Please pull rdma.git

Hi Linus,

This comprises a number of various fixes (including the ones you've requested):

The following changes since commit bc197aad0daa:

Linux 4.2-rc2 (2015-07-12 15:10:30 -0700)

are available in the git repository at:

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

for you to fetch changes up to d8b2ba7c5928173fe1c12bd2545f5ed85d1c3c7a:

IB/core: Destroy ocrdma_dev_id IDR on module exit (2015-07-14 13:20:16
-0400)

----------------------------------------------------------------
Changes for 4.2-rc

- Mainly fix-ups for the various 4.2 items

----------------------------------------------------------------
Amir Vadai (1):
IB/IPoIB: Fix bad error flow in ipoib_add_port()

Carol L Soto (1):
IB/ucm: Fix bitmap wrap when devnum > IB_UCM_MAX_DEVICES

Doug Ledford (2):
IB/mlx4: Fix memory leak in do_slave_init
IB/mlx4: Optimize do_slave_init

Erez Shitrit (2):
IB/cm: Do not queue work to a device that's going away IB/ipoib: Set MTU to max allowed by mode when mode changes

Haggai Eran (2):
IB/ucma: Fix lockdep warning in ucma_lock_files IB/ipoib: Prevent lockdep warning in __ipoib_ib_dev_flush

Hal Rosenstock (1):
IB: Add rdma_cap_ib_switch helper and use where appropriate

Ira Weiny (2):
IB/mad: Fix compare between big endian and cpu endian IB/mad: Remove improper use of BUG_ON

Johannes Thumshirn (2):
IB/core: Destroy multcast_idr on module exit IB/core: Destroy ocrdma_dev_id IDR on module exit

Maninder Singh (1):
IB/mlx4: Optimize freeing of items on error unwind

Matan Barak (1):
IB/mlx4: Do not attemp to report HCA clock offset on VFs

Or Gerlitz (1):
IB/mlx4: Fix use of flow-counters for process_mad

Sagi Grimberg (1):
IB/srp: Avoid using uninitialized variable

Tatyana Nikolova (3):
RDMA/core: Fixes for port mapper client registration RDMA/nes: Fix for resolving the neigh RDMA/nes: Fix for incorrect recording of the MAC address

Vaishali Thakkar (2):
IB/srpt: Convert use of __constant_cpu_to_beXX to cpu_to_beXX IB/ipath: Convert use of __constant_<foo> to <foo>

Wengang Wang (1):
rds: rds_ib_device.refcount overflow

Yuval Shaia (1):
IB/ipoib: Scatter-Gather support in connected mode

 drivers/infiniband/core/agent.c            |  4 +-
 drivers/infiniband/core/cm.c               | 61 ++++++++++++++++++++++---
 drivers/infiniband/core/iwpm_msg.c         | 33 +++++++-------
 drivers/infiniband/core/iwpm_util.c        | 12 ++++-
 drivers/infiniband/core/iwpm_util.h        | 28 +++++++++---
 drivers/infiniband/core/mad.c              | 47 +++++++-------------
 drivers/infiniband/core/multicast.c        |  8 +---
 drivers/infiniband/core/opa_smi.h          |  4 +-
 drivers/infiniband/core/sa_query.c         |  8 +---
 drivers/infiniband/core/smi.c              | 37 ++++++++--------
 drivers/infiniband/core/smi.h              |  4 +-
 drivers/infiniband/core/sysfs.c            |  2 +-
 drivers/infiniband/core/ucm.c              |  4 +-
 drivers/infiniband/core/ucma.c             |  5 ++-
 drivers/infiniband/hw/ehca/ehca_sqp.c      |  5 ++-
 drivers/infiniband/hw/ipath/ipath_mad.c    |  5 ++-
 drivers/infiniband/hw/ipath/ipath_verbs.c  |  4 +-
 drivers/infiniband/hw/mlx4/mad.c           | 34 +++++++++-----
 drivers/infiniband/hw/mlx4/main.c          | 33 +++++++-------
 drivers/infiniband/hw/mlx5/mad.c           |  5 ++-
 drivers/infiniband/hw/mthca/mthca_mad.c    |  5 ++-
 drivers/infiniband/hw/nes/nes_cm.c         |  5 ++-
 drivers/infiniband/hw/nes/nes_hw.c         |  2 +-
 drivers/infiniband/hw/ocrdma/ocrdma_ah.c   |  5 ++-
 drivers/infiniband/hw/ocrdma/ocrdma_main.c |  1 +
 drivers/infiniband/hw/qib/qib_mad.c        |  5 ++-
 drivers/infiniband/ulp/ipoib/ipoib.h       | 29 +++++++++++-
 drivers/infiniband/ulp/ipoib/ipoib_cm.c    | 33 ++++++--------
 drivers/infiniband/ulp/ipoib/ipoib_ib.c    | 49 ++++++++-------------
 drivers/infiniband/ulp/ipoib/ipoib_main.c  | 21 ++++-----
 drivers/infiniband/ulp/srp/ib_srp.c        | 23 +++-------
 drivers/infiniband/ulp/srpt/ib_srpt.c      | 71
+++++++++++++++---------------
 drivers/scsi/scsi_transport_srp.c          |  3 +-
 include/rdma/ib_verbs.h                    | 20 +++++++--
 include/scsi/scsi_transport_srp.h          |  1 +
 net/rds/ib_rdma.c                          |  4 +-
 36 files changed, 351 insertions(+), 269 deletions(-)

Hmm. I've pulled this, but quite frankly, I don't think this was appropriate for post-merge-window. It's not at all obvious that that "rdma_cap_ib_switch helper" thing is a bugfix, and that goes for a few of the other commits too. Some of them are _clearly_ not actual bug-fixes, but just random commits.

If this is going to be a pattern, I'm going to stop pulling things. But as a one-off, I let it pass.

Linus
--
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

rdma_cap_ib_switch isn't even a bug fix, it clearly adds (or shuffles) dead code..
--
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

In the 4.2 merge window we added the code that mostly eliminated our reliance on node_type as anything of value (instead of checking node_type + link_layer to determine if certain features were supported, we added a bitmap of the features themselves and now check that bitmap).
Then I went on PTO for several weeks during the merge window. While I
was gone, the rdma_cap_ib_switch helper patch was discussed and written on the mailing list. When I got back (which was just before the first rc), it took me a bit of time to get all of the appropriate patches pulled together and submitted. I included that code because it was the last remnant of node_type still in use, was a use that the original patchset didn't address, and taking it allowed us to go from using node_type to not using node_type at all in one kernel version instead of going from using node_type, to using node_type for switches only, to not using node_type at all across a series of two kernels.

Doug Ledford <[email protected]>
              GPG KeyID: 0E572FDD

Hi Doug,

the point is that there is no driver that even set the is_switch flag in-tree and I can't find a publicly available one elsewhere either.

So it's plain and simple dead code. I'll send a patch to kill this untestable code off once I find a little spare time.
--
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

Try looking on Mellanox.com for their source code. I know they use upstream linux kernels as the base for their switch OS and in their switch-X driver it should set this flag.

However, this is a point that I don't really care about, but I inherited (and I know other people care about). This code has supported switches via out-of-tree drivers since as long as I know. That was an early decision that Roland made. Removing this code is more than just removing this code, it is reversing that decision.

Doug Ledford <[email protected]>
              GPG KeyID: 0E572FDD

There are 2 vendors that I'm aware of that use linux for their embedded IB switch platforms: Mellanox and Bay Microsystems and there may be others.

Mellanox has switch drivers for the various different switch ICs that have been used over time (InfiniScale IV, SwitchX/SwitchX-2, and most Switch-IB).

-- Hal
--
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 Hal, we indeed do. And we would like to continue to do so, be it by setting a flag or otherwise. I hope removing that functionality from the kernel is not an option, as it would be a major inconvenience for us. Thanks,
Suri

-----Original Message-----
From: Hal Rosenstock [mailto:[email protected]]
Sent: Thursday, July 16, 2015 9:23 AM
To: Doug Ledford
Cc: Christoph Hellwig; Linus Torvalds; linux-rdma; Suri Shelvapille
Subject: Re: [PULL REQUEST] Please pull rdma.git

There are 2 vendors that I'm aware of that use linux for their embedded IB switch platforms: Mellanox and Bay Microsystems and there may be others.

Mellanox has switch drivers for the various different switch ICs that have been used over time (InfiniScale IV, SwitchX/SwitchX-2, and most Switch-IB).

-- Hal

This correspondence, and any attachments or files transmitted with this correspondence, contains information which may be confidential and privileged and is intended solely for the use of the addressee. Unless you are the addressee or are authorized to receive messages for the addressee, you may not use, copy, disseminate, or disclose this correspondence or any information contained in this correspondence to any third party. If you have received this correspondence in error, please notify the sender immediately and delete this correspondence and any attachments or files transmitted with this correspondence from your system, and destroy any and all copies thereof, electronic or otherwise. Your cooperation and understanding are greatly appreciated. --
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

Hi Suri,

how about you submit the driver to the kernel tree?
--
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

Christoph:

Is that a requirement? Because, we have a lot of proprietary things that is tied to our FPGA that is in the driver and to rip all of that to make it more consumable to the community would be another effort that we don't have resources for. Please let me know.
Thanks,
Suri

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Thursday, July 16, 2015 10:02 AM
To: Suri Shelvapille
Cc: Hal Rosenstock; Doug Ledford; Christoph Hellwig; Linus Torvalds; linux-rdma
Subject: Re: [PULL REQUEST] Please pull rdma.git

Hi Suri,

how about you submit the driver to the kernel tree?

This correspondence, and any attachments or files transmitted with this correspondence, contains information which may be confidential and privileged and is intended solely for the use of the addressee. Unless you are the addressee or are authorized to receive messages for the addressee, you may not use, copy, disseminate, or disclose this correspondence or any information contained in this correspondence to any third party. If you have received this correspondence in error, please notify the sender immediately and delete this correspondence and any attachments or files transmitted with this correspondence from your system, and destroy any and all copies thereof, electronic or otherwise. Your cooperation and understanding are greatly appreciated. --
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