[PATCH 08/32] aio: replace kiocb_set_cancel_fn with a cancel_kiocb file operation

[Christoph Hellwig <hch@lst.de>]

The current kiocb_set_cancel_fn implementation assumes the kiocb is
embedded into an aio_kiocb, which is fundamentally unsafe as it might
have been submitted by non-aio callers.  Instead add a cancel_kiocb
file operation that replaced the ki_cancel function pointer set by
kiocb_set_cancel_fn, and only adds iocbs to the active list when
the read/write_iter methods return -EIOCBQUEUED and the file has
a cancel_kiocb method.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 drivers/usb/gadget/function/f_fs.c | 10 ++--------
 drivers/usb/gadget/legacy/inode.c  |  5 ++---
 fs/aio.c                           | 31 ++++++++++++------------------
 fs/orangefs/orangefs-kernel.h      |  1 -
 include/linux/aio.h                |  7 -------
 include/linux/fs.h                 |  1 +
 6 files changed, 17 insertions(+), 38 deletions(-)

diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c
index 0294e4f18873..531a7206407b 100644
--- a/drivers/usb/gadget/function/f_fs.c
+++ b/drivers/usb/gadget/function/f_fs.c
@@ -26,7 +26,6 @@
 #include <linux/usb/composite.h>
 #include <linux/usb/functionfs.h>
 
-#include <linux/aio.h>
 #include <linux/mmu_context.h>
 #include <linux/poll.h>
 #include <linux/eventfd.h>
@@ -1072,7 +1071,7 @@ ffs_epfile_open(struct inode *inode, struct file *file)
 	return 0;
 }
 
-static int ffs_aio_cancel(struct kiocb *kiocb)
+static int ffs_epfile_cancel_kiocb(struct kiocb *kiocb)
 {
 	struct ffs_io_data *io_data = kiocb->private;
 	struct ffs_epfile *epfile = kiocb->ki_filp->private_data;
@@ -1115,9 +1114,6 @@ static ssize_t ffs_epfile_write_iter(struct kiocb *kiocb, struct iov_iter *from)
 
 	kiocb->private = p;
 
-	if (p->aio)
-		kiocb_set_cancel_fn(kiocb, ffs_aio_cancel);
-
 	res = ffs_epfile_io(kiocb->ki_filp, p);
 	if (res == -EIOCBQUEUED)
 		return res;
@@ -1160,9 +1156,6 @@ static ssize_t ffs_epfile_read_iter(struct kiocb *kiocb, struct iov_iter *to)
 
 	kiocb->private = p;
 
-	if (p->aio)
-		kiocb_set_cancel_fn(kiocb, ffs_aio_cancel);
-
 	res = ffs_epfile_io(kiocb->ki_filp, p);
 	if (res == -EIOCBQUEUED)
 		return res;
@@ -1274,6 +1267,7 @@ static const struct file_operations ffs_epfile_operations = {
 	.read_iter =	ffs_epfile_read_iter,
 	.release =	ffs_epfile_release,
 	.unlocked_ioctl =	ffs_epfile_ioctl,
+	.cancel_kiocb = ffs_epfile_cancel_kiocb,
 };
 
 
diff --git a/drivers/usb/gadget/legacy/inode.c b/drivers/usb/gadget/legacy/inode.c
index 37ca0e669bd8..02ea83c8861a 100644
--- a/drivers/usb/gadget/legacy/inode.c
+++ b/drivers/usb/gadget/legacy/inode.c
@@ -21,7 +21,6 @@
 #include <linux/slab.h>
 #include <linux/poll.h>
 #include <linux/mmu_context.h>
-#include <linux/aio.h>
 #include <linux/uio.h>
 #include <linux/refcount.h>
 #include <linux/delay.h>
@@ -435,7 +434,7 @@ struct kiocb_priv {
 	unsigned		actual;
 };
 
-static int ep_aio_cancel(struct kiocb *iocb)
+static int ep_cancel_kiocb(struct kiocb *iocb)
 {
 	struct kiocb_priv	*priv = iocb->private;
 	struct ep_data		*epdata;
@@ -528,7 +527,6 @@ static ssize_t ep_aio(struct kiocb *iocb,
 	iocb->private = priv;
 	priv->iocb = iocb;
 
-	kiocb_set_cancel_fn(iocb, ep_aio_cancel);
 	get_ep(epdata);
 	priv->epdata = epdata;
 	priv->actual = 0;
@@ -700,6 +698,7 @@ static const struct file_operations ep_io_operations = {
 	.unlocked_ioctl = ep_ioctl,
 	.read_iter =	ep_read_iter,
 	.write_iter =	ep_write_iter,
+	.cancel_kiocb = ep_cancel_kiocb,
 };
 
 /* ENDPOINT INITIALIZATION
diff --git a/fs/aio.c b/fs/aio.c
index 8991baa38d5d..be10dde20c8e 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -171,7 +171,6 @@ struct aio_kiocb {
 	};
 
 	struct kioctx		*ki_ctx;
-	kiocb_cancel_fn		*ki_cancel;
 
 	struct iocb __user	*ki_user_iocb;	/* user's aiocb */
 	__u64			ki_user_data;	/* user's data for completion */
@@ -545,22 +544,6 @@ static int aio_setup_ring(struct kioctx *ctx, unsigned int nr_events)
 #define AIO_EVENTS_FIRST_PAGE	((PAGE_SIZE - sizeof(struct aio_ring)) / sizeof(struct io_event))
 #define AIO_EVENTS_OFFSET	(AIO_EVENTS_PER_PAGE - AIO_EVENTS_FIRST_PAGE)
 
-void kiocb_set_cancel_fn(struct kiocb *iocb, kiocb_cancel_fn *cancel)
-{
-	struct aio_kiocb *req = container_of(iocb, struct aio_kiocb, rw);
-	struct kioctx *ctx = req->ki_ctx;
-	unsigned long flags;
-
-	if (WARN_ON_ONCE(!list_empty(&req->ki_list)))
-		return;
-
-	spin_lock_irqsave(&ctx->ctx_lock, flags);
-	list_add_tail(&req->ki_list, &ctx->active_reqs);
-	req->ki_cancel = cancel;
-	spin_unlock_irqrestore(&ctx->ctx_lock, flags);
-}
-EXPORT_SYMBOL(kiocb_set_cancel_fn);
-
 /*
  * free_ioctx() should be RCU delayed to synchronize against the RCU
  * protected lookup_ioctx() and also needs process context to call
@@ -608,7 +591,7 @@ static void free_ioctx_users(struct percpu_ref *ref)
 		req = list_first_entry(&ctx->active_reqs,
 				       struct aio_kiocb, ki_list);
 		list_del_init(&req->ki_list);
-		req->ki_cancel(&req->rw);
+		req->rw.ki_filp->f_op->cancel_kiocb(&req->rw);
 	}
 
 	spin_unlock_irq(&ctx->ctx_lock);
@@ -1447,6 +1430,16 @@ static inline ssize_t aio_rw_ret(struct kiocb *req, ssize_t ret)
 {
 	switch (ret) {
 	case -EIOCBQUEUED:
+		if (req->ki_filp->f_op->cancel_kiocb) {
+			struct aio_kiocb *iocb =
+				container_of(req, struct aio_kiocb, rw);
+			struct kioctx *ctx = iocb->ki_ctx;
+			unsigned long flags;
+
+			spin_lock_irqsave(&ctx->ctx_lock, flags);
+			list_add_tail(&iocb->ki_list, &ctx->active_reqs);
+			spin_unlock_irqrestore(&ctx->ctx_lock, flags);
+		}
 		return ret;
 	case -ERESTARTSYS:
 	case -ERESTARTNOINTR:
@@ -1824,7 +1817,7 @@ SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb,
 	kiocb = lookup_kiocb(ctx, iocb);
 	if (kiocb) {
 		list_del_init(&kiocb->ki_list);
-		ret = kiocb->ki_cancel(&kiocb->rw);
+		ret = kiocb->rw.ki_filp->f_op->cancel_kiocb(&kiocb->rw);
 	}
 	spin_unlock_irq(&ctx->ctx_lock);
 
diff --git a/fs/orangefs/orangefs-kernel.h b/fs/orangefs/orangefs-kernel.h
index c29bb0ebc6bb..5cd598c59bdc 100644
--- a/fs/orangefs/orangefs-kernel.h
+++ b/fs/orangefs/orangefs-kernel.h
@@ -34,7 +34,6 @@
 #include <linux/fs.h>
 #include <linux/vmalloc.h>
 
-#include <linux/aio.h>
 #include <linux/posix_acl.h>
 #include <linux/posix_acl_xattr.h>
 #include <linux/compat.h>
diff --git a/include/linux/aio.h b/include/linux/aio.h
index b83e68dd006f..42e3cc953005 100644
--- a/include/linux/aio.h
+++ b/include/linux/aio.h
@@ -4,20 +4,13 @@
 
 #include <linux/aio_abi.h>
 
-struct kioctx;
-struct kiocb;
 struct mm_struct;
 
-typedef int (kiocb_cancel_fn)(struct kiocb *);
-
 /* prototypes */
 #ifdef CONFIG_AIO
 extern void exit_aio(struct mm_struct *mm);
-void kiocb_set_cancel_fn(struct kiocb *req, kiocb_cancel_fn *cancel);
 #else
 static inline void exit_aio(struct mm_struct *mm) { }
-static inline void kiocb_set_cancel_fn(struct kiocb *req,
-				       kiocb_cancel_fn *cancel) { }
 #endif /* CONFIG_AIO */
 
 /* for sysctl: */
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 4b6045ebb2f2..7fe64df9eb09 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1742,6 +1742,7 @@ struct file_operations {
 			u64);
 	ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
 			u64);
+	int (*cancel_kiocb)(struct kiocb *);
 } __randomize_layout;
 
 struct inode_operations {
-- 
2.17.0