[GIT PULL] Please pull RDMA subsystem changes

[Jason Gunthorpe <jgg@nvidia.com>]

[-- Attachment #1: Type: text/plain, Size: 3859 bytes --]

Hi Linus,

Usual rc pull, a whole bunch of driver fixes. bnxt_re has started
testing driver unload and fixed a bunch of bugs there.

Thanks,
Jason

The following changes since commit 2014c95afecee3e76ca4a56956a936e23283f05b:

  Linux 6.14-rc1 (2025-02-02 15:39:26 -0800)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma.git tags/for-linus

for you to fetch changes up to b66535356a4834a234f99e16a97eb51f2c6c5a7d:

  RDMA/bnxt_re: Fix the page details for the srq created by kernel consumers (2025-02-23 06:57:56 -0500)

----------------------------------------------------------------
RDMA v6.14 first rc pull request

- Fix a mlx5 malfunction if the UMR QP gets an error

- Return the correct port number to userspace for a mlx5 DCT

- Don't cause a UMR QP error if DMABUF teardown races with invalidation

- Fix a WARN splat when unregisering so mlx5 device memory MR types

- Use the correct alignment for the mana doorbell so that two processes
  do not share the same physical page on non-4k page systems

- MAINTAINERS updates for MANA

- Retry failed HNS FW commands because some can take a long time

- Cast void * handle to the correct type in bnxt to fix corruption

- Avoid a NULL pointer crash in bnxt_re

- Fix skipped ib_device_unregsiter() for bnxt_re due to some earlier
  rework

- Correctly detect if the bnxt supports extended statistics

- Fix refcount leak in mlx5 odp introduced by a previous fix

- Map the FW result for the port rate to the userspace values properly in
  mlx5, returns correct values for newer 800G ports

- Don't wrongly destroy counters objects that were not automatically
  created during mlx5 bind qp

- Set page size/shift members of kernel owned SRQs to fix a crash in nvme
  target

----------------------------------------------------------------
Junxian Huang (1):
      RDMA/hns: Fix mbox timing out by adding retry mechanism

Kalesh AP (3):
      RDMA/bnxt_re: Fix an issue in bnxt_re_async_notifier
      RDMA/bnxt_re: Add sanity checks on rdev validity
      RDMA/bnxt_re: Fix issue in the unload path

Kashyap Desai (1):
      RDMA/bnxt_re: Fix the page details for the srq created by kernel consumers

Konstantin Taranov (1):
      RDMA/mana_ib: Allocate PAGE aligned doorbell index

Long Li (1):
      MAINTAINERS: update maintainer for Microsoft MANA RDMA driver

Mark Zhang (1):
      IB/mlx5: Set and get correct qp_num for a DCT QP

Patrisious Haddad (2):
      RDMA/mlx5: Fix AH static rate parsing
      RDMA/mlx5: Fix bind QP error cleanup flow

Selvin Xavier (1):
      RDMA/bnxt_re: Fix the statistics for Gen P7 VF

Yishai Hadas (4):
      RDMA/mlx5: Fix the recovery flow of the UMR QP
      RDMA/mlx5: Fix a race for DMABUF MR which can lead to CQE with error
      RDMA/mlx5: Fix a WARN during dereg_mr for DM type
      RDMA/mlx5: Fix implicit ODP hang on parent deregistration

 MAINTAINERS                                 |  2 +-
 drivers/infiniband/hw/bnxt_re/bnxt_re.h     |  1 -
 drivers/infiniband/hw/bnxt_re/hw_counters.c |  4 +-
 drivers/infiniband/hw/bnxt_re/ib_verbs.c    |  2 +
 drivers/infiniband/hw/bnxt_re/main.c        | 22 ++++----
 drivers/infiniband/hw/bnxt_re/qplib_res.h   |  8 +++
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  | 64 ++++++++++++++++------
 drivers/infiniband/hw/hns/hns_roce_hw_v2.h  |  2 +
 drivers/infiniband/hw/mana/main.c           |  2 +-
 drivers/infiniband/hw/mlx5/ah.c             |  3 +-
 drivers/infiniband/hw/mlx5/counters.c       |  8 ++-
 drivers/infiniband/hw/mlx5/mr.c             | 16 +++++-
 drivers/infiniband/hw/mlx5/odp.c            |  1 +
 drivers/infiniband/hw/mlx5/qp.c             | 10 ++--
 drivers/infiniband/hw/mlx5/qp.h             |  1 +
 drivers/infiniband/hw/mlx5/umr.c            | 83 +++++++++++++++++++----------
 16 files changed, 161 insertions(+), 68 deletions(-)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]