[PATCH] aio: fix request cancelling

[Robert Baldyga]

This this patch fix kiocb_cancel() function.
If cancel() function ends successfully:
- kiocb request is removed from ctx->active_reqs,
- new event is added to ring buffer (this behavior is described in comment
  in io_cancel function, but was not implemented)
- kiocb_free function is called
- percpu_ref_put is called for ctx->reqs refcount

To do this, part of io_complete function, responsible for adding new event
to ring buffer, was moved into new function named add_event().

This patch solves problem related with lack of possibility to call
aio_complete() from cancel callback. It was because it needs to lock
ctx->ctx_lock which is always locked before kiocb_cancel() call. So there
was no way to complete request properly. Now, after cancel() function call,
request and related resources are freed, and event is added to ring buffer,
so there is no need to call aio_complete() in cancel callback function.

Signed-off-by: Robert Baldyga <r.baldyga@xxxxxxxxxxx>
---
 fs/aio.c |   77 ++++++++++++++++++++++++++++++++++++++------------------------
 1 file changed, 47 insertions(+), 30 deletions(-)

diff --git a/fs/aio.c b/fs/aio.c
index 062a5f6..ea275fa 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -471,8 +471,12 @@ void kiocb_set_cancel_fn(struct kiocb *req, kiocb_cancel_fn *cancel)
 }
 EXPORT_SYMBOL(kiocb_set_cancel_fn);
 
+void add_event(struct kiocb *iocb, long res, long res2);
+static void kiocb_free(struct kiocb *req);
+
 static int kiocb_cancel(struct kioctx *ctx, struct kiocb *kiocb)
 {
+	int ret;
 	kiocb_cancel_fn *old, *cancel;
 
 	/*
@@ -488,8 +492,15 @@ static int kiocb_cancel(struct kioctx *ctx, struct kiocb *kiocb)
 		old = cancel;
 		cancel = cmpxchg(&kiocb->ki_cancel, old, KIOCB_CANCELLED);
 	} while (cancel != old);
-
-	return cancel(kiocb);
+	
+	ret = cancel(kiocb);
+	if (!ret) {
+		list_del_init(&kiocb->ki_list);
+		add_event(kiocb, -EIDRM, -EIDRM);
+		kiocb_free(kiocb);
+		percpu_ref_put(&ctx->reqs);
+	}
+	return ret;
 }
 
 static void free_ioctx(struct work_struct *work)
@@ -906,40 +917,13 @@ out:
 	return ret;
 }
 
-/* aio_complete
- *	Called when the io request on the given iocb is complete.
- */
-void aio_complete(struct kiocb *iocb, long res, long res2)
+void add_event(struct kiocb *iocb, long res, long res2)
 {
 	struct kioctx	*ctx = iocb->ki_ctx;
 	struct aio_ring	*ring;
 	struct io_event	*ev_page, *event;
 	unsigned long	flags;
 	unsigned tail, pos;
-
-	/*
-	 * Special case handling for sync iocbs:
-	 *  - events go directly into the iocb for fast handling
-	 *  - the sync task with the iocb in its stack holds the single iocb
-	 *    ref, no other paths have a way to get another ref
-	 *  - the sync task helpfully left a reference to itself in the iocb
-	 */
-	if (is_sync_kiocb(iocb)) {
-		iocb->ki_user_data = res;
-		smp_wmb();
-		iocb->ki_ctx = ERR_PTR(-EXDEV);
-		wake_up_process(iocb->ki_obj.tsk);
-		return;
-	}
-
-	if (iocb->ki_list.next) {
-		unsigned long flags;
-
-		spin_lock_irqsave(&ctx->ctx_lock, flags);
-		list_del(&iocb->ki_list);
-		spin_unlock_irqrestore(&ctx->ctx_lock, flags);
-	}
-
 	/*
 	 * Add a completion event to the ring buffer. Must be done holding
 	 * ctx->completion_lock to prevent other code from messing with the tail
@@ -983,6 +967,39 @@ void aio_complete(struct kiocb *iocb, long res, long res2)
 	spin_unlock_irqrestore(&ctx->completion_lock, flags);
 
 	pr_debug("added to ring %p at [%u]\n", iocb, tail);
+}
+
+/* aio_complete
+ *	Called when the io request on the given iocb is complete.
+ */
+void aio_complete(struct kiocb *iocb, long res, long res2)
+{
+	struct kioctx	*ctx = iocb->ki_ctx;
+
+	/*
+	 * Special case handling for sync iocbs:
+	 *  - events go directly into the iocb for fast handling
+	 *  - the sync task with the iocb in its stack holds the single iocb
+	 *    ref, no other paths have a way to get another ref
+	 *  - the sync task helpfully left a reference to itself in the iocb
+	 */
+	if (is_sync_kiocb(iocb)) {
+		iocb->ki_user_data = res;
+		smp_wmb();
+		iocb->ki_ctx = ERR_PTR(-EXDEV);
+		wake_up_process(iocb->ki_obj.tsk);
+		return;
+	}
+
+	if (iocb->ki_list.next) {
+		unsigned long flags;
+
+		spin_lock_irqsave(&ctx->ctx_lock, flags);
+		list_del(&iocb->ki_list);
+		spin_unlock_irqrestore(&ctx->ctx_lock, flags);
+	}
+
+	add_event(iocb, res, res2);
 
 	/*
 	 * Check if the user asked us to deliver the result through an
-- 
1.7.9.5

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/