[PATCH for-next 00/27] IB/hfi1, rdmavt, core, etc: patches for next 08/04/2017

[PATCH for-next 00/27] IB/hfi1, rdmavt, core, etc: patches for next 08/04/2017

[Dennis Dalessandro]

Hi Doug,

Here is the next set of patches for our drivers for 4.14. Along with the usual
fixes and such this includes our extended LID patches which were posted a
while back [1] [2]. Please drop those patch sets as they needed rebased.
Along with those core changes we are now submitting the driver changes that
are needed to make use of extended LIDs. 21 of these 27 patches are related
to that effort. The other 6 are usual fixes.

Also note the first patch we did mark stable. That may be something to consider
for the 4.13-rc.

Patches can can also be found in my GitHub repo at:
https://github.com/ddalessa/kernel/tree/for-4.14

[1] http://marc.info/?l=linux-rdma&m=149694352021933&w=2
[2] http://marc.info/?l=linux-rdma&m=149694355721946&w=2

---

Alex Estrin (1):
      IB/hfi1: Revert egress pkey check enforcement

Bartlomiej Dudek (1):
      IB/hfi1: Use host_link_state to read state when DC is shut down

Byczkowski, Jakub (1):
      IB/hfi1: Remove lstate from hfi1_pportdata

Dasaratharaman Chandramouli (10):
      IB/core: Convert ah_attr from OPA to IB when copying to user
      IB/srpt: Increase lid and sm_lid to 32 bits
      IB/IPoIB: Increase local_lid to 32 bits
      IB/mad: Change slid in RMPP recv from 16 to 32 bits
      IB/core: Change port_attr.lid size from 16 to 32 bits
      IB/core: Change port_attr.sm_lid from 16 to 32 bits
      IB/CM: Create appropriate path records when handling CM request
      IB/CM: Set appropriate slid and dlid when handling CM request
      IB/rdmavt,hfi1,qib: Enhance rdmavt and hfi1 to use 32 bit lids
      IB/hfi1: Enable RDMA_CAP_OPA_AH in hfi driver to support extended LIDs

Don Hiatt (11):
      IB/core: Change wc.slid from 16 to 32 bits
      IB/CM: Add OPA Path record support to CM
      IB/rdmavt,hfi1,qib: Modify check_ah() to account for extended LIDs
      IB/hfi1: Add support to receive 16B bypass packets
      IB/hfi1: Add support to send 16B bypass packets
      IB/hfi1: Add support to process 16B header errors
      IB/hfi1: Determine 9B/16B L2 header type based on Address handle
      IB/hfi1: Add 16B UD support
      IB/hfi1: Add 16B trace support
      IB/hfi1: Add 16B RC/UC support
      IB/hfi1: Enhance PIO/SDMA send for 16B

Michael J. Ruhl (2):
      IB/hfi1: Protect context array set/clear with spinlock
      IB/hf1: User context locking is inconsistent

Sebastian Sanchez (1):
      IB/hfi1: Remove pmtu from the QP structure


 drivers/infiniband/core/cm.c              |  159 +++++++++--
 drivers/infiniband/core/core_priv.h       |    1 
 drivers/infiniband/core/mad_rmpp.c        |    2 
 drivers/infiniband/core/sa_query.c        |   21 +
 drivers/infiniband/core/ucm.c             |    2 
 drivers/infiniband/core/ucma.c            |   10 -
 drivers/infiniband/core/user_mad.c        |    2 
 drivers/infiniband/core/uverbs_cmd.c      |   22 +
 drivers/infiniband/core/uverbs_marshall.c |   48 +++
 drivers/infiniband/hw/hfi1/aspm.h         |   35 ++
 drivers/infiniband/hw/hfi1/chip.c         |  110 +++++--
 drivers/infiniband/hw/hfi1/chip.h         |    3 
 drivers/infiniband/hw/hfi1/common.h       |   10 -
 drivers/infiniband/hw/hfi1/debugfs.c      |   32 +-
 drivers/infiniband/hw/hfi1/driver.c       |  291 +++++++++++++++-----
 drivers/infiniband/hw/hfi1/file_ops.c     |  340 +++++++++++++----------
 drivers/infiniband/hw/hfi1/hfi.h          |  422 ++++++++++++++++++++++++++---
 drivers/infiniband/hw/hfi1/init.c         |  306 ++++++++++++++-------
 drivers/infiniband/hw/hfi1/mad.c          |  108 ++++++-
 drivers/infiniband/hw/hfi1/qp.c           |   28 ++
 drivers/infiniband/hw/hfi1/rc.c           |  389 +++++++++++++++++---------
 drivers/infiniband/hw/hfi1/ruc.c          |  237 ++++++++++++----
 drivers/infiniband/hw/hfi1/trace.c        |  153 ++++++++++
 drivers/infiniband/hw/hfi1/trace_ibhdrs.h |  364 +++++++++++++++++--------
 drivers/infiniband/hw/hfi1/trace_rx.h     |   12 -
 drivers/infiniband/hw/hfi1/uc.c           |   42 ++-
 drivers/infiniband/hw/hfi1/ud.c           |  427 ++++++++++++++++++++++-------
 drivers/infiniband/hw/hfi1/user_sdma.c    |    7 
 drivers/infiniband/hw/hfi1/verbs.c        |  280 +++++++++++++------
 drivers/infiniband/hw/hfi1/verbs.h        |   45 ++-
 drivers/infiniband/hw/hfi1/vnic.h         |   15 -
 drivers/infiniband/hw/hfi1/vnic_main.c    |   32 --
 drivers/infiniband/hw/mlx4/alias_GUID.c   |    2 
 drivers/infiniband/hw/mlx4/mad.c          |    8 -
 drivers/infiniband/hw/mlx5/mad.c          |    2 
 drivers/infiniband/hw/mthca/mthca_cmd.c   |    4 
 drivers/infiniband/hw/mthca/mthca_mad.c   |    4 
 drivers/infiniband/hw/qib/qib_mad.c       |    4 
 drivers/infiniband/hw/qib/qib_verbs.c     |    9 +
 drivers/infiniband/sw/rdmavt/ah.c         |   10 -
 drivers/infiniband/sw/rdmavt/cq.c         |    2 
 drivers/infiniband/sw/rdmavt/qp.c         |   32 ++
 drivers/infiniband/ulp/ipoib/ipoib.h      |    2 
 drivers/infiniband/ulp/srpt/ib_srpt.h     |    4 
 include/rdma/ib_marshall.h                |    6 
 include/rdma/ib_verbs.h                   |   33 ++
 include/rdma/opa_addr.h                   |   42 +++
 include/rdma/opa_vnic.h                   |    3 
 include/rdma/rdma_vt.h                    |    2 
 include/rdma/rdmavt_qp.h                  |    1 
 50 files changed, 2986 insertions(+), 1139 deletions(-)

--
-Denny

[PATCH for-next 01/27] IB/hfi1: Revert egress pkey check enforcement

[Dennis Dalessandro]

From: Alex Estrin <alex.estrin@intel.com>

Current code has some serious flaws. Disarm the flag
pending an appropriate patch.

Fixes: 53526500f301 ("IB/hfi1: Permanently enable P_Key checking in HFI")
Cc: stable@vger.kernel.org
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Alex Estrin <alex.estrin@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
---
 drivers/infiniband/hw/hfi1/init.c |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)

diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c
index be027c9..da7cd5b 100644
--- a/drivers/infiniband/hw/hfi1/init.c
+++ b/drivers/infiniband/hw/hfi1/init.c
@@ -519,7 +519,6 @@ void hfi1_init_pportdata(struct pci_dev *pdev, struct hfi1_pportdata *ppd,
 
 	ppd->pkeys[default_pkey_idx] = DEFAULT_P_KEY;
 	ppd->part_enforce |= HFI1_PART_ENFORCE_IN;
-	ppd->part_enforce |= HFI1_PART_ENFORCE_OUT;
 
 	if (loopback) {
 		hfi1_early_err(&pdev->dev,

Re: [PATCH for-next 00/27] IB/hfi1, rdmavt, core, etc: patches for next 08/04/2017

[Dennis Dalessandro]

On 8/4/2017 4:52 PM, Dennis Dalessandro wrote:

> Dasaratharaman Chandramouli (10):
>        IB/core: Convert ah_attr from OPA to IB when copying to user
>        IB/srpt: Increase lid and sm_lid to 32 bits
>        IB/IPoIB: Increase local_lid to 32 bits
>        IB/mad: Change slid in RMPP recv from 16 to 32 bits
>        IB/core: Change port_attr.lid size from 16 to 32 bits
>        IB/core: Change port_attr.sm_lid from 16 to 32 bits
>        IB/CM: Create appropriate path records when handling CM request
>        IB/CM: Set appropriate slid and dlid when handling CM request
>        IB/rdmavt,hfi1,qib: Enhance rdmavt and hfi1 to use 32 bit lids
>        IB/hfi1: Enable RDMA_CAP_OPA_AH in hfi driver to support extended LIDs

Leon had some comments with these patches and Don is going to submit an 
updated version of these 10. These were submitted previously back in 
early June and sat on the list without comment. Hearing no feedback I 
rebased them and sent with my other patches so we could get the driver 
patches out.

In hindsight we probably should have just submitted another revision to 
these 10 and sent the following 11 and the other 6 as a separate series. 
If that makes things easier we can still do that. Let me know the 
easiest way for you to consume these.

> Don Hiatt (11):
>        IB/core: Change wc.slid from 16 to 32 bits
>        IB/CM: Add OPA Path record support to CM
>        IB/rdmavt,hfi1,qib: Modify check_ah() to account for extended LIDs
>        IB/hfi1: Add support to receive 16B bypass packets
>        IB/hfi1: Add support to send 16B bypass packets
>        IB/hfi1: Add support to process 16B header errors
>        IB/hfi1: Determine 9B/16B L2 header type based on Address handle
>        IB/hfi1: Add 16B UD support
>        IB/hfi1: Add 16B trace support
>        IB/hfi1: Add 16B RC/UC support
>        IB/hfi1: Enhance PIO/SDMA send for 16B

These will still apply, as will the other 6 patches in this series.

-Denny

Re: [PATCH for-next 00/27] IB/hfi1, rdmavt, core, etc: patches for next 08/04/2017

[Don Hiatt]

On 8/10/2017 10:05 AM, Dennis Dalessandro wrote:
> On 8/4/2017 4:52 PM, Dennis Dalessandro wrote:
>
>> Dasaratharaman Chandramouli (10):
>>        IB/core: Convert ah_attr from OPA to IB when copying to user
>>        IB/srpt: Increase lid and sm_lid to 32 bits
>>        IB/IPoIB: Increase local_lid to 32 bits
>>        IB/mad: Change slid in RMPP recv from 16 to 32 bits
>>        IB/core: Change port_attr.lid size from 16 to 32 bits
>>        IB/core: Change port_attr.sm_lid from 16 to 32 bits
>>        IB/CM: Create appropriate path records when handling CM request
>>        IB/CM: Set appropriate slid and dlid when handling CM request
>>        IB/rdmavt,hfi1,qib: Enhance rdmavt and hfi1 to use 32 bit lids
>>        IB/hfi1: Enable RDMA_CAP_OPA_AH in hfi driver to support 
>> extended LIDs
>

I submitted a v4 of just the 'extended lid' changes which is 8 patches 
in total.
These patches from Denny's original patch series (from Dasa) are still 
valid:

        IB/CM: Set appropriate slid and dlid when handling CM request
        IB/rdmavt,hfi1,qib: Enhance rdmavt and hfi1 to use 32 bit lids
        IB/hfi1: Enable RDMA_CAP_OPA_AH in hfi driver to support 
extended LIDs


> Leon had some comments with these patches and Don is going to submit 
> an updated version of these 10. These were submitted previously back 
> in early June and sat on the list without comment. Hearing no feedback 
> I rebased them and sent with my other patches so we could get the 
> driver patches out.
>
> In hindsight we probably should have just submitted another revision 
> to these 10 and sent the following 11 and the other 6 as a separate 
> series. If that makes things easier we can still do that. Let me know 
> the easiest way for you to consume these.
>
>> Don Hiatt (11):
>>        IB/core: Change wc.slid from 16 to 32 bits
>>        IB/CM: Add OPA Path record support to CM
>>        IB/rdmavt,hfi1,qib: Modify check_ah() to account for extended 
>> LIDs
>>        IB/hfi1: Add support to receive 16B bypass packets
>>        IB/hfi1: Add support to send 16B bypass packets
>>        IB/hfi1: Add support to process 16B header errors
>>        IB/hfi1: Determine 9B/16B L2 header type based on Address handle
>>        IB/hfi1: Add 16B UD support
>>        IB/hfi1: Add 16B trace support
>>        IB/hfi1: Add 16B RC/UC support
>>        IB/hfi1: Enhance PIO/SDMA send for 16B
>
> These will still apply, as will the other 6 patches in this series.
>
> -Denny

Re: [PATCH for-next 00/27] IB/hfi1, rdmavt, core, etc: patches for next 08/04/2017

[Doug Ledford]

On Thu, 2017-08-10 at 11:17 -0700, Don Hiatt wrote:
> 
> On 8/10/2017 10:05 AM, Dennis Dalessandro wrote:
> > On 8/4/2017 4:52 PM, Dennis Dalessandro wrote:
> > 
> > > Dasaratharaman Chandramouli (10):
> > >        IB/core: Convert ah_attr from OPA to IB when copying to
> > > user
> > >        IB/srpt: Increase lid and sm_lid to 32 bits
> > >        IB/IPoIB: Increase local_lid to 32 bits
> > >        IB/mad: Change slid in RMPP recv from 16 to 32 bits
> > >        IB/core: Change port_attr.lid size from 16 to 32 bits
> > >        IB/core: Change port_attr.sm_lid from 16 to 32 bits
> > >        IB/CM: Create appropriate path records when handling CM
> > > request
> > >        IB/CM: Set appropriate slid and dlid when handling CM
> > > request
> > >        IB/rdmavt,hfi1,qib: Enhance rdmavt and hfi1 to use 32 bit
> > > lids
> > >        IB/hfi1: Enable RDMA_CAP_OPA_AH in hfi driver to support 
> > > extended LIDs
> 
> I submitted a v4 of just the 'extended lid' changes which is 8
> patches 
> in total.
> These patches from Denny's original patch series (from Dasa) are
> still 
> valid:
> 
>         IB/CM: Set appropriate slid and dlid when handling CM request
>         IB/rdmavt,hfi1,qib: Enhance rdmavt and hfi1 to use 32 bit
> lids
>         IB/hfi1: Enable RDMA_CAP_OPA_AH in hfi driver to support 
> extended LIDs

OK, I *think* I've got things figured out, but I need to request that
you guys submit things differently in the future.

In general, if you submit a series of 7 patches that does some "thing",
then whether or not a resubmit adds or reduces the patch count, it
should still be just that "thing".  And once you do a submission for a
"thing" as a series, that series should never get subsumed into another
"thing".  In order for me to keep v1 versus v2 of any given "thing"
straight and not miss stuff, or waste a ton of time trying to apply
things that are already applied, a "thing" must remain the same "thing"
unless it is clearly withdrawn entirely and submitted as part of a new
"thing".

That said, I have the two series that updated the lid size and added
the support to the CM.  I think that means that patches 7 through 16
are now all applied, but I still need to process the others (and
digging that "thing" out of the middle is even hard than if it were
just at the beginning or the end).  Let me know if that is not correct.

> 
> > Leon had some comments with these patches and Don is going to
> > submit 
> > an updated version of these 10. These were submitted previously
> > back 
> > in early June and sat on the list without comment. Hearing no
> > feedback 
> > I rebased them and sent with my other patches so we could get the 
> > driver patches out.
> > 
> > In hindsight we probably should have just submitted another
> > revision 
> > to these 10 and sent the following 11 and the other 6 as a
> > separate 
> > series. If that makes things easier we can still do that. Let me
> > know 
> > the easiest way for you to consume these.
> > 
> > > Don Hiatt (11):
> > >        IB/core: Change wc.slid from 16 to 32 bits
> > >        IB/CM: Add OPA Path record support to CM
> > >        IB/rdmavt,hfi1,qib: Modify check_ah() to account for
> > > extended 
> > > LIDs
> > >        IB/hfi1: Add support to receive 16B bypass packets
> > >        IB/hfi1: Add support to send 16B bypass packets
> > >        IB/hfi1: Add support to process 16B header errors
> > >        IB/hfi1: Determine 9B/16B L2 header type based on Address
> > > handle
> > >        IB/hfi1: Add 16B UD support
> > >        IB/hfi1: Add 16B trace support
> > >        IB/hfi1: Add 16B RC/UC support
> > >        IB/hfi1: Enhance PIO/SDMA send for 16B
> > 
> > These will still apply, as will the other 6 patches in this series.
> > 
> > -Denny
> 
> 
-- 
Doug Ledford <dledford@redhat.com>
    GPG KeyID: B826A3330E572FDD
    Key fingerprint = AE6B 1BDA 122B 23B4 265B  1274 B826 A333 0E57 2FDD

Re: [PATCH for-next 00/27] IB/hfi1, rdmavt, core, etc: patches for next 08/04/2017

[Doug Ledford]

On Fri, 2017-08-18 at 15:07 -0400, Doug Ledford wrote:
> That said, I have the two series that updated the lid size and added
> the support to the CM.  I think that means that patches 7 through 16
> are now all applied, but I still need to process the others (and
> digging that "thing" out of the middle is even hard than if it were
> just at the beginning or the end).  Let me know if that is not
> correct.

Patches 1-6 and 17-27 processed and applied.  Thanks.

-- 
Doug Ledford <dledford@redhat.com>
    GPG KeyID: B826A3330E572FDD
    Key fingerprint = AE6B 1BDA 122B 23B4 265B  1274 B826 A333 0E57 2FDD