Re: [Xen-devel] [PATCH] pvcalls-front: fixes incorrect error handling

From: Stefano Stabellini
Date: Tue Nov 27 2018 - 16:09:02 EST


On Tue, 27 Nov 2018, Boris Ostrovsky wrote:
> On 11/27/18 3:37 PM, Stefano Stabellini wrote:
> > On Tue, 27 Nov 2018, PanBian wrote:
> >> On Mon, Nov 26, 2018 at 03:31:39PM -0500, Boris Ostrovsky wrote:
> >>> On 11/21/18 9:07 PM, Pan Bian wrote:
> >>>> kfree() is incorrectly used to release the pages allocated by
> >>>> __get_free_page() and __get_free_pages(). Use the matching deallocators
> >>>> i.e., free_page() and free_pages(), respectively.
> >>>>
> >>>> Signed-off-by: Pan Bian <bianpan2016@xxxxxxx>
> >>>> ---
> >>>> drivers/xen/pvcalls-front.c | 4 ++--
> >>>> 1 file changed, 2 insertions(+), 2 deletions(-)
> >>>>
> >>>> diff --git a/drivers/xen/pvcalls-front.c b/drivers/xen/pvcalls-front.c
> >>>> index 2f11ca7..77224d8 100644
> >>>> --- a/drivers/xen/pvcalls-front.c
> >>>> +++ b/drivers/xen/pvcalls-front.c
> >>>> @@ -385,8 +385,8 @@ static int create_active(struct sock_mapping *map, int *evtchn)
> >>>> out_error:
> >>>> if (*evtchn >= 0)
> >>>> xenbus_free_evtchn(pvcalls_front_dev, *evtchn);
> >>>> - kfree(map->active.data.in);
> >>>> - kfree(map->active.ring);
> >>>> + free_pages((unsigned long)map->active.data.in, PVCALLS_RING_ORDER);
> >>> Is map->active.data.in guaranteed to be NULL when entering this routine?
> >> I am not sure yet. Sorry for that. I observed the mismatches between
> >> __get_free_page and kfree, and submitted the patch.
> >>
> >> But I think your consideration is reasonable. A better solution is to
> >> directly free bytes, a local variable that holds __get_free_pages return
> >> value. If you agree, I will rewrite the patch.
> > Like Boris said, map->active.ring and map->active.data.in are not
> > guaranteed to be NULL or != NULL here. For instance,map->active.ring can
> > be != NULL and map->active.data.in can be NULL. However, free_pages and
> > free_page should be able to cope with it, the same way that kfree is
> > able to cope with it?
>
> If map->active.data.in can be non-NULL on entry to this routine then I
> think this has been a problem all along. Pan's suggestion to use bytes
> for freeing is going to solve this (assuming bytes will be initialized
> to NULL).

Why is it a problem? map->active.data.in and map->active.ring are only
!= NULL if they need to be freed. Otherwise, they are NULL. All structs
are always initialized to zero. I don't think there are any issues.