On Mon, Apr 02, 2018 at 03:18:26PM +0000, Ruhl, Michael J wrote: > >-----Original Message----- > >From: linux-rdma-owner@xxxxxxxxxxxxxxx [mailto:linux-rdma- > >owner@xxxxxxxxxxxxxxx] On Behalf Of Leon Romanovsky > >Sent: Sunday, April 1, 2018 8:08 AM > >To: Doug Ledford <dledford@xxxxxxxxxx>; Jason Gunthorpe > ><jgg@xxxxxxxxxxxx> > >Cc: Leon Romanovsky <leonro@xxxxxxxxxxxx>; RDMA mailing list <linux- > >rdma@xxxxxxxxxxxxxxx>; Ariel Levkovich <lariel@xxxxxxxxxxxx>; Saeed > >Mahameed <saeedm@xxxxxxxxxxxx> > >Subject: [PATCH rdma-next v1 4/8] IB/uverbs: Add device memory registration > >ioctl support > > > >From: Ariel Levkovich <lariel@xxxxxxxxxxxx> > > > >Adding new ioctl method for the MR object - REG_DM_MR. > > > >This command can be used by users to register an allocated > >device memory buffer as an MR and receive lkey and rkey > >to be used within work requests. > > > >It is added as a new method under the MR object and using a new > >ib_device callback - reg_dm_mr. > >The command creates a standard ib_mr object which represents the > >registered memory. > > > >Signed-off-by: Ariel Levkovich <lariel@xxxxxxxxxxxx> > >Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx> > >--- > >++++++++++++++++++++++++++ > > include/rdma/uverbs_ioctl.h | 12 +++ > > 7 files changed, 186 insertions(+), 12 deletions(-) > > create mode 100644 drivers/infiniband/core/uverbs_std_types_mr.c > > > >diff --git a/include/rdma/uverbs_ioctl.h b/include/rdma/uverbs_ioctl.h > >index 3d6ac684b8f0..4a4201d997a7 100644 > >--- a/include/rdma/uverbs_ioctl.h > >+++ b/include/rdma/uverbs_ioctl.h > >@@ -408,6 +408,18 @@ static inline int uverbs_attr_get_enum_id(const struct > >uverbs_attr_bundle *attrs > > return attr->ptr_attr.enum_id; > > } > > > >+static inline void *uverbs_attr_get_obj(const struct uverbs_attr_bundle > >*attrs_bundle, > >+ u16 idx) > >+{ > >+ struct ib_uobject *uobj = > >+ uverbs_attr_get(attrs_bundle, idx)->obj_attr.uobject; > >+ > >+ if (IS_ERR(uobj)) > >+ return uobj; > >+ > >+ return uobj->object; > >+} > >+ > > I know that this patch set is being added augment the functionality of > the base IOCTL interface, but this change appears to be very useful to > the overall usage of the IOCTL interface. > > Should new, general functions like this, be split into separate patches, > so that if the main patches are not accepted (or they morph into something > else), utilities like uverbs_attr_get_object() are not lost in the shuffle? Let's be positive and hope that the current patch set in good shape to be accepted. Anyway I don't expect to see acceptance of "sub-patches" without main user. Thanks
Attachment:
signature.asc
Description: PGP signature
