ksmbd: don't hold ci->m_lock while waiting for a lease break ack

When a cifs.ko client caches a read-handle (RH) lease via deferred close
and a conflicting open arrives, ksmbd breaks the lease and waits for the
acknowledgment in wait_for_break_ack() for up to OPLOCK_WAIT_TIME (35s).
__smb_break_all_levII_oplock() runs that wait while holding ci->m_lock
for read.

cifs.ko reacts to a handle-lease break by closing the deferred handle
rather than sending a lease break acknowledgment. That close path
(close_id_del_oplock() -> opinfo_del()) takes ci->m_lock for write and
is exactly what would wake the waiter, but it blocks on the read lock
held by the waiting thread. The break is then resolved only by the 35s
timeout, so xfstests generic/001 takes ~78s with leases enabled versus
~4s with oplocks only.

Collect the target opinfos (each pinned with a reference) while holding
ci->m_lock, then break them after releasing it, matching how
smb_grant_oplock() already breaks a conflicting lease using only a
reference.  The reference keeps the opinfo (and its conn and lease)
alive across the unlocked window, and a close racing the break is
handled by the existing OPLOCK_CLOSING state check. Apply the same fix
to the parent lease break paths.

Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>

diff --git a/fs/smb/server/oplock.c b/fs/smb/server/oplock.c
index 31dd9f3479b2158b6748ac6384b9f23ac7531a6b..64cc8c5805e3aa908a06d6a3b2ed27d9cc4f6a6d 100644 (file)
--- a/fs/smb/server/oplock.c
+++ b/fs/smb/server/oplock.c
@@ -1180,6 +1180,36 @@ again:
        return err;
 }
 
+struct oplock_break_entry {
+       struct list_head        list;
+       struct oplock_info      *opinfo;
+};
+
+static int oplock_break_add(struct list_head *head, struct oplock_info *opinfo)
+{
+       struct oplock_break_entry *ent;
+
+       ent = kmalloc_obj(struct oplock_break_entry, KSMBD_DEFAULT_GFP);
+       if (!ent)
+               return -ENOMEM;
+
+       ent->opinfo = opinfo;
+       list_add_tail(&ent->list, head);
+       return 0;
+}
+
+static void oplock_break_drain_none(struct list_head *head)
+{
+       struct oplock_break_entry *ent, *tmp;
+
+       list_for_each_entry_safe(ent, tmp, head, list) {
+               oplock_break(ent->opinfo, SMB2_OPLOCK_LEVEL_NONE, NULL, false);
+               list_del(&ent->list);
+               opinfo_put(ent->opinfo);
+               kfree(ent);
+       }
+}
+
 void destroy_lease_table(struct ksmbd_conn *conn)
 {
        struct lease_table *lb, *lbtmp;
@@ -1289,6 +1319,7 @@ void smb_send_parent_lease_break_noti(struct ksmbd_file *fp,
 {
        struct oplock_info *opinfo;
        struct ksmbd_inode *p_ci = NULL;
+       LIST_HEAD(brk_list);
 
        if (lctx->version != 2)
                return;
@@ -1314,12 +1345,14 @@ void smb_send_parent_lease_break_noti(struct ksmbd_file *fp,
                                continue;
                        }
 
-                       oplock_break(opinfo, SMB2_OPLOCK_LEVEL_NONE, NULL, false);
-                       opinfo_put(opinfo);
+                       if (oplock_break_add(&brk_list, opinfo))
+                               opinfo_put(opinfo);
                }
        }
        up_read(&p_ci->m_lock);
 
+       oplock_break_drain_none(&brk_list);
+
        ksmbd_inode_put(p_ci);
 }
 
@@ -1327,6 +1360,7 @@ void smb_lazy_parent_lease_break_close(struct ksmbd_file *fp)
 {
        struct oplock_info *opinfo;
        struct ksmbd_inode *p_ci = NULL;
+       LIST_HEAD(brk_list);
 
        rcu_read_lock();
        opinfo = rcu_dereference(fp->f_opinfo);
@@ -1355,12 +1389,14 @@ void smb_lazy_parent_lease_break_close(struct ksmbd_file *fp)
                                continue;
                        }
 
-                       oplock_break(opinfo, SMB2_OPLOCK_LEVEL_NONE, NULL, false);
-                       opinfo_put(opinfo);
+                       if (oplock_break_add(&brk_list, opinfo))
+                               opinfo_put(opinfo);
                }
        }
        up_read(&p_ci->m_lock);
 
+       oplock_break_drain_none(&brk_list);
+
        ksmbd_inode_put(p_ci);
 }
 
@@ -1602,9 +1638,11 @@ static void __smb_break_all_levII_oplock(struct ksmbd_work *work,
                                         bool send_interim, bool send_oplock_break)
 {
        struct oplock_info *op, *brk_op;
+       struct oplock_break_entry *ent, *tmp;
        struct ksmbd_inode *ci;
        struct ksmbd_conn *conn = work->conn;
        bool sent_interim = false;
+       LIST_HEAD(brk_list);
 
        if (!test_share_config_flag(work->tcon->share_conf,
                                    KSMBD_SHARE_FLAG_OPLOCKS))
@@ -1646,6 +1684,22 @@ static void __smb_break_all_levII_oplock(struct ksmbd_work *work,
                            SMB2_LEASE_KEY_SIZE))
                        goto next;
                brk_op->open_trunc = is_trunc;
+
+               /*
+                * Defer the break until ci->m_lock is released: oplock_break()
+                * may block waiting for the lease break acknowledgment, and the
+                * close that wakes that wait needs ci->m_lock for write.
+                */
+               if (!oplock_break_add(&brk_list, brk_op))
+                       continue;
+next:
+               opinfo_put(brk_op);
+       }
+       up_read(&ci->m_lock);
+
+       list_for_each_entry_safe(ent, tmp, &brk_list, list) {
+               brk_op = ent->opinfo;
+
                if (!brk_op->is_lease && !send_oplock_break) {
                        brk_op->level = SMB2_OPLOCK_LEVEL_NONE;
                        brk_op->op_state = OPLOCK_STATE_NONE;
@@ -1657,10 +1711,10 @@ static void __smb_break_all_levII_oplock(struct ksmbd_work *work,
                                     false);
                }
                sent_interim = true;
-next:
+               list_del(&ent->list);
                opinfo_put(brk_op);
+               kfree(ent);
        }
-       up_read(&ci->m_lock);
 
        if (op)
                opinfo_put(op);