Merge tag 'for-5.19/io_uring-2022-05-22' of git://git.kernel.dk/linux-block

Pull io_uring updates from Jens Axboe:
 "Here are the main io_uring changes for 5.19. This contains:

   - Fixes for sparse type warnings (Christoph, Vasily)

   - Support for multi-shot accept (Hao)

   - Support for io_uring managed fixed files, rather than always
     needing the applicationt o manage the indices (me)

   - Fix for a spurious poll wakeup (Dylan)

   - CQE overflow fixes (Dylan)

   - Support more types of cancelations (me)

   - Support for co-operative task_work signaling, rather than always
     forcing an IPI (me)

   - Support for doing poll first when appropriate, rather than always
     attempting a transfer first (me)

   - Provided buffer cleanups and support for mapped buffers (me)

   - Improve how io_uring handles inflight SCM files (Pavel)

   - Speedups for registered files (Pavel, me)

   - Organize the completion data in a struct in io_kiocb rather than
     keep it in separate spots (Pavel)

   - task_work improvements (Pavel)

   - Cleanup and optimize the submission path, in general and for
     handling links (Pavel)

   - Speedups for registered resource handling (Pavel)

   - Support sparse buffers and file maps (Pavel, me)

   - Various fixes and cleanups (Almog, Pavel, me)"

* tag 'for-5.19/io_uring-2022-05-22' of git://git.kernel.dk/linux-block: (111 commits)
  io_uring: fix incorrect __kernel_rwf_t cast
  io_uring: disallow mixed provided buffer group registrations
  io_uring: initialize io_buffer_list head when shared ring is unregistered
  io_uring: add fully sparse buffer registration
  io_uring: use rcu_dereference in io_close
  io_uring: consistently use the EPOLL* defines
  io_uring: make apoll_events a __poll_t
  io_uring: drop a spurious inline on a forward declaration
  io_uring: don't use ERR_PTR for user pointers
  io_uring: use a rwf_t for io_rw.flags
  io_uring: add support for ring mapped supplied buffers
  io_uring: add io_pin_pages() helper
  io_uring: add buffer selection support to IORING_OP_NOP
  io_uring: fix locking state for empty buffer group
  io_uring: implement multishot mode for accept
  io_uring: let fast poll support multishot
  io_uring: add REQ_F_APOLL_MULTISHOT for requests
  io_uring: add IORING_ACCEPT_MULTISHOT for accept
  io_uring: only wake when the correct events are set
  io_uring: avoid io-wq -EAGAIN looping for !IOPOLL
  ...

1 files changed, 19 insertions, 6 deletions

diff --git a/kernel/task_work.c b/kernel/task_work.c
index c59e1a49bc40..dff75bcde151 100644
--- a/kernel/task_work.c
+++ b/kernel/task_work.c
@@ -12,12 +12,22 @@ static struct callback_head work_exited; /* all we need is ->next == NULL */
  * @notify: how to notify the targeted task
  *
  * Queue @work for task_work_run() below and notify the @task if @notify
- * is @TWA_RESUME or @TWA_SIGNAL. @TWA_SIGNAL works like signals, in that the
- * it will interrupt the targeted task and run the task_work. @TWA_RESUME
- * work is run only when the task exits the kernel and returns to user mode,
- * or before entering guest mode. Fails if the @task is exiting/exited and thus
- * it can't process this @work. Otherwise @work->func() will be called when the
- * @task goes through one of the aforementioned transitions, or exits.
+ * is @TWA_RESUME, @TWA_SIGNAL, or @TWA_SIGNAL_NO_IPI.
+ *
+ * @TWA_SIGNAL works like signals, in that the it will interrupt the targeted
+ * task and run the task_work, regardless of whether the task is currently
+ * running in the kernel or userspace.
+ * @TWA_SIGNAL_NO_IPI works like @TWA_SIGNAL, except it doesn't send a
+ * reschedule IPI to force the targeted task to reschedule and run task_work.
+ * This can be advantageous if there's no strict requirement that the
+ * task_work be run as soon as possible, just whenever the task enters the
+ * kernel anyway.
+ * @TWA_RESUME work is run only when the task exits the kernel and returns to
+ * user mode, or before entering guest mode.
+ *
+ * Fails if the @task is exiting/exited and thus it can't process this @work.
+ * Otherwise @work->func() will be called when the @task goes through one of
+ * the aforementioned transitions, or exits.
  *
  * If the targeted task is exiting, then an error is returned and the work item
  * is not queued. It's up to the caller to arrange for an alternative mechanism
@@ -53,6 +63,9 @@ int task_work_add(struct task_struct *task, struct callback_head *work,
 	case TWA_SIGNAL:
 		set_notify_signal(task);
 		break;
+	case TWA_SIGNAL_NO_IPI:
+		__set_notify_signal(task);
+		break;
 	default:
 		WARN_ON_ONCE(1);
 		break;