 			   fsopt->caps_wanted_delay_max);
 	if (fsopt->max_readdir != CEPH_MAX_READDIR_DEFAULT)
-		seq_printf(m, ",readdir_max_entries=%d", fsopt->max_readdir);
+		seq_printf(m, ",readdir_max_entries=%u", fsopt->max_readdir);
 	if (fsopt->max_readdir_bytes != CEPH_MAX_READDIR_BYTES_DEFAULT)
-		seq_printf(m, ",readdir_max_bytes=%d", fsopt->max_readdir_bytes);
+		seq_printf(m, ",readdir_max_bytes=%u", fsopt->max_readdir_bytes);
 	if (strcmp(fsopt->snapdir_name, CEPH_SNAPDIRNAME_DEFAULT))
 		seq_show_option(m, "snapdirname", fsopt->snapdir_name);
 
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index f0f9cb7447ac..3bf1a01cd536 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -73,16 +73,16 @@
 #define CEPH_CAPS_WANTED_DELAY_MAX_DEFAULT     60  /* cap release delay */
 
 struct ceph_mount_options {
-	int flags;
+	unsigned int flags;
 
-	int wsize;            /* max write size */
-	int rsize;            /* max read size */
-	int rasize;           /* max readahead */
-	int congestion_kb;    /* max writeback in flight */
-	int caps_wanted_delay_min, caps_wanted_delay_max;
+	unsigned int wsize;            /* max write size */
+	unsigned int rsize;            /* max read size */
+	unsigned int rasize;           /* max readahead */
+	unsigned int congestion_kb;    /* max writeback in flight */
+	unsigned int caps_wanted_delay_min, caps_wanted_delay_max;
 	int caps_max;
-	int max_readdir;       /* max readdir result (entires) */
-	int max_readdir_bytes; /* max readdir result (bytes) */
+	unsigned int max_readdir;       /* max readdir result (entries) */
+	unsigned int max_readdir_bytes; /* max readdir result (bytes) */
 
 	/*
 	 * everything above this point can be memcmp'd; everything below