[PATCH] vfio: fix vfio_info_cap_add/shift

[PATCH] vfio: fix vfio_info_cap_add/shift

[Eric Auger]

Capability header next field is an offset relative to the start of
the INFO buffer. tmp->next is assigned the proper value but iterations
implemented in vfio_info_cap_add and vfio_info_cap_shift use next
as an offset between the headers. When coping with multiple capabilities
this leads to an Oops.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
---
 drivers/vfio/vfio.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/vfio/vfio.c b/drivers/vfio/vfio.c
index d1d70e0..1e838d1 100644
--- a/drivers/vfio/vfio.c
+++ b/drivers/vfio/vfio.c
@@ -1763,7 +1763,7 @@ struct vfio_info_cap_header *vfio_info_cap_add(struct vfio_info_cap *caps,
 	header->version = version;
 
 	/* Add to the end of the capability chain */
-	for (tmp = caps->buf; tmp->next; tmp = (void *)tmp + tmp->next)
+	for (tmp = buf; tmp->next; tmp = buf + tmp->next)
 		; /* nothing */
 
 	tmp->next = caps->size;
@@ -1776,8 +1776,9 @@ EXPORT_SYMBOL_GPL(vfio_info_cap_add);
 void vfio_info_cap_shift(struct vfio_info_cap *caps, size_t offset)
 {
 	struct vfio_info_cap_header *tmp;
+	void *buf = (void *)caps->buf;
 
-	for (tmp = caps->buf; tmp->next; tmp = (void *)tmp + tmp->next - offset)
+	for (tmp = buf; tmp->next; tmp = buf + tmp->next - offset)
 		tmp->next += offset;
 }
 EXPORT_SYMBOL_GPL(vfio_info_cap_shift);
-- 
2.5.5

Re: [PATCH] vfio: fix vfio_info_cap_add/shift

[David Hildenbrand]

Am 21.11.2016 um 07:21 schrieb Eric Auger:
> Capability header next field is an offset relative to the start of
> the INFO buffer. tmp->next is assigned the proper value but iterations
> implemented in vfio_info_cap_add and vfio_info_cap_shift use next
> as an offset between the headers. When coping with multiple capabilities
> this leads to an Oops.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> ---
>  drivers/vfio/vfio.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/vfio/vfio.c b/drivers/vfio/vfio.c
> index d1d70e0..1e838d1 100644
> --- a/drivers/vfio/vfio.c
> +++ b/drivers/vfio/vfio.c
> @@ -1763,7 +1763,7 @@ struct vfio_info_cap_header *vfio_info_cap_add(struct vfio_info_cap *caps,
>  	header->version = version;
>
>  	/* Add to the end of the capability chain */
> -	for (tmp = caps->buf; tmp->next; tmp = (void *)tmp + tmp->next)
> +	for (tmp = buf; tmp->next; tmp = buf + tmp->next)
>  		; /* nothing */
>

Looks sane to me, tmp->next is always set to the old size, which 
corresponds to the offset from the start of the buffer.

>  	tmp->next = caps->size;
> @@ -1776,8 +1776,9 @@ EXPORT_SYMBOL_GPL(vfio_info_cap_add);
>  void vfio_info_cap_shift(struct vfio_info_cap *caps, size_t offset)
>  {
>  	struct vfio_info_cap_header *tmp;
> +	void *buf = (void *)caps->buf;
>
> -	for (tmp = caps->buf; tmp->next; tmp = (void *)tmp + tmp->next - offset)
> +	for (tmp = buf; tmp->next; tmp = buf + tmp->next - offset)
>  		tmp->next += offset;


Reviewed-by: David Hildenbrand <david@redhat.com>

-- 

David

Re: [PATCH] vfio: fix vfio_info_cap_add/shift

[Alex Williamson]

On Mon, 21 Nov 2016 07:21:02 +0100
Eric Auger <eric.auger@redhat.com> wrote:

> Capability header next field is an offset relative to the start of
> the INFO buffer. tmp->next is assigned the proper value but iterations
> implemented in vfio_info_cap_add and vfio_info_cap_shift use next
> as an offset between the headers. When coping with multiple capabilities
> this leads to an Oops.
> 
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> ---
>  drivers/vfio/vfio.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/vfio/vfio.c b/drivers/vfio/vfio.c
> index d1d70e0..1e838d1 100644
> --- a/drivers/vfio/vfio.c
> +++ b/drivers/vfio/vfio.c
> @@ -1763,7 +1763,7 @@ struct vfio_info_cap_header *vfio_info_cap_add(struct vfio_info_cap *caps,
>  	header->version = version;
>  
>  	/* Add to the end of the capability chain */
> -	for (tmp = caps->buf; tmp->next; tmp = (void *)tmp + tmp->next)
> +	for (tmp = buf; tmp->next; tmp = buf + tmp->next)
>  		; /* nothing */
>  
>  	tmp->next = caps->size;
> @@ -1776,8 +1776,9 @@ EXPORT_SYMBOL_GPL(vfio_info_cap_add);
>  void vfio_info_cap_shift(struct vfio_info_cap *caps, size_t offset)
>  {
>  	struct vfio_info_cap_header *tmp;
> +	void *buf = (void *)caps->buf;
>  
> -	for (tmp = caps->buf; tmp->next; tmp = (void *)tmp + tmp->next - offset)
> +	for (tmp = buf; tmp->next; tmp = buf + tmp->next - offset)
>  		tmp->next += offset;
>  }
>  EXPORT_SYMBOL_GPL(vfio_info_cap_shift);

Thanks, good to get this fixed before we start getting multiple entries
in the chain and expose it via mdev.  For that same reason I also don't
see that this needs any sort of stable backport as the existing code
works for the single entries in the tree.  Applied to next branch for
4.10 with David's R-b.  Thanks,

Alex