Merge tag 'nfs-for-6.1-1' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client updates from Anna Schumaker:
 "New Features:
   - Add NFSv4.2 xattr tracepoints
   - Replace xprtiod WQ in rpcrdma
   - Flexfiles cancels I/O on layout recall or revoke

  Bugfixes and Cleanups:
   - Directly use ida_alloc() / ida_free()
   - Don't open-code max_t()
   - Prefer using strscpy over strlcpy
   - Remove unused forward declarations
   - Always return layout states on flexfiles layout return
   - Have LISTXATTR treat NFS4ERR_NOXATTR as an empty reply instead of
     error
   - Allow more xprtrdma memory allocations to fail without triggering a
     reclaim
   - Various other xprtrdma clean ups
   - Fix rpc_killall_tasks() races"

* tag 'nfs-for-6.1-1' of git://git.linux-nfs.org/projects/anna/linux-nfs: (27 commits)
  NFSv4/flexfiles: Cancel I/O if the layout is recalled or revoked
  SUNRPC: Add API to force the client to disconnect
  SUNRPC: Add a helper to allow pNFS drivers to selectively cancel RPC calls
  SUNRPC: Fix races with rpc_killall_tasks()
  xprtrdma: Fix uninitialized variable
  xprtrdma: Prevent memory allocations from driving a reclaim
  xprtrdma: Memory allocation should be allowed to fail during connect
  xprtrdma: MR-related memory allocation should be allowed to fail
  xprtrdma: Clean up synopsis of rpcrdma_regbuf_alloc()
  xprtrdma: Clean up synopsis of rpcrdma_req_create()
  svcrdma: Clean up RPCRDMA_DEF_GFP
  SUNRPC: Replace the use of the xprtiod WQ in rpcrdma
  NFSv4.2: Add a tracepoint for listxattr
  NFSv4.2: Add tracepoints for getxattr, setxattr, and removexattr
  NFSv4.2: Move TRACE_DEFINE_ENUM(NFS4_CONTENT_*) under CONFIG_NFS_V4_2
  NFSv4.2: Add special handling for LISTXATTR receiving NFS4ERR_NOXATTR
  nfs: remove nfs_wait_atomic_killable() and nfs_write_prepare() declaration
  NFSv4: remove nfs4_renewd_prepare_shutdown() declaration
  fs/nfs/pnfs_nfs.c: fix spelling typo and syntax error in comment
  NFSv4/pNFS: Always return layout stats on layout return for flexfiles
  ...

1 files changed, 36 insertions, 15 deletions

diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c
index 46cbf151a50b1a..be587a308e05a5 100644
--- a/net/sunrpc/sched.c
+++ b/net/sunrpc/sched.c
@@ -65,6 +65,13 @@ gfp_t rpc_task_gfp_mask(void)
 }
 EXPORT_SYMBOL_GPL(rpc_task_gfp_mask);
 
+bool rpc_task_set_rpc_status(struct rpc_task *task, int rpc_status)
+{
+	if (cmpxchg(&task->tk_rpc_status, 0, rpc_status) == 0)
+		return true;
+	return false;
+}
+
 unsigned long
 rpc_task_timeout(const struct rpc_task *task)
 {
@@ -853,12 +860,25 @@ void rpc_signal_task(struct rpc_task *task)
 	if (!RPC_IS_ACTIVATED(task))
 		return;
 
+	if (!rpc_task_set_rpc_status(task, -ERESTARTSYS))
+		return;
 	trace_rpc_task_signalled(task, task->tk_action);
 	set_bit(RPC_TASK_SIGNALLED, &task->tk_runstate);
 	smp_mb__after_atomic();
 	queue = READ_ONCE(task->tk_waitqueue);
 	if (queue)
-		rpc_wake_up_queued_task_set_status(queue, task, -ERESTARTSYS);
+		rpc_wake_up_queued_task(queue, task);
+}
+
+void rpc_task_try_cancel(struct rpc_task *task, int error)
+{
+	struct rpc_wait_queue *queue;
+
+	if (!rpc_task_set_rpc_status(task, error))
+		return;
+	queue = READ_ONCE(task->tk_waitqueue);
+	if (queue)
+		rpc_wake_up_queued_task(queue, task);
 }
 
 void rpc_exit(struct rpc_task *task, int status)
@@ -905,10 +925,16 @@ static void __rpc_execute(struct rpc_task *task)
 		 * Perform the next FSM step or a pending callback.
 		 *
 		 * tk_action may be NULL if the task has been killed.
-		 * In particular, note that rpc_killall_tasks may
-		 * do this at any time, so beware when dereferencing.
 		 */
 		do_action = task->tk_action;
+		/* Tasks with an RPC error status should exit */
+		if (do_action != rpc_exit_task &&
+		    (status = READ_ONCE(task->tk_rpc_status)) != 0) {
+			task->tk_status = status;
+			if (do_action != NULL)
+				do_action = rpc_exit_task;
+		}
+		/* Callbacks override all actions */
 		if (task->tk_callback) {
 			do_action = task->tk_callback;
 			task->tk_callback = NULL;
@@ -931,14 +957,6 @@ static void __rpc_execute(struct rpc_task *task)
 		}
 
 		/*
-		 * Signalled tasks should exit rather than sleep.
-		 */
-		if (RPC_SIGNALLED(task)) {
-			task->tk_rpc_status = -ERESTARTSYS;
-			rpc_exit(task, -ERESTARTSYS);
-		}
-
-		/*
 		 * The queue->lock protects against races with
 		 * rpc_make_runnable().
 		 *
@@ -953,6 +971,12 @@ static void __rpc_execute(struct rpc_task *task)
 			spin_unlock(&queue->lock);
 			continue;
 		}
+		/* Wake up any task that has an exit status */
+		if (READ_ONCE(task->tk_rpc_status) != 0) {
+			rpc_wake_up_task_queue_locked(queue, task);
+			spin_unlock(&queue->lock);
+			continue;
+		}
 		rpc_clear_running(task);
 		spin_unlock(&queue->lock);
 		if (task_is_async)
@@ -970,10 +994,7 @@ static void __rpc_execute(struct rpc_task *task)
 			 * clean up after sleeping on some queue, we don't
 			 * break the loop here, but go around once more.
 			 */
-			trace_rpc_task_signalled(task, task->tk_action);
-			set_bit(RPC_TASK_SIGNALLED, &task->tk_runstate);
-			task->tk_rpc_status = -ERESTARTSYS;
-			rpc_exit(task, -ERESTARTSYS);
+			rpc_signal_task(task);
 		}
 		trace_rpc_task_sync_wake(task, task->tk_action);
 	}