> -----Original Message----- > From: Steve Wise [mailto:swise@xxxxxxxxxxxxxxxxxxxxx] > Sent: Wednesday, February 28, 2018 9:53 AM > To: Parav Pandit <parav@xxxxxxxxxxxx>; Jason Gunthorpe > <jgg@xxxxxxxxxxxx> > Cc: 'Leon Romanovsky' <leon@xxxxxxxxxx>; dledford@xxxxxxxxxx; Yishai Hadas > <yishaih@xxxxxxxxxxxx>; linux-rdma@xxxxxxxxxxxxxxx > Subject: RE: [PATCH v3 rdma-next 07/10] mlx4_ib: set user mr attributes in struct > ib_mr > > > > > > > > > > > On Mon, Feb 26, 2018 at 03:22:38PM -0800, Steve Wise wrote: > > > > > > Setting iova, length, and page_size allows this information to > > > > > > be seen via NLDEV netlink queries, which can aid in user rdma > debugging. > > > > > > > > > > > > Signed-off-by: Steve Wise <swise@xxxxxxxxxxxxxxxxxxxxx> > > > > > > drivers/infiniband/hw/mlx4/mr.c | 3 +++ > > > > > > 1 file changed, 3 insertions(+) > > > > > > > > > > > > diff --git a/drivers/infiniband/hw/mlx4/mr.c > > > > > b/drivers/infiniband/hw/mlx4/mr.c > > > > > > index 4975f3e..17f4f15 100644 > > > > > > +++ b/drivers/infiniband/hw/mlx4/mr.c > > > > > > @@ -407,6 +407,9 @@ struct ib_mr *mlx4_ib_reg_user_mr(struct > > > > > > ib_pd *pd, > > > > > u64 start, u64 length, > > > > > > goto err_mr; > > > > > > > > > > > > mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key; > > > > > > + mr->ibmr.length = length; > > > > > > + mr->ibmr.iova = virt_addr; > > > > > > + mr->ibmr.page_size = 1U << shift; > > > > > > > > > > I wonder how many other drivers will have updates like this. > > > > > > > > I can't test the other drivers for lack of hardware. They can be > > > > fixed as the vendors discover them. OR, we don't dump any of this > > > > for MRs. And with the up and coming provider-specific resources, > > > > that will probably be dumped anyway. > > > > > > > > Should I just remove dumping of length/iova/page_size? > > > > > > We certainly have a problem if iova/length is uninited kernel memory > > > in > any > > > driver. :( > > > > > Given that iova and length are user provided inputs to register a MR, > shouldn't > > they be stored to mr->ibmr by ib_core and not by every provider drivers? > > The page_size is computed by the provider, so that field, at least should be filled > in by the provider. > Yes, so that can be in ib_mr.provider.page_size field that provider driver can fill up. > -- 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
