[PATCH] RDMA/ucma: check fd type in ucma_migrate_id()

[PATCH] RDMA/ucma: check fd type in ucma_migrate_id()

[Jann Horn]

The current code grabs the private_data of whatever file descriptor
userspace has supplied and implicitly casts it to a `struct ucma_file *`,
potentially causing a type confusion.

This is probably fine in practice because the pointer is only used for
comparisons, it is never actually dereferenced; and even in the
comparisons, it is unlikely that a file from another filesystem would have
a ->private_data pointer that happens to also be valid in this context.
But ->private_data is not always guaranteed to be a valid pointer to an
object owned by the file's filesystem; for example, some filesystems just
cram numbers in there.

Check the type of the supplied file descriptor to be safe, analogous to how
other places in the kernel do it.

Fixes: 88314e4dda1e ("RDMA/cma: add support for rdma_migrate_id()")
Signed-off-by: Jann Horn <jannh@google.com>
---
Only compile-tested, because I don't have an environment in which I
could test this.

 drivers/infiniband/core/ucma.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c
index ec8fb289621f..5f437d1570fb 100644
--- a/drivers/infiniband/core/ucma.c
+++ b/drivers/infiniband/core/ucma.c
@@ -124,6 +124,8 @@ static DEFINE_MUTEX(mut);
 static DEFINE_IDR(ctx_idr);
 static DEFINE_IDR(multicast_idr);
 
+static const struct file_operations ucma_fops;
+
 static inline struct ucma_context *_ucma_find_context(int id,
 						      struct ucma_file *file)
 {
@@ -1581,6 +1583,10 @@ static ssize_t ucma_migrate_id(struct ucma_file *new_file,
 	f = fdget(cmd.fd);
 	if (!f.file)
 		return -ENOENT;
+	if (f.file->f_op != &ucma_fops) {
+		ret = -EINVAL;
+		goto file_put;
+	}
 
 	/* Validate current fd and prevent destruction of id. */
 	ctx = ucma_get_ctx(f.file->private_data, cmd.id);
-- 
2.19.0.rc1.350.ge57e33dbd1-goog

Re: [PATCH] RDMA/ucma: check fd type in ucma_migrate_id()

[Jason Gunthorpe]

On Mon, Sep 03, 2018 at 06:54:14PM +0200, Jann Horn wrote:
> The current code grabs the private_data of whatever file descriptor
> userspace has supplied and implicitly casts it to a `struct ucma_file *`,
> potentially causing a type confusion.
> 
> This is probably fine in practice because the pointer is only used for
> comparisons, it is never actually dereferenced; and even in the
> comparisons, it is unlikely that a file from another filesystem would have
> a ->private_data pointer that happens to also be valid in this context.
> But ->private_data is not always guaranteed to be a valid pointer to an
> object owned by the file's filesystem; for example, some filesystems just
> cram numbers in there.
> 
> Check the type of the supplied file descriptor to be safe, analogous to how
> other places in the kernel do it.
> 
> Fixes: 88314e4dda1e ("RDMA/cma: add support for rdma_migrate_id()")
> Signed-off-by: Jann Horn <jannh@google.com>
> ---
> Only compile-tested, because I don't have an environment in which I
> could test this.
> 
>  drivers/infiniband/core/ucma.c | 6 ++++++
>  1 file changed, 6 insertions(+)

Yep, this looks right to me also, applied to for-rc, thanks

Jason