1 files changed, 140 insertions, 38 deletions

diff --git a/fs/dlm/plock.c b/fs/dlm/plock.c
index 70a4752ed913..e6b4c1a21446 100644
--- a/fs/dlm/plock.c
+++ b/fs/dlm/plock.c
@@ -11,6 +11,8 @@
 #include <linux/dlm_plock.h>
 #include <linux/slab.h>
 
+#include <trace/events/dlm.h>
+
 #include "dlm_internal.h"
 #include "lockspace.h"
 
@@ -42,6 +44,27 @@ static inline void set_version(struct dlm_plock_info *info)
 	info->version[2] = DLM_PLOCK_VERSION_PATCH;
 }
 
+static struct plock_op *plock_lookup_waiter(const struct dlm_plock_info *info)
+{
+	struct plock_op *op = NULL, *iter;
+
+	list_for_each_entry(iter, &recv_list, list) {
+		if (iter->info.fsid == info->fsid &&
+		    iter->info.number == info->number &&
+		    iter->info.owner == info->owner &&
+		    iter->info.pid == info->pid &&
+		    iter->info.start == info->start &&
+		    iter->info.end == info->end &&
+		    iter->info.ex == info->ex &&
+		    iter->info.wait) {
+			op = iter;
+			break;
+		}
+	}
+
+	return op;
+}
+
 static int check_version(struct dlm_plock_info *info)
 {
 	if ((DLM_PLOCK_VERSION_MAJOR != info->version[0]) ||
@@ -74,30 +97,26 @@ static void send_op(struct plock_op *op)
 	wake_up(&send_wq);
 }
 
-/* If a process was killed while waiting for the only plock on a file,
-   locks_remove_posix will not see any lock on the file so it won't
-   send an unlock-close to us to pass on to userspace to clean up the
-   abandoned waiter.  So, we have to insert the unlock-close when the
-   lock call is interrupted. */
-
-static void do_unlock_close(const struct dlm_plock_info *info)
+static int do_lock_cancel(const struct dlm_plock_info *orig_info)
 {
 	struct plock_op *op;
+	int rv;
 
 	op = kzalloc(sizeof(*op), GFP_NOFS);
 	if (!op)
-		return;
+		return -ENOMEM;
+
+	op->info = *orig_info;
+	op->info.optype = DLM_PLOCK_OP_CANCEL;
+	op->info.wait = 0;
 
-	op->info.optype		= DLM_PLOCK_OP_UNLOCK;
-	op->info.pid		= info->pid;
-	op->info.fsid		= info->fsid;
-	op->info.number		= info->number;
-	op->info.start		= 0;
-	op->info.end		= OFFSET_MAX;
-	op->info.owner		= info->owner;
-
-	op->info.flags |= DLM_PLOCK_FL_CLOSE;
 	send_op(op);
+	wait_event(recv_wq, (op->done != 0));
+
+	rv = op->info.rv;
+
+	dlm_release_plock_op(op);
+	return rv;
 }
 
 int dlm_posix_lock(dlm_lockspace_t *lockspace, u64 number, struct file *file,
@@ -156,7 +175,7 @@ int dlm_posix_lock(dlm_lockspace_t *lockspace, u64 number, struct file *file,
 	send_op(op);
 
 	if (op->info.wait) {
-		rv = wait_event_killable(recv_wq, (op->done != 0));
+		rv = wait_event_interruptible(recv_wq, (op->done != 0));
 		if (rv == -ERESTARTSYS) {
 			spin_lock(&ops_lock);
 			/* recheck under ops_lock if we got a done != 0,
@@ -166,17 +185,37 @@ int dlm_posix_lock(dlm_lockspace_t *lockspace, u64 number, struct file *file,
 				spin_unlock(&ops_lock);
 				goto do_lock_wait;
 			}
-			list_del(&op->list);
 			spin_unlock(&ops_lock);
 
+			rv = do_lock_cancel(&op->info);
+			switch (rv) {
+			case 0:
+				/* waiter was deleted in user space, answer will never come
+				 * remove original request. The original request must be
+				 * on recv_list because the answer of do_lock_cancel()
+				 * synchronized it.
+				 */
+				spin_lock(&ops_lock);
+				list_del(&op->list);
+				spin_unlock(&ops_lock);
+				rv = -EINTR;
+				break;
+			case -ENOENT:
+				/* cancellation wasn't successful but op should be done */
+				fallthrough;
+			default:
+				/* internal error doing cancel we need to wait */
+				goto wait;
+			}
+
 			log_debug(ls, "%s: wait interrupted %x %llx pid %d",
 				  __func__, ls->ls_global_id,
 				  (unsigned long long)number, op->info.pid);
-			do_unlock_close(&op->info);
 			dlm_release_plock_op(op);
 			goto out;
 		}
 	} else {
+wait:
 		wait_event(recv_wq, (op->done != 0));
 	}
 
@@ -240,8 +279,8 @@ static int dlm_plock_callback(struct plock_op *op)
 	rv = notify(fl, 0);
 	if (rv) {
 		/* XXX: We need to cancel the fs lock here: */
-		log_print("dlm_plock_callback: lock granted after lock request "
-			  "failed; dangling lock!\n");
+		log_print("%s: lock granted after lock request failed; dangling lock!",
+			  __func__);
 		goto out;
 	}
 
@@ -318,6 +357,75 @@ out:
 }
 EXPORT_SYMBOL_GPL(dlm_posix_unlock);
 
+/*
+ * NOTE: This implementation can only handle async lock requests as nfs
+ * do it. It cannot handle cancellation of a pending lock request sitting
+ * in wait_event(), but for now only nfs is the only user local kernel
+ * user.
+ */
+int dlm_posix_cancel(dlm_lockspace_t *lockspace, u64 number, struct file *file,
+		     struct file_lock *fl)
+{
+	struct dlm_plock_info info;
+	struct plock_op *op;
+	struct dlm_ls *ls;
+	int rv;
+
+	/* this only works for async request for now and nfs is the only
+	 * kernel user right now.
+	 */
+	if (WARN_ON_ONCE(!fl->fl_lmops || !fl->fl_lmops->lm_grant))
+		return -EOPNOTSUPP;
+
+	ls = dlm_find_lockspace_local(lockspace);
+	if (!ls)
+		return -EINVAL;
+
+	memset(&info, 0, sizeof(info));
+	info.pid = fl->fl_pid;
+	info.ex = (fl->fl_type == F_WRLCK);
+	info.fsid = ls->ls_global_id;
+	dlm_put_lockspace(ls);
+	info.number = number;
+	info.start = fl->fl_start;
+	info.end = fl->fl_end;
+	info.owner = (__u64)fl->fl_pid;
+
+	rv = do_lock_cancel(&info);
+	switch (rv) {
+	case 0:
+		spin_lock(&ops_lock);
+		/* lock request to cancel must be on recv_list because
+		 * do_lock_cancel() synchronizes it.
+		 */
+		op = plock_lookup_waiter(&info);
+		if (WARN_ON_ONCE(!op)) {
+			spin_unlock(&ops_lock);
+			rv = -ENOLCK;
+			break;
+		}
+
+		list_del(&op->list);
+		spin_unlock(&ops_lock);
+		WARN_ON(op->info.optype != DLM_PLOCK_OP_LOCK);
+		op->data->callback(op->data->fl, -EINTR);
+		dlm_release_plock_op(op);
+		rv = -EINTR;
+		break;
+	case -ENOENT:
+		/* if cancel wasn't successful we probably were to late
+		 * or it was a non-blocking lock request, so just unlock it.
+		 */
+		rv = dlm_posix_unlock(lockspace, number, file, fl);
+		break;
+	default:
+		break;
+	}
+
+	return rv;
+}
+EXPORT_SYMBOL_GPL(dlm_posix_cancel);
+
 int dlm_posix_get(dlm_lockspace_t *lockspace, u64 number, struct file *file,
 		  struct file_lock *fl)
 {
@@ -403,6 +511,8 @@ static ssize_t dev_read(struct file *file, char __user *u, size_t count,
 	if (!op)
 		return -EAGAIN;
 
+	trace_dlm_plock_read(&info);
+
 	/* there is no need to get a reply from userspace for unlocks
 	   that were generated by the vfs cleaning up for a close
 	   (the process did not make an unlock call). */
@@ -430,6 +540,8 @@ static ssize_t dev_write(struct file *file, const char __user *u, size_t count,
 	if (copy_from_user(&info, u, sizeof(info)))
 		return -EFAULT;
 
+	trace_dlm_plock_write(&info);
+
 	if (check_version(&info))
 		return -EINVAL;
 
@@ -441,22 +553,11 @@ static ssize_t dev_write(struct file *file, const char __user *u, size_t count,
 	 */
 	spin_lock(&ops_lock);
 	if (info.wait) {
-		list_for_each_entry(iter, &recv_list, list) {
-			if (iter->info.fsid == info.fsid &&
-			    iter->info.number == info.number &&
-			    iter->info.owner == info.owner &&
-			    iter->info.pid == info.pid &&
-			    iter->info.start == info.start &&
-			    iter->info.end == info.end &&
-			    iter->info.ex == info.ex &&
-			    iter->info.wait) {
-				op = iter;
-				break;
-			}
-		}
+		op = plock_lookup_waiter(&info);
 	} else {
 		list_for_each_entry(iter, &recv_list, list) {
-			if (!iter->info.wait) {
+			if (!iter->info.wait &&
+			    iter->info.fsid == info.fsid) {
 				op = iter;
 				break;
 			}
@@ -468,8 +569,7 @@ static ssize_t dev_write(struct file *file, const char __user *u, size_t count,
 		if (info.wait)
 			WARN_ON(op->info.optype != DLM_PLOCK_OP_LOCK);
 		else
-			WARN_ON(op->info.fsid != info.fsid ||
-				op->info.number != info.number ||
+			WARN_ON(op->info.number != info.number ||
 				op->info.owner != info.owner ||
 				op->info.optype != info.optype);
 
@@ -534,5 +634,7 @@ int dlm_plock_init(void)
 void dlm_plock_exit(void)
 {
 	misc_deregister(&plock_dev_misc);
+	WARN_ON(!list_empty(&send_list));
+	WARN_ON(!list_empty(&recv_list));
 }