Merge tag 'io_uring-5.19-2022-06-24' of git://git.kernel.dk/linux-block

Pull io_uring fixes from Jens Axboe:
 "A few fixes that should go into the 5.19 release. All are fixing
  issues that either happened in this release, or going to stable.

  In detail:

   - A small series of fixlets for the poll handling, all destined for
     stable (Pavel)

   - Fix a merge error from myself that caused a potential -EINVAL for
     the recv/recvmsg flag setting (me)

   - Fix a kbuf recycling issue for partial IO (me)

   - Use the original request for the inflight tracking (me)

   - Fix an issue introduced this merge window with trace points using a
     custom decoder function, which won't work for perf (Dylan)"

* tag 'io_uring-5.19-2022-06-24' of git://git.kernel.dk/linux-block:
  io_uring: use original request task for inflight tracking
  io_uring: move io_uring_get_opcode out of TP_printk
  io_uring: fix double poll leak on repolling
  io_uring: fix wrong arm_poll error handling
  io_uring: fail links when poll fails
  io_uring: fix req->apoll_events
  io_uring: fix merge error in checking send/recv addr2 flags
  io_uring: mark reissue requests with REQ_F_PARTIAL_IO

fs/io_uring.c

@@ -1975,7 +1975,7 @@ static inline void io_req_track_inflight(struct io_kiocb *req)
 {
 	if (!(req->flags & REQ_F_INFLIGHT)) {
 		req->flags |= REQ_F_INFLIGHT;
-		atomic_inc(&current->io_uring->inflight_tracked);
+		atomic_inc(&req->task->io_uring->inflight_tracked);
 	}
 }
 
@@ -3437,7 +3437,7 @@ static bool __io_complete_rw_common(struct io_kiocb *req, long res)
 	if (unlikely(res != req->cqe.res)) {
 		if ((res == -EAGAIN || res == -EOPNOTSUPP) &&
 		    io_rw_should_reissue(req)) {
-			req->flags |= REQ_F_REISSUE;
+			req->flags |= REQ_F_REISSUE | REQ_F_PARTIAL_IO;
 			return true;
 		}
 		req_set_fail(req);
@@ -3487,7 +3487,7 @@ static void io_complete_rw_iopoll(struct kiocb *kiocb, long res)
 		kiocb_end_write(req);
 	if (unlikely(res != req->cqe.res)) {
 		if (res == -EAGAIN && io_rw_should_reissue(req)) {
-			req->flags |= REQ_F_REISSUE;
+			req->flags |= REQ_F_REISSUE | REQ_F_PARTIAL_IO;
 			return;
 		}
 		req->cqe.res = res;
@@ -6077,8 +6077,6 @@ static int io_sendmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 
 	if (unlikely(sqe->file_index))
 		return -EINVAL;
-	if (unlikely(sqe->addr2 || sqe->file_index))
-		return -EINVAL;
 
 	sr->umsg = u64_to_user_ptr(READ_ONCE(sqe->addr));
 	sr->len = READ_ONCE(sqe->len);
@@ -6315,8 +6313,6 @@ static int io_recvmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 
 	if (unlikely(sqe->file_index))
 		return -EINVAL;
-	if (unlikely(sqe->addr2 || sqe->file_index))
-		return -EINVAL;
 
 	sr->umsg = u64_to_user_ptr(READ_ONCE(sqe->addr));
 	sr->len = READ_ONCE(sqe->len);
@@ -6954,7 +6950,8 @@ static void io_apoll_task_func(struct io_kiocb *req, bool *locked)
 		io_req_complete_failed(req, ret);
 }
 
-static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
+static void __io_poll_execute(struct io_kiocb *req, int mask,
+			      __poll_t __maybe_unused events)
 {
 	req->cqe.res = mask;
 	/*
@@ -6963,7 +6960,6 @@ static void __io_poll_execute(struct io_kiocb *req, int mask, __poll_t events)
 	 * CPU. We want to avoid pulling in req->apoll->events for that
 	 * case.
 	 */
-	req->apoll_events = events;
 	if (req->opcode == IORING_OP_POLL_ADD)
 		req->io_task_work.func = io_poll_task_func;
 	else
@@ -7114,6 +7110,8 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
 	io_init_poll_iocb(poll, mask, io_poll_wake);
 	poll->file = req->file;
 
+	req->apoll_events = poll->events;
+
 	ipt->pt._key = mask;
 	ipt->req = req;
 	ipt->error = 0;
@@ -7144,8 +7142,11 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
 
 	if (mask) {
 		/* can't multishot if failed, just queue the event we've got */
-		if (unlikely(ipt->error || !ipt->nr_entries))
+		if (unlikely(ipt->error || !ipt->nr_entries)) {
 			poll->events |= EPOLLONESHOT;
+			req->apoll_events |= EPOLLONESHOT;
+			ipt->error = 0;
+		}
 		__io_poll_execute(req, mask, poll->events);
 		return 0;
 	}
@@ -7207,6 +7208,7 @@ static int io_arm_poll_handler(struct io_kiocb *req, unsigned issue_flags)
 		mask |= EPOLLEXCLUSIVE;
 	if (req->flags & REQ_F_POLLED) {
 		apoll = req->apoll;
+		kfree(apoll->double_poll);
 	} else if (!(issue_flags & IO_URING_F_UNLOCKED) &&
 		   !list_empty(&ctx->apoll_cache)) {
 		apoll = list_first_entry(&ctx->apoll_cache, struct async_poll,
@@ -7392,7 +7394,7 @@ static int io_poll_add_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe
 		return -EINVAL;
 
 	io_req_set_refcount(req);
-	req->apoll_events = poll->events = io_poll_parse_events(sqe, flags);
+	poll->events = io_poll_parse_events(sqe, flags);
 	return 0;
 }
 
@@ -7405,6 +7407,8 @@ static int io_poll_add(struct io_kiocb *req, unsigned int issue_flags)
 	ipt.pt._qproc = io_poll_queue_proc;
 
 	ret = __io_arm_poll_handler(req, &req->poll, &ipt, poll->events);
+	if (!ret && ipt.error)
+		req_set_fail(req);
 	ret = ret ?: ipt.error;
 	if (ret)
 		__io_req_complete(req, issue_flags, ret, 0);

include/trace/events/io_uring.h

@@ -158,6 +158,8 @@ TRACE_EVENT(io_uring_queue_async_work,
 		__field(  unsigned int,			flags		)
 		__field(  struct io_wq_work *,		work		)
 		__field(  int,				rw		)
+
+		__string( op_str, io_uring_get_opcode(opcode)	)
 	),
 
 	TP_fast_assign(
@@ -168,11 +170,13 @@ TRACE_EVENT(io_uring_queue_async_work,
 		__entry->opcode		= opcode;
 		__entry->work		= work;
 		__entry->rw		= rw;
+
+		__assign_str(op_str, io_uring_get_opcode(opcode));
 	),
 
 	TP_printk("ring %p, request %p, user_data 0x%llx, opcode %s, flags 0x%x, %s queue, work %p",
 		__entry->ctx, __entry->req, __entry->user_data,
-		io_uring_get_opcode(__entry->opcode),
+		__get_str(op_str),
 		__entry->flags, __entry->rw ? "hashed" : "normal", __entry->work)
 );
 
@@ -198,6 +202,8 @@ TRACE_EVENT(io_uring_defer,
 		__field(  void *,		req	)
 		__field(  unsigned long long,	data	)
 		__field(  u8,			opcode	)
+
+		__string( op_str, io_uring_get_opcode(opcode) )
 	),
 
 	TP_fast_assign(
@@ -205,11 +211,13 @@ TRACE_EVENT(io_uring_defer,
 		__entry->req	= req;
 		__entry->data	= user_data;
 		__entry->opcode	= opcode;
+
+		__assign_str(op_str, io_uring_get_opcode(opcode));
 	),
 
 	TP_printk("ring %p, request %p, user_data 0x%llx, opcode %s",
 		__entry->ctx, __entry->req, __entry->data,
-		io_uring_get_opcode(__entry->opcode))
+		__get_str(op_str))
 );
 
 /**
@@ -298,6 +306,8 @@ TRACE_EVENT(io_uring_fail_link,
 		__field(  unsigned long long,	user_data	)
 		__field(  u8,			opcode		)
 		__field(  void *,		link		)
+
+		__string( op_str, io_uring_get_opcode(opcode) )
 	),
 
 	TP_fast_assign(
@@ -306,11 +316,13 @@ TRACE_EVENT(io_uring_fail_link,
 		__entry->user_data	= user_data;
 		__entry->opcode		= opcode;
 		__entry->link		= link;
+
+		__assign_str(op_str, io_uring_get_opcode(opcode));
 	),
 
 	TP_printk("ring %p, request %p, user_data 0x%llx, opcode %s, link %p",
 		__entry->ctx, __entry->req, __entry->user_data,
-		io_uring_get_opcode(__entry->opcode), __entry->link)
+		__get_str(op_str), __entry->link)
 );
 
 /**
@@ -390,6 +402,8 @@ TRACE_EVENT(io_uring_submit_sqe,
 		__field(  u32,			flags		)
 		__field(  bool,			force_nonblock	)
 		__field(  bool,			sq_thread	)
+
+		__string( op_str, io_uring_get_opcode(opcode) )
 	),
 
 	TP_fast_assign(
@@ -400,11 +414,13 @@ TRACE_EVENT(io_uring_submit_sqe,
 		__entry->flags		= flags;
 		__entry->force_nonblock	= force_nonblock;
 		__entry->sq_thread	= sq_thread;
+
+		__assign_str(op_str, io_uring_get_opcode(opcode));
 	),
 
 	TP_printk("ring %p, req %p, user_data 0x%llx, opcode %s, flags 0x%x, "
 		  "non block %d, sq_thread %d", __entry->ctx, __entry->req,
-		  __entry->user_data, io_uring_get_opcode(__entry->opcode),
+		  __entry->user_data, __get_str(op_str),
 		  __entry->flags, __entry->force_nonblock, __entry->sq_thread)
 );
 
@@ -435,6 +451,8 @@ TRACE_EVENT(io_uring_poll_arm,
 		__field(  u8,			opcode		)
 		__field(  int,			mask		)
 		__field(  int,			events		)
+
+		__string( op_str, io_uring_get_opcode(opcode) )
 	),
 
 	TP_fast_assign(
@@ -444,11 +462,13 @@ TRACE_EVENT(io_uring_poll_arm,
 		__entry->opcode		= opcode;
 		__entry->mask		= mask;
 		__entry->events		= events;
+
+		__assign_str(op_str, io_uring_get_opcode(opcode));
 	),
 
 	TP_printk("ring %p, req %p, user_data 0x%llx, opcode %s, mask 0x%x, events 0x%x",
 		  __entry->ctx, __entry->req, __entry->user_data,
-		  io_uring_get_opcode(__entry->opcode),
+		  __get_str(op_str),
 		  __entry->mask, __entry->events)
 );
 
@@ -474,6 +494,8 @@ TRACE_EVENT(io_uring_task_add,
 		__field(  unsigned long long,	user_data	)
 		__field(  u8,			opcode		)
 		__field(  int,			mask		)
+
+		__string( op_str, io_uring_get_opcode(opcode) )
 	),
 
 	TP_fast_assign(
@@ -482,11 +504,13 @@ TRACE_EVENT(io_uring_task_add,
 		__entry->user_data	= user_data;
 		__entry->opcode		= opcode;
 		__entry->mask		= mask;
+
+		__assign_str(op_str, io_uring_get_opcode(opcode));
 	),
 
 	TP_printk("ring %p, req %p, user_data 0x%llx, opcode %s, mask %x",
 		__entry->ctx, __entry->req, __entry->user_data,
-		io_uring_get_opcode(__entry->opcode),
+		__get_str(op_str),
 		__entry->mask)
 );
 
@@ -523,6 +547,8 @@ TRACE_EVENT(io_uring_req_failed,
 		__field( u64,			pad1		)
 		__field( u64,			addr3		)
 		__field( int,			error		)
+
+		__string( op_str, io_uring_get_opcode(sqe->opcode) )
 	),
 
 	TP_fast_assign(
@@ -542,6 +568,8 @@ TRACE_EVENT(io_uring_req_failed,
 		__entry->pad1		= sqe->__pad2[0];
 		__entry->addr3		= sqe->addr3;
 		__entry->error		= error;
+
+		__assign_str(op_str, io_uring_get_opcode(sqe->opcode));
 	),
 
 	TP_printk("ring %p, req %p, user_data 0x%llx, "
@@ -550,7 +578,7 @@ TRACE_EVENT(io_uring_req_failed,
 		  "personality=%d, file_index=%d, pad=0x%llx, addr3=%llx, "
 		  "error=%d",
 		  __entry->ctx, __entry->req, __entry->user_data,
-		  io_uring_get_opcode(__entry->opcode),
+		  __get_str(op_str),
 		  __entry->flags, __entry->ioprio,
 		  (unsigned long long)__entry->off,
 		  (unsigned long long) __entry->addr, __entry->len,