Date:	Wed, 24 Feb 2016 13:59:50 +1100
From:	Gavin Shan <gwshan@...ux.vnet.ibm.com>
To:	Gavin Shan <gwshan@...ux.vnet.ibm.com>
Cc:	netdev@...r.kernel.org, benh@...nel.crashing.org,
	davem@...emloft.net, sergei.shtylyov@...entembedded.com
Subject: Re: [PATCH RFC 0/6] NCSI Support

On Mon, Nov 09, 2015 at 11:10:00AM +1100, Gavin Shan wrote:

Hello David, did you get time to go through this series? It has been there
for a while. Looking forward to your comments and feedback :-)

Thanks,
Gavin

>This series of patches are prototype requesting for comments. Please focus on
>the big picture at current stage, but any comments to improve the implementaion
>are welcomed.
>
>The following figure gives an example about how NCSI is deployed: The NCSI is
>specified by DSP0222, which can be downloaded from the following link here
>(http://www.dmtf.org/sites/default/files/standards/documents/DSP0222_1.0.0.pdf).
>
>   * The NC-SI (aka NCSI) is defined as the interface between a (Base) Management
>     Controller (BMC) and one or multiple Network Controlers (NC) on host side.
>     The interface is responsible for providing external network connectivity
>     for BMC.
>   * Each BMC can connect to multiple packages, up to 8. Each package can have
>     multiple channels, up to 32. Every package and channel are identified by
>     3-bits and 5-bits in NCSI packet. At one moment, one channel is active to
>     provide service.
>   * NCSI packet, encapsulated in ethernet frame, has 0x88F8 in the protocol
>     field. The destination MAC address should be 0xFF's while the source MAC
>     address can be arbitrary one.
>   * NCSI packets are classified to command, response, AEN (Asynchronous Event
>     Notification). Commands are sent from BMC to host for configuration and
>     information retrival. Responses, corresponding to commands, are sent from
>     host to BMC for confirmation and requested information. One command should
>     have one and only one response. AEN is sent from host to BMC for notification
>     (e.g. link down on active channel) so that BMC can take appropriate action.
>
>   +------------------+        +----------------------------------------------+
>   |                  |        |                     Host                     |
>   |        BMC       |        |                                              |
>   |                  |        | +-------------------+  +-------------------+ |
>   |    +---------+   |        | |     Package-A     |  |     Package-B     | |
>   |    |         |   |        | +---------+---------+  +-------------------+ |
>   |    |   NIC   |   |        | | Channel | Channel |  | Channel | Channel | |
>   +----+----+----+---+        +-+---------+---------+--+---------+---------+-+
>             |                             |                      |
>             |                             |                      |
>             +-----------------------------+----------------------+           
>
>The design for the patchset is highlighted as below:
>
>   * All the code included in this patchset runs on BMC side. The major target
>     of this patchset is to hide the NCSI details to BMC's NIC driver. The patchset
>     abstracts a NCSI device (struct ncsi_dev), which is created when the driver
>     is loaded and the BMC NIC (struct net_device) instance is created. When the
>     NIC is opened (ndo_open()), all available packages and channels are probed
>     through NCSI packets. At the same time, one active channel is selected to
>     provide service. It means the NIC driver can receive/transmit packets before
>     ndo_open().
>   * When NCSI device (struct ncsi_dev) is created when NIC driver is loaded,
>     NCSI protocl handler is added by dev_add_pack() so that the ingress NCSI
>     packets can be routed to the stack for further processing. On the other
>     hand, dev_queue_xmit_sk() is called to transmit skb that has been bound
>     with the BMC's NIC.
>   * The NCSI device is closed when the BMC's NIC is closed.
>   * NCSI stack should be configurable through netlink, but it's not implemented
>     in this patchset. It's something TBD.
>   * The first NIC driver that is aware of NCSI: drivers/net/ethernet/faraday/ftgmac100.c
>
>Gavin Shan (6):
>  net/ncsi: Resource management
>  net/ncsi: Packet handler
>  net/ncsi: Manage NCSI device
>  net/faraday: Replace use_nc_si with use_ncsi
>  net/faraday: Enable NCSI interface
>  net/faraday: Enable offload checksum according to device-tree
>
> drivers/net/ethernet/faraday/ftgmac100.c |  108 ++-
> include/net/ncsi.h                       |   57 ++
> include/uapi/linux/if_ether.h            |    1 +
> net/Kconfig                              |    1 +
> net/Makefile                             |    1 +
> net/ncsi/Kconfig                         |   10 +
> net/ncsi/Makefile                        |    5 +
> net/ncsi/internal.h                      |  313 ++++++++
> net/ncsi/ncsi-aen.c                      |  197 +++++
> net/ncsi/ncsi-cmd.c                      |  372 ++++++++++
> net/ncsi/ncsi-manage.c                   |  899 +++++++++++++++++++++++
> net/ncsi/ncsi-pkt.h                      |  391 ++++++++++
> net/ncsi/ncsi-rsp.c                      | 1166 ++++++++++++++++++++++++++++++
> 13 files changed, 3490 insertions(+), 31 deletions(-)
> create mode 100644 include/net/ncsi.h
> create mode 100644 net/ncsi/Kconfig
> create mode 100644 net/ncsi/Makefile
> create mode 100644 net/ncsi/internal.h
> create mode 100644 net/ncsi/ncsi-aen.c
> create mode 100644 net/ncsi/ncsi-cmd.c
> create mode 100644 net/ncsi/ncsi-manage.c
> create mode 100644 net/ncsi/ncsi-pkt.h
> create mode 100644 net/ncsi/ncsi-rsp.c
>
>-- 
>2.1.0
>

Powered by blists - more mailing lists