ceph: clean up readdir caps reservation

Use a global counter for the minimum number of allocated caps instead of
hard coding a check against readdir_max.  This takes into account multiple
client instances, and avoids examining the superblock mount options when a
cap is dropped.

Signed-off-by: Sage Weil <sage@newdream.net>

4 files changed, 33 insertions, 13 deletions

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index ab9b571dda11..f94b56faba3b 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -128,6 +128,7 @@ static int caps_total_count;        /* total caps allocated */
 static int caps_use_count;          /* in use */
 static int caps_reserve_count;      /* unused, reserved */
 static int caps_avail_count;        /* unused, unreserved */
+static int caps_min_count;          /* keep at least this many (unreserved) */
 
 void __init ceph_caps_init(void)
 {
@@ -149,6 +150,15 @@ void ceph_caps_finalize(void)
         caps_avail_count = 0;
         caps_use_count = 0;
         caps_reserve_count = 0;
+        caps_min_count = 0;
+        spin_unlock(&caps_list_lock);
+}
+
+void ceph_adjust_min_caps(int delta)
+{
+        spin_lock(&caps_list_lock);
+        caps_min_count += delta;
+        BUG_ON(caps_min_count < 0);
         spin_unlock(&caps_list_lock);
 }
 
@@ -265,12 +275,10 @@ static void put_cap(struct ceph_cap *cap,
              caps_reserve_count, caps_avail_count);
         caps_use_count--;
         /*
-         * Keep some preallocated caps around, at least enough to do a
-         * readdir (which needs to preallocate lots of them), to avoid
-         * lots of free/alloc churn.
+         * Keep some preallocated caps around (ceph_min_count), to
+         * avoid lots of free/alloc churn.
          */
-        if (caps_avail_count >= caps_reserve_count +
-            ceph_client(cap->ci->vfs_inode.i_sb)->mount_args->max_readdir) {
+        if (caps_avail_count >= caps_reserve_count + caps_min_count) {
                 caps_total_count--;
                 kmem_cache_free(ceph_cap_cachep, cap);
         } else {
@@ -289,7 +297,8 @@ static void put_cap(struct ceph_cap *cap,
 }
 
 void ceph_reservation_status(struct ceph_client *client,
-                             int *total, int *avail, int *used, int *reserved)
+                             int *total, int *avail, int *used, int *reserved,
+                             int *min)
 {
         if (total)
                 *total = caps_total_count;
@@ -299,6 +308,8 @@ void ceph_reservation_status(struct ceph_client *client,
                 *used = caps_use_count;
         if (reserved)
                 *reserved = caps_reserve_count;
+        if (min)
+                *min = caps_min_count;
 }
 
 /*
diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c
index b58bd9188692..1a47b5c25b5f 100644
--- a/fs/ceph/debugfs.c
+++ b/fs/ceph/debugfs.c
@@ -255,14 +255,15 @@ static int osdc_show(struct seq_file *s, void *pp)
 static int caps_show(struct seq_file *s, void *p)
 {
         struct ceph_client *client = p;
-        int total, avail, used, reserved;
+        int total, avail, used, reserved, min;
 
-        ceph_reservation_status(client, &total, &avail, &used, &reserved);
+        ceph_reservation_status(client, &total, &avail, &used, &reserved, &min);
         seq_printf(s, "total\t\t%d\n"
-                      "avail\t\t%d\n"
-                      "used\t\t%d\n"
-                      "reserved\t%d\n",
-                   total, avail, used, reserved);
+                   "avail\t\t%d\n"
+                   "used\t\t%d\n"
+                   "reserved\t%d\n"
+                   "min\t%d\n",
+                   total, avail, used, reserved, min);
         return 0;
 }
 
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
index 39aaf29a04a0..74953be75f8f 100644
--- a/fs/ceph/super.c
+++ b/fs/ceph/super.c
@@ -578,6 +578,9 @@ static struct ceph_client *ceph_create_client(struct ceph_mount_args *args)
         if (!client->wb_pagevec_pool)
                 goto fail_trunc_wq;
 
+        /* caps */
+        client->min_caps = args->max_readdir;
+        ceph_adjust_min_caps(client->min_caps);
 
         /* subsystems */
         err = ceph_monc_init(&client->monc, client);
@@ -619,6 +622,8 @@ static void ceph_destroy_client(struct ceph_client *client)
         ceph_monc_stop(&client->monc);
         ceph_osdc_stop(&client->osdc);
 
+        ceph_adjust_min_caps(-client->min_caps);
+
         ceph_debugfs_client_cleanup(client);
         destroy_workqueue(client->wb_wq);
         destroy_workqueue(client->pg_inv_wq);
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index 1f3928785e12..3b5faf9980f8 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -129,6 +129,8 @@ struct ceph_client {
 
         int auth_err;
 
+        int min_caps;                  /* min caps i added */
+
         struct ceph_messenger *msgr;   /* messenger instance */
         struct ceph_mon_client monc;
         struct ceph_mds_client mdsc;
@@ -557,11 +559,12 @@ extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
 
 extern void ceph_caps_init(void);
 extern void ceph_caps_finalize(void);
+extern void ceph_adjust_min_caps(int delta);
 extern int ceph_reserve_caps(struct ceph_cap_reservation *ctx, int need);
 extern int ceph_unreserve_caps(struct ceph_cap_reservation *ctx);
 extern void ceph_reservation_status(struct ceph_client *client,
                                     int *total, int *avail, int *used,
-                                    int *reserved);
+                                    int *reserved, int *min);
 
 static inline struct ceph_client *ceph_inode_to_client(struct inode *inode)
 {