On Wed, May 06, 2020 at 07:56:08PM +0300, Leon Romanovsky wrote:
> On Wed, May 06, 2020 at 11:43:44AM -0300, Jason Gunthorpe wrote:
> > On Wed, May 06, 2020 at 08:32:13AM +0300, Leon Romanovsky wrote:
> > > From: Jack Morgenstein <jackm@xxxxxxxxxxxxxxxxxx>
> > >
> > > The IB core pkey cache is populated by procedure ib_cache_update().
> > > Initially, the pkey cache pointer is NULL. ib_cache_update allocates
> > > a buffer and populates it with the device's pkeys, via repeated calls
> > > to procedure ib_query_pkey().
> > >
> > > If there is a failure in populating the pkey buffer via ib_query_pkey(),
> > > ib_cache_update does not replace the old pkey buffer cache with the
> > > updated one -- it leaves the old cache as is.
> > >
> > > Since initially the pkey buffer cache is NULL, when calling
> > > ib_cache_update the first time, a failure in ib_query_pkey() will cause
> > > the pkey buffer cache pointer to remain NULL.
> > >
> > > In this situation, any calls subsequent to ib_get_cached_pkey(),
> > > ib_find_cached_pkey(), or ib_find_cached_pkey_exact() will try to
> > > dereference the NULL pkey cache pointer, causing a kernel panic.
> > >
> > > Fix this by checking the ib_cache_update() return value.
> > >
> > > Fixes: 8faea9fd4a39 ("RDMA/cache: Move the cache per-port data into the main ib_port_data")
> > > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> > > Signed-off-by: Jack Morgenstein <jackm@xxxxxxxxxxxxxxxxxx>
> > > Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
> > > Changelog:
> > > v1: I rewrote the patch to take care of ib_cache_update() return value.
> > > v0: https://lore.kernel.org/linux-rdma/20200426075811.129814-1-leon@xxxxxxxxxx
> > > drivers/infiniband/core/cache.c | 11 +++++++++--
> > > 1 file changed, 9 insertions(+), 2 deletions(-)
> > >
> > >
> > > diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c
> > > index 717b798cddad..1cbebfa374a5 100644
> > > +++ b/drivers/infiniband/core/cache.c
> > > @@ -1553,10 +1553,17 @@ int ib_cache_setup_one(struct ib_device *device)
> > > if (err)
> > > return err;
> > >
> > > - rdma_for_each_port (device, p)
> > > - ib_cache_update(device, p, true);
> > > + rdma_for_each_port (device, p) {
> > > + err = ib_cache_update(device, p, true);
> > > + if (err)
> > > + goto out;
> > > + }
> > >
> > > return 0;
> > > +
> > > +out:
> > > + ib_cache_release_one(device);
> > > + return err;
> >
> > ib_cache_release_once can be called only once, and it is always called
> > by ib_device_release(), it should not be called here
>
> It doesn't sound right if we rely on ib_device_release() to unwind error
> in ib_cache_setup_one(). I don't think that we need to return from
> ib_cache_setup_one() without cleaning it.
We do as ib_cache_release_one() cannot be called multiple times
The general design of all this pre-registration stuff is that the
release function does the clean up and the individual functions should
not error unwind cleanup done in the unconditional release.
Other schemes were too complicated
Jason
