mbox series

[for-next,V5,0/7] Add support for Broadcom's 57500 series of adapters

Message ID 1549521088-6903-1-git-send-email-devesh.sharma@broadcom.com (mailing list archive)
Headers show
Series Add support for Broadcom's 57500 series of adapters | expand

Message

Devesh Sharma Feb. 7, 2019, 6:31 a.m. UTC
This is to enable RoCE on Broadcom's 57500 series of adapters.
Patch 0001, 0002 and 0003 are handing the control path changes.
Patch 0004 and 0005 are related to kernel space fast path.
Patch 0006 handles the user-kernel ABI changes. Patch 0007 is
to enable RoCE driver load on next gen chips.

This patch series requires a patch from Linus git:
commit 78793afbb0b9 ("bnxt_en: Increase context memory allocations on
                      57500 chips for RDMA.")

Changelog:

V4->V5
rebased the series to the tip of for-next.
Patch 0002:
	- cleaned up the DB ring code to address muti-platform
          mmio-write compatibility issue.
V3->V4
rebased the series to the tip of for-next.
Patch 0002:
        - dropped use of __iowrite64_copy, use writeq instead.
Patch 0006:
        - use __aligned_u64 instead of __u64.
        - fix sizeof(resp) when copying response to user.
V2->V3
Patch 0006:
        - Implemented the comp_mask approach instead of changing the
          ABI version number.
V1->V2
Rebased the series to the tip of for-next and fixed widespread
code alignment.
Patch 0001:
        - Replace chip_ctx pointer with static member in bnxt_re_dev
          structure.
Patch 0002:
        - Fixed kbuild error on i386 arch. Using "depends on 64BIT"
        - Removed wmb before calling writeq in db-ring functions
Patch 0003:
        - Fixed typo in the commit message
Patch 0004:
	- Removed endian-ness fix from feature series.
Patch 0005:
        - Fixed code formatting issues
Patch 0006:
        - Implemented ABI range check as suggested by Jason and Leon

Devesh Sharma (7):
  RDMA/bnxt_re: Add chip context to identify 57500 series
  RDMA/bnxt_re: Add 64bit doorbells for 57500 series
  RDMA/bnxt_re: Skip backing store allocation for 57500 series
  RDMA/bnxt_re: Enable GSI QP support for 57500 series
  RDMA/bnxt_re: Add extended psn structure for 57500 adapters
  RDMA/bnxt_re: Update kernel user abi to pass chip context
  RDMA/bnxt_en: Enable RDMA driver support for 57500 chip

 drivers/infiniband/hw/bnxt_re/Kconfig         |   1 +
 drivers/infiniband/hw/bnxt_re/bnxt_re.h       |   1 +
 drivers/infiniband/hw/bnxt_re/ib_verbs.c      | 132 +++++++++++++-------
 drivers/infiniband/hw/bnxt_re/main.c          | 122 +++++++++++++-----
 drivers/infiniband/hw/bnxt_re/qplib_fp.c      | 173 +++++++++++++++-----------
 drivers/infiniband/hw/bnxt_re/qplib_fp.h      |  47 ++++++-
 drivers/infiniband/hw/bnxt_re/qplib_rcfw.c    |  37 ++++--
 drivers/infiniband/hw/bnxt_re/qplib_rcfw.h    |  45 ++++++-
 drivers/infiniband/hw/bnxt_re/qplib_res.c     |  10 +-
 drivers/infiniband/hw/bnxt_re/qplib_res.h     |  30 ++++-
 drivers/infiniband/hw/bnxt_re/qplib_sp.c      |   3 +-
 drivers/infiniband/hw/bnxt_re/roce_hsi.h      | 160 +++++++++++++++---------
 drivers/net/ethernet/broadcom/bnxt/bnxt_ulp.c |   3 -
 include/uapi/rdma/bnxt_re-abi.h               |  11 ++
 14 files changed, 535 insertions(+), 240 deletions(-)

Comments

Jason Gunthorpe Feb. 7, 2019, 8:51 p.m. UTC | #1
On Thu, Feb 07, 2019 at 01:31:21AM -0500, Devesh Sharma wrote:
> This is to enable RoCE on Broadcom's 57500 series of adapters.
> Patch 0001, 0002 and 0003 are handing the control path changes.
> Patch 0004 and 0005 are related to kernel space fast path.
> Patch 0006 handles the user-kernel ABI changes. Patch 0007 is
> to enable RoCE driver load on next gen chips.
> 
> This patch series requires a patch from Linus git:
> commit 78793afbb0b9 ("bnxt_en: Increase context memory allocations on
>                       57500 chips for RDMA.")

This is the wrong commit ID

> Changelog:
> 
> V4->V5
> rebased the series to the tip of for-next.
> Patch 0002:
> 	- cleaned up the DB ring code to address muti-platform
>           mmio-write compatibility issue.
> V3->V4
> rebased the series to the tip of for-next.
> Patch 0002:
>         - dropped use of __iowrite64_copy, use writeq instead.
> Patch 0006:
>         - use __aligned_u64 instead of __u64.
>         - fix sizeof(resp) when copying response to user.
> V2->V3
> Patch 0006:
>         - Implemented the comp_mask approach instead of changing the
>           ABI version number.
> V1->V2
> Rebased the series to the tip of for-next and fixed widespread
> code alignment.
> Patch 0001:
>         - Replace chip_ctx pointer with static member in bnxt_re_dev
>           structure.
> Patch 0002:
>         - Fixed kbuild error on i386 arch. Using "depends on 64BIT"
>         - Removed wmb before calling writeq in db-ring functions
> Patch 0003:
>         - Fixed typo in the commit message
> Patch 0004:
> 	- Removed endian-ness fix from feature series.
> Patch 0005:
>         - Fixed code formatting issues
> Patch 0006:
>         - Implemented ABI range check as suggested by Jason and Leon
> 
> Devesh Sharma (7):
>   RDMA/bnxt_re: Add chip context to identify 57500 series
>   RDMA/bnxt_re: Add 64bit doorbells for 57500 series
>   RDMA/bnxt_re: Skip backing store allocation for 57500 series
>   RDMA/bnxt_re: Enable GSI QP support for 57500 series
>   RDMA/bnxt_re: Add extended psn structure for 57500 adapters
>   RDMA/bnxt_re: Update kernel user abi to pass chip context
>   RDMA/bnxt_en: Enable RDMA driver support for 57500 chip

This seems fine now, applied to for-next

Thanks,
Jason