On Tue, Nov 20, 2018 at 03:47:38PM +0000, Ruhl, Michael J wrote: > >From: linux-rdma-owner@xxxxxxxxxxxxxxx [mailto:linux-rdma- > >owner@xxxxxxxxxxxxxxx] On Behalf Of Leon Romanovsky > >Sent: Monday, November 19, 2018 3:11 AM > >To: Doug Ledford <dledford@xxxxxxxxxx>; Jason Gunthorpe > ><jgg@xxxxxxxxxxxx> > >Cc: Leon Romanovsky <leonro@xxxxxxxxxxxx>; RDMA mailing list <linux- > >rdma@xxxxxxxxxxxxxxx> > >Subject: [PATCH rdma-next 03/10] RDMA/uverbs: Replace ib_uverbs_file with > >uverbs_attr_bundle for write > > > >From: Jason Gunthorpe <jgg@xxxxxxxxxxxx> > > > >Now that we can add meta-data to the description of write() methods we > >need to pass the uverbs_attr_bundle into all write based handlers so > >future patches can use it as a container for any new data transferred out > >of the core. > > > >This is the first step to bringing the write() and ioctl() methods to a > >common interface signature. > > > >This is a simple search/replace, and we push the attr down into the uobj > >and other APIs to keep changes minimal. > > > >Signed-off-by: Jason Gunthorpe <jgg@xxxxxxxxxxxx> > >Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx> > > drivers/infiniband/core/rdma_core.c | 13 +- > > drivers/infiniband/core/rdma_core.h | 8 +- > > drivers/infiniband/core/uverbs_cmd.c | 301 +++++++++++++------------- > > drivers/infiniband/core/uverbs_main.c | 12 +- > > drivers/infiniband/core/uverbs_uapi.c | 4 +- > > drivers/infiniband/hw/mlx5/devx.c | 2 +- > > include/rdma/ib_verbs.h | 2 +- > > include/rdma/uverbs_ioctl.h | 10 +- > > include/rdma/uverbs_std_types.h | 48 ++-- > > 9 files changed, 208 insertions(+), 192 deletions(-) > > > >diff --git a/drivers/infiniband/core/rdma_core.c > >b/drivers/infiniband/core/rdma_core.c > >index 752a55c6bdce..cf671a643d4c 100644 > >+++ b/drivers/infiniband/core/rdma_core.c > >@@ -224,12 +224,14 @@ int uobj_destroy(struct ib_uobject *uobj) > > * uverbs_put_destroy. > > */ > > struct ib_uobject *__uobj_get_destroy(const struct uverbs_api_object *obj, > >- u32 id, struct ib_uverbs_file *ufile) > >+ u32 id, > >+ const struct uverbs_attr_bundle *attrs) > > { > > struct ib_uobject *uobj; > > int ret; > > > >- uobj = rdma_lookup_get_uobject(obj, ufile, id, > >UVERBS_LOOKUP_DESTROY); > >+ uobj = rdma_lookup_get_uobject(obj, attrs->ufile, id, > >+ UVERBS_LOOKUP_DESTROY); > > Is there a reason that you didn't push attrs into rdma_lookup_get_uobject()? Not really.. In this case only uobj_* is being called from the write() handlers so I stopped pushing at that layer. It makes some sense that the rdma_* layer is more general than the uobj_* layer, but it could also be argued the other way. There are a lot of places where attr_bundle should be pushed down into, particularly into drivers. I was expecting all of that to be its own series. This thing is already very big. > Reviewed-by: Mike Ruhl <michael.j.ruhl@xxxxxxxxx> Thanks Jason
