On Mon, Mar 12, 2018 at 09:22:32AM -0600, Jason Gunthorpe wrote:
> On Mon, Mar 12, 2018 at 04:16:16PM +0200, Leon Romanovsky wrote:
> > From: Leon Romanovsky <leonro@xxxxxxxxxxxx>
> >
> > Hi,
> >
> > The reason to this patchset is described in "RDMA/mlx4: Clear all
> > allocated memory by default" accommodated with relevant crash report.
>
> Well, but you never explained what the actual problem is, and I've failed
> to guess...
109 void rdma_restrack_add(struct rdma_restrack_entry *res)
110 {
111 struct ib_device *dev = res_to_dev(res);
112
113 if (!dev)
114 return;
115
116 if (res_is_user(res)) {
117 if (!res->task)
^^^^^^^^^^^ this part wasn't zero in new objects,
sometimes it failed in QP (mlx4), sometimes it failed in CQ (mlx5) and my
grep showed that other drivers in worst situation.
>
> So I can't judge if changing every alloc in every driver is a sane
> idea or some kind of overkill..
Please proceed at least with mlx4/mlx5/rxe/cxgb3/cxbg4 drivers, I want
them correct without any relation to current state of supported
objects.
Thanks
>
> Jason
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Attachment:
signature.asc
Description: PGP signature
