On 10/29/10 5:44 PM, Kiran Patil wrote:
> The following series implements...
>
> 1) Plug into TCM and extend lport's params
> 2) Enhanced exchangeID selection algorithm
> 3) Enhanced EMA selection algorithm
> 4) Added support for "offload" in TCM target and fix related bug
> 5) Enhancement in base driver to support offload for targets.
>
> It also contain RFC (Comments primarily from Joe) fixes..
> 1) RPORT state machine fix is not required. Removed the respective patch.
> 2) Cleanup of function fcoe_oem_match, fc_find_ema, and fc_exch_recv.
> 3) Tab spacing fix.
> 4) Removal of comment from function fcoe_oem_match.
>
> Dependencies: This series depends on TCM and Joe's FC4 patches for TCM.
>
> ---
>
> Kiran Patil (5):
> Base Driver: Added support for receive offload in base driver for
> storage target.
> TCM/LIO: Support for large FC offload (write exchange)
> fcoe: Unable to select the exchangeID from offload pool for target
> when write request coming from initiator.
> libfc: Enhanced exchange ID selection mechanism and fix related EMA
> selection logic.
> libfc: Target roles isn't target even though there is a registered
> target.
>
>
> drivers/net/ixgbe/ixgbe_fcoe.c | 26 +++++++++++++
> drivers/scsi/fcoe/fcoe.c | 15 +++++++-
> drivers/scsi/libfc/fc_exch.c | 64
> ++++++++++++++++++++++----------
> drivers/scsi/libfc/fc_libfc.c | 19 ++++++++++
> drivers/scsi/libfc/fc_libfc.h | 1 +
> drivers/scsi/libfc/fc_lport.c | 1 +
> drivers/target/target_core_transport.c | 8 ++++
> drivers/target/tcm_fc/tcm_fc.h | 3 ++
> drivers/target/tcm_fc/tfc_cmd.c | 33 ++++++++++++++++-
> drivers/target/tcm_fc/tfc_conf.c | 4 ++
> drivers/target/tcm_fc/tfc_io.c | 64
> ++++++++++++++++++++++++++++++++
> 11 files changed, 212 insertions(+), 26 deletions(-)
>
I won't get to review the rest of this until Monday or Tuesday.
Just a thought: TCM/LIO could be modified to do the write to the
back end device from the buffers we give it. That way, no DDP is needed.
This would work without a copy in most cases, except for ram disk,
perhaps, where it may want to keep the buffers itself.
Joe
_______________________________________________
devel mailing list
[email protected]
http://www.open-fcoe.org/mailman/listinfo/devel
