IBoE support to Infiniband

IBoE allows running the IB transport protocol using Ethernet frames, enabling the deployment of IB semantics on lossless Ethernet fabrics.

IBoE packets are standard Ethernet frames with an IEEE assigned Ethertype, a GRH, unmodified IB transport headers and payload. IB subnet management and SA services are not required for IBoE operation; Ethernet management practices are used instead. IBoE resolves MAC addresses using the host IP stack. For multicast GIDs, standard IP to MAC mappings apply.

The OFA RDMA Verbs API is syntactically unmodified. The CMA is adapted to support IBoE ports allowing existing RDMA applications to run over IBoE with no changes.

Address handles for IBoE are required to contain valid L3 addresses (GIDs) and the IB L2 address fields become reserved. The complementary Ethernet L2 address information is subsequently derived below the API (currently, the Eth L2 information is encoded in the GID).

As there is no SA in IBoE, the CMA code is adapted to locally fill-in corresponding path record attributes for IBoE address handles. Also, the CMA provides the required address handle attributes for SIDR requests and joining of multicast groups.

With this patch set, each IBoE port is assigned a GID equal to the link local address of its corresponding net device, and one more GID for each one of the VLAN devices which are derived from it. iboe packets are tagged with the VLAN ID of the corresponding netdevice through which they are generated.

The priority field in the 802.1q header of IBoE packets is derived from the SL field in the address vector. rdma_cm applications can set the TOS value of the rdma_cm_id object through the rdma_set_option() API which then maps to SL.

With these patches, IBoE multicast frames may be broadcast as there is currently no use of a L2 multicast group membership protocol.

To enable IBoE with the mlx4 driver stack, both the mlx4_en and mlx4_ib drivers must be loaded, and the netdevice for the corresponding IBoE port must be running. Individual ports of a multi port HCA can be independently configured as Ethernet (with support for IBoE) or as IB, as it was already the case.

We have successfully tested MPI, SDP, RDS, and native Verbs applications over IBoE.

Following is a series of 12 patches based on Roland's iboe branch. This new series reflects changes based on feedback from the community on the previous patch set.

Changes from v9
1. Allow VLAN ID = 0 in kernel.
2. Small code modifications (details in the patches).

---

 drivers/infiniband/core/cma.c          |  282 ++++++++++++++++-
 drivers/infiniband/core/sa_query.c     |    5 
 drivers/infiniband/core/ucma.c         |   54 ++-
 drivers/infiniband/core/ud_header.c    |  158 +++++++--
 drivers/infiniband/core/user_mad.c     |   11 
 drivers/infiniband/core/uverbs_cmd.c   |    1 
 drivers/infiniband/hw/mlx4/ah.c        |  171 ++++++++--
 drivers/infiniband/hw/mlx4/mad.c       |   32 +
 drivers/infiniband/hw/mlx4/main.c      |  546 ++++++++++++++++++++++++++++++---
 drivers/infiniband/hw/mlx4/mlx4_ib.h   |   32 +
 drivers/infiniband/hw/mlx4/qp.c        |  201 +++++++++---
 drivers/infiniband/hw/mthca/mthca_qp.c |    2 
 drivers/net/mlx4/en_main.c             |   15 
 drivers/net/mlx4/en_netdev.c           |   10 
 drivers/net/mlx4/en_port.c             |    4 
 drivers/net/mlx4/en_port.h             |    3 
 drivers/net/mlx4/fw.c                  |    3 
 drivers/net/mlx4/intf.c                |   21 +
 drivers/net/mlx4/mlx4.h                |    1 
 drivers/net/mlx4/mlx4_en.h             |    1 
 drivers/net/mlx4/port.c                |   19 +
 include/linux/mlx4/cmd.h               |    1 
 include/linux/mlx4/device.h            |   31 +
 include/linux/mlx4/driver.h            |   16 
 include/linux/mlx4/qp.h                |    9 
 include/rdma/ib_addr.h                 |  134 ++++++++
 include/rdma/ib_pack.h                 |   29 +
 include/rdma/ib_user_verbs.h           |    3 
 28 files changed, 1598 insertions(+), 197 deletions(-)

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

May be? They are broadcast because there is no IGMP like support? Any hopes of multicast support soon?

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

I came across your ofed commit e5414cccaa13e6dd80d8d6fc3dafe95355facdef "sdp: module parameter to disable SDP over ROCEE" and wasn't sure what's behind it, can you clarify that?

Or.

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

(Resending without HTML stuff)

It is from the days that SDP over RoCE wasn't stable.
In addition, when customers had a very long delay before TCP connection established, in the following scenario:

1. in libsdp.conf, setting mode to 'both' (Try SDP and fallback to TCP)
2. application tries to open socket to a remote peer connected using 10G ethernet
3. Remote host don't support RoCE.

It took few seconds till the CMA gave up trying to connect, and SDP connection failed, and TCP connection was established.

We plan to change the default state of this module parameter to enable SDP over RoCE, and to add the option to disable SDP per interface. Currently a user can only disable SDP on a subnet/port/application.

- Amir

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

thanks for the clarification re remote host not supporting IBoE, anyway, I don't see if/what this has to do with sdp stability, its just a delay on the connection establishment and you say the default is to be changed to off for this param.

Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

Yes, it will be changed.

- Amir

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html