locks: fix tracking of inprogress lease breaks

We currently use a bit in fl_flags to record whether a lease is being
broken, and set fl_type to the type (RDLCK or UNLCK) that it will
eventually have.  This means that once the lease break starts, we forget
what the lease's type *used* to be.  Breaking a read lease will then
result in blocking read opens, even though there's no conflict--because
the lease type is now F_UNLCK and we can no longer tell whether it was
previously a read or write lease.

So, instead keep fl_type as the original type (the type which we
enforce), and keep track of whether we're unlocking or merely
downgrading by replacing the single FL_INPROGRESS flag by
FL_UNLOCK_PENDING and FL_DOWNGRADE_PENDING flags.

To get this right we also need to track separate downgrade and break
times, to handle the case where a write-leased file gets conflicting
opens first for read, then later for write.

(I first considered just eliminating the downgrade behavior
completely--nfsv4 doesn't need it, and nobody as far as I can tell
actually uses it currently--but Jeremy Allison tells me that Windows
oplocks do behave this way, so Samba will probably use this some day.)

Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>

1 files changed, 55 insertions, 32 deletions

diff --git a/fs/locks.c b/fs/locks.c
index c4215418bca3..c525aa4de234 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -135,7 +135,16 @@
 
 static bool lease_breaking(struct file_lock *fl)
 {
-        return fl->fl_flags & FL_INPROGRESS;
+        return fl->fl_flags & (FL_UNLOCK_PENDING | FL_DOWNGRADE_PENDING);
+}
+
+static int target_leasetype(struct file_lock *fl)
+{
+        if (fl->fl_flags & FL_UNLOCK_PENDING)
+                return F_UNLCK;
+        if (fl->fl_flags & FL_DOWNGRADE_PENDING)
+                return F_RDLCK;
+        return fl->fl_type;
 }
 
 int leases_enable = 1;
@@ -1124,6 +1133,17 @@ int locks_mandatory_area(int read_write, struct inode *inode,
 
 EXPORT_SYMBOL(locks_mandatory_area);
 
+static void lease_clear_pending(struct file_lock *fl, int arg)
+{
+        switch (arg) {
+        case F_UNLCK:
+                fl->fl_flags &= ~FL_UNLOCK_PENDING;
+                /* fall through: */
+        case F_RDLCK:
+                fl->fl_flags &= ~FL_DOWNGRADE_PENDING;
+        }
+}
+
 /* We already had a lease on this file; just change its type */
 int lease_modify(struct file_lock **before, int arg)
 {
@@ -1132,7 +1152,7 @@ int lease_modify(struct file_lock **before, int arg)
 
         if (error)
                 return error;
-        fl->fl_flags &= ~FL_INPROGRESS;
+        lease_clear_pending(fl, arg);
         locks_wake_up_blocks(fl);
         if (arg == F_UNLCK)
                 locks_delete_lock(before);
@@ -1141,6 +1161,14 @@ int lease_modify(struct file_lock **before, int arg)
 
 EXPORT_SYMBOL(lease_modify);
 
+static bool past_time(unsigned long then)
+{
+        if (!then)
+                /* 0 is a special value meaning "this never expires": */
+                return false;
+        return time_after(jiffies, then);
+}
+
 static void time_out_leases(struct inode *inode)
 {
         struct file_lock **before;
@@ -1148,12 +1176,10 @@ static void time_out_leases(struct inode *inode)
 
         before = &inode->i_flock;
         while ((fl = *before) && IS_LEASE(fl) && lease_breaking(fl)) {
-                if ((fl->fl_break_time == 0)
-                                || time_before(jiffies, fl->fl_break_time)) {
-                        before = &fl->fl_next;
-                        continue;
-                }
-                lease_modify(before, fl->fl_type);
+                if (past_time(fl->fl_downgrade_time))
+                        lease_modify(before, F_RDLCK);
+                if (past_time(fl->fl_break_time))
+                        lease_modify(before, F_UNLCK);
                 if (fl == *before)      /* lease_modify may have freed fl */
                         before = &fl->fl_next;
         }
@@ -1171,7 +1197,7 @@ static void time_out_leases(struct inode *inode)
  */
 int __break_lease(struct inode *inode, unsigned int mode)
 {
-        int error = 0, future;
+        int error = 0;
         struct file_lock *new_fl, *flock;
         struct file_lock *fl;
         unsigned long break_time;
@@ -1188,24 +1214,13 @@ int __break_lease(struct inode *inode, unsigned int mode)
         if ((flock == NULL) || !IS_LEASE(flock))
                 goto out;
 
+        if (!locks_conflict(flock, new_fl))
+                goto out;
+
         for (fl = flock; fl && IS_LEASE(fl); fl = fl->fl_next)
                 if (fl->fl_owner == current->files)
                         i_have_this_lease = 1;
 
-        if (want_write) {
-                /* If we want write access, we have to revoke any lease. */
-                future = F_UNLCK;
-        } else if (lease_breaking(flock)) {
-                /* If the lease is already being broken, we just leave it */
-                future = flock->fl_type;
-        } else if (flock->fl_type & F_WRLCK) {
-                /* Downgrade the exclusive lease to a read-only lease. */
-                future = F_RDLCK;
-        } else {
-                /* the existing lease was read-only, so we can read too. */
-                goto out;
-        }
-
         if (IS_ERR(new_fl) && !i_have_this_lease
                         && ((mode & O_NONBLOCK) == 0)) {
                 error = PTR_ERR(new_fl);
@@ -1220,13 +1235,18 @@ int __break_lease(struct inode *inode, unsigned int mode)
         }
 
         for (fl = flock; fl && IS_LEASE(fl); fl = fl->fl_next) {
-                if (fl->fl_type != future) {
-                        fl->fl_type = future;
-                        fl->fl_flags |= FL_INPROGRESS;
+                if (want_write) {
+                        if (fl->fl_flags & FL_UNLOCK_PENDING)
+                                continue;
+                        fl->fl_flags |= FL_UNLOCK_PENDING;
                         fl->fl_break_time = break_time;
-                        /* lease must have lmops break callback */
-                        fl->fl_lmops->lm_break(fl);
+                } else {
+                        if (lease_breaking(flock))
+                                continue;
+                        fl->fl_flags |= FL_DOWNGRADE_PENDING;
+                        fl->fl_downgrade_time = break_time;
                 }
+                fl->fl_lmops->lm_break(fl);
         }
 
         if (i_have_this_lease || (mode & O_NONBLOCK)) {
@@ -1250,10 +1270,13 @@ restart:
         if (error >= 0) {
                 if (error == 0)
                         time_out_leases(inode);
-                /* Wait for the next lease that has not been broken yet */
+                /*
+                 * Wait for the next conflicting lease that has not been
+                 * broken yet
+                 */
                 for (flock = inode->i_flock; flock && IS_LEASE(flock);
                                 flock = flock->fl_next) {
-                        if (lease_breaking(flock))
+                        if (locks_conflict(new_fl, flock))
                                 goto restart;
                 }
                 error = 0;
@@ -1321,7 +1344,7 @@ int fcntl_getlease(struct file *filp)
         for (fl = filp->f_path.dentry->d_inode->i_flock; fl && IS_LEASE(fl);
                         fl = fl->fl_next) {
                 if (fl->fl_file == filp) {
-                        type = fl->fl_type;
+                        type = target_leasetype(fl);
                         break;
                 }
         }
@@ -1386,7 +1409,7 @@ int generic_setlease(struct file *filp, long arg, struct file_lock **flp)
                         before = &fl->fl_next) {
                 if (fl->fl_file == filp)
                         my_before = before;
-                else if ((fl->fl_type == F_UNLCK) && lease_breaking(fl))
+                else if (fl->fl_flags & FL_UNLOCK_PENDING)
                         /*
                          * Someone is in the process of opening this
                          * file for writing so we may not take an