fs/aio: Check IOCB_AIO_RW before the struct aio_kiocb conversion
authorBart Van Assche <bvanassche@acm.org>
Mon, 4 Mar 2024 23:57:15 +0000 (15:57 -0800)
committerChristian Brauner <brauner@kernel.org>
Tue, 5 Mar 2024 09:40:58 +0000 (10:40 +0100)
The first kiocb_set_cancel_fn() argument may point at a struct kiocb
that is not embedded inside struct aio_kiocb. With the current code,
depending on the compiler, the req->ki_ctx read happens either before
the IOCB_AIO_RW test or after that test. Move the req->ki_ctx read such
that it is guaranteed that the IOCB_AIO_RW test happens first.

Reported-by: Eric Biggers <ebiggers@kernel.org>
Cc: Benjamin LaHaise <ben@communityfibre.ca>
Cc: Eric Biggers <ebiggers@google.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Avi Kivity <avi@scylladb.com>
Cc: Sandeep Dhavale <dhavale@google.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: stable@vger.kernel.org
Fixes: b820de741ae4 ("fs/aio: Restrict kiocb_set_cancel_fn() to I/O submitted via libaio")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240304235715.3790858-1-bvanassche@acm.org
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
fs/aio.c

index da18dbc..9cdaa2f 100644 (file)
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -589,8 +589,8 @@ static int aio_setup_ring(struct kioctx *ctx, unsigned int nr_events)
 
 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;
+       struct aio_kiocb *req;
+       struct kioctx *ctx;
        unsigned long flags;
 
        /*
@@ -600,9 +600,13 @@ void kiocb_set_cancel_fn(struct kiocb *iocb, kiocb_cancel_fn *cancel)
        if (!(iocb->ki_flags & IOCB_AIO_RW))
                return;
 
+       req = container_of(iocb, struct aio_kiocb, rw);
+
        if (WARN_ON_ONCE(!list_empty(&req->ki_list)))
                return;
 
+       ctx = req->ki_ctx;
+
        spin_lock_irqsave(&ctx->ctx_lock, flags);
        list_add_tail(&req->ki_list, &ctx->active_reqs);
        req->ki_cancel = cancel;