| Message ID | 20181128182034.32618.71330.stgit@scvm10.sc.intel.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | IB/hfi1: Start to add in OPFN | expand |
On Wed, Nov 28, 2018 at 10:21:54AM -0800, Dennis Dalessandro wrote: > Hi Doug and Jason, > > This is the start of some of the prep work for TID RDMA. We have other patches > to follow but this seems to stand on it's own so we wanted to post while we are > still reworking the other patches. Holiday's and travels, and vacations means > things are a bit more delayed, but the rest of the patches are on the way. > > I think this could go in now to the driver, even though it won't be used yet. > The reason is that as other TID patches continue to flow in there is going to > be a lot of conflicts and it will make things easier to take them in as we go. > > > Kaike Wan (2): > IB/hfi1: Unreserve a reserved request when it is completed > IB/hfi1: Add OPFN and TID RDMA capability bits > > Mike Marciniszyn (1): > IB/hfi1: Allow the driver to initialize QP priv struct Okay, applied to for-next, but the 'Unreserve a reserved request' looks like another -rc patch if it fixes a WARN_ON? Jason
On 12/6/2018 10:12 PM, Jason Gunthorpe wrote: > On Wed, Nov 28, 2018 at 10:21:54AM -0800, Dennis Dalessandro wrote: >> Hi Doug and Jason, >> >> This is the start of some of the prep work for TID RDMA. We have other patches >> to follow but this seems to stand on it's own so we wanted to post while we are >> still reworking the other patches. Holiday's and travels, and vacations means >> things are a bit more delayed, but the rest of the patches are on the way. >> >> I think this could go in now to the driver, even though it won't be used yet. >> The reason is that as other TID patches continue to flow in there is going to >> be a lot of conflicts and it will make things easier to take them in as we go. >> >> >> Kaike Wan (2): >> IB/hfi1: Unreserve a reserved request when it is completed >> IB/hfi1: Add OPFN and TID RDMA capability bits >> >> Mike Marciniszyn (1): >> IB/hfi1: Allow the driver to initialize QP priv struct > > Okay, applied to for-next, but the 'Unreserve a reserved request' > looks like another -rc patch if it fixes a WARN_ON? This all falls out of the TID submission that Kaike is reworking. I don't think there is any issue until TID as it is the first thing that will do this reserved operations stuff so as long as this fix goes in before the other TID patches (not yet re-posted) we should be good. -Denny
Hi Doug and Jason, This is the start of some of the prep work for TID RDMA. We have other patches to follow but this seems to stand on it's own so we wanted to post while we are still reworking the other patches. Holiday's and travels, and vacations means things are a bit more delayed, but the rest of the patches are on the way. I think this could go in now to the driver, even though it won't be used yet. The reason is that as other TID patches continue to flow in there is going to be a lot of conflicts and it will make things easier to take them in as we go. --- Kaike Wan (2): IB/hfi1: Unreserve a reserved request when it is completed IB/hfi1: Add OPFN and TID RDMA capability bits Mike Marciniszyn (1): IB/hfi1: Allow the driver to initialize QP priv struct drivers/infiniband/hw/hfi1/Makefile | 1 + drivers/infiniband/hw/hfi1/common.h | 19 ++++++++----- drivers/infiniband/hw/hfi1/rc.c | 2 + drivers/infiniband/hw/hfi1/tid_rdma.c | 48 +++++++++++++++++++++++++++++++++ drivers/infiniband/hw/hfi1/tid_rdma.h | 13 +++++++++ drivers/infiniband/hw/hfi1/verbs.c | 1 + drivers/infiniband/hw/hfi1/verbs.h | 2 + drivers/infiniband/sw/rdmavt/qp.c | 9 ++++++ include/rdma/rdma_vt.h | 7 +++++ include/uapi/rdma/hfi/hfi1_user.h | 6 ++-- 10 files changed, 96 insertions(+), 12 deletions(-) create mode 100644 drivers/infiniband/hw/hfi1/tid_rdma.c create mode 100644 drivers/infiniband/hw/hfi1/tid_rdma.h -- -Denny