Re: [PATCH] libceph, ceph: potential dereference of null pointer

[Xiubo Li]

On 12/9/21 7:20 PM, Jeff Layton wrote:
> On Thu, 2021-12-09 at 10:50 +0800, Jiasheng Jiang wrote:
> > The return value of kzalloc() needs to be checked.
> > To avoid use of null pointer in case of the failure of alloc.
> >
> > Fixes: 3d14c5d2b6e1 ("ceph: factor out libceph from Ceph file system")
> > Signed-off-by: Jiasheng Jiang <jiasheng@xxxxxxxxxxx>
> > ---
> >   net/ceph/osd_client.c | 2 ++
> >   1 file changed, 2 insertions(+)
> >
> > diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c
> > index ff8624a7c964..3203e8a34370 100644
> > --- a/net/ceph/osd_client.c
> > +++ b/net/ceph/osd_client.c
> > @@ -1234,6 +1234,8 @@ static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum)
> >   	WARN_ON(onum == CEPH_HOMELESS_OSD);
> >   
> >   	osd = kzalloc(sizeof(*osd), GFP_NOIO | __GFP_NOFAIL);
> > +	if (!osd)
> > +		return NULL;
> >   	osd_init(osd);
> >   	osd->o_osdc = osdc;
> >   	osd->o_osd = onum;
> __GFP_NOFAIL should ensure that it never returns NULL, right?

Yeah, from the comment, it make no sense to test for failure here:


204  * %__GFP_NOFAIL: The VM implementation _must_ retry infinitely: the 
caller
205  * cannot handle allocation failures. The allocation could block
206  * indefinitely but will never return with failure. Testing for
207  * failure is pointless.
208  * New users should be evaluated carefully (and the flag should be
209  * used only when there is no reasonable failure policy) but it is
210  * definitely preferable to use the flag rather than opencode endless
211  * loop around allocator.
212  * Using this flag for costly allocations is _highly_ discouraged.
213  */



> Also, if you're going to fix this up to handle that error then you
> probably also need to fix lookup_create_osd to handle a NULL return from
> create_osd as well.