OSDN Git Service

tomoyo/tomoyo-test1.git
5 years agoIB/hfi1: Allow for extra entries in QP's s_ack_queue
Kaike Wan [Thu, 24 Jan 2019 05:48:38 +0000 (21:48 -0800)]
IB/hfi1: Allow for extra entries in QP's s_ack_queue

The TID RDMA WRITE protocol differs from normal IB RDMA WRITE
in that TID RDMA WRITE requests do require responses, not just
ACKs.

Therefore, TID RDMA WRITE requests need to be treated as RDMA
READ requests from the point of view of the QPs' s_ack_queue.
In other words, the QPs' need to allow for TID RDMA WRITE
requests to be stored in their s_ack_queue.

However, because the user does not know anything about the TID
RDMA capability and/or protocols, these extra entries in the
queue cannot be advertized to the user.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Build TID RDMA WRITE request
Kaike Wan [Thu, 24 Jan 2019 05:48:28 +0000 (21:48 -0800)]
IB/hfi1: Build TID RDMA WRITE request

This patch adds the functions to build TID RDMA WRITE request.
The work request opcode, packet opcode, and packet formats for TID
RDMA WRITE protocol are also defined in this patch.

Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoMerge branch 'tid-read' into hfi1-tid
Doug Ledford [Tue, 5 Feb 2019 22:59:43 +0000 (17:59 -0500)]
Merge branch 'tid-read' into hfi1-tid

This is the series for adding TID RDMA read. Kaike put in a lot of
effort into making this more consumable for review so special thanks to
him.

Allocating resources and tracing are separated out followed by patches
which build up the read request. Then we have the patches to receive
incoming TID RDMA read requests and handle integration with the RC
protocol.

See the cover letter of the original posting for more of a detailed
overview of TID.

https://www.spinics.net/lists/linux-rdma/msg66611.html

* tid-read:
  IB/hfi1: Add static trace for TID RDMA READ protocol
  IB/hfi1: Enable TID RDMA READ protocol
  IB/hfi1: Add interlock between a TID RDMA request and other requests
  IB/hfi1: Integrate TID RDMA READ protocol into RC protocol
  IB/hfi1: Increment the retry timeout value for TID RDMA READ request
  IB/hfi1: Add functions for restarting TID RDMA READ request
  IB/hfi1: Add TID RDMA handlers
  IB/hfi1: Add functions to receive TID RDMA READ response
  IB/hfi1: Add a function to build TID RDMA READ response
  IB/hfi1: Add functions to receive TID RDMA READ request
  IB/hfi1: Set PbcInsertHcrc for TID RDMA packets
  IB/hfi1: Add functions to build TID RDMA READ request
  IB/hfi1: Add static trace for flow and TID management functions
  IB/hfi1: Add the counter n_tidwait
  IB/hfi1: TID RDMA RcvArray programming and TID allocation
  IB/hfi1: TID RDMA flow allocation
  IB/hfi: Move RC functions into a header file

Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add static trace for TID RDMA READ protocol
Kaike Wan [Thu, 24 Jan 2019 03:32:44 +0000 (19:32 -0800)]
IB/hfi1: Add static trace for TID RDMA READ protocol

This patch makes the following changes to the static trace:
1. Adds the decoding of TID RDMA READ packets in IB header trace;
2. Tracks qpriv->s_flags and iow_flags in qpsleepwakeup trace;
3. Adds a new event to track RC ACK receiving;
4. Adds trace events for various stages of the TID RDMA READ
protocol. These events provide a fine-grained control for monitoring
and debugging the hfi1 driver in the filed.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Enable TID RDMA READ protocol
Kaike Wan [Thu, 24 Jan 2019 03:32:30 +0000 (19:32 -0800)]
IB/hfi1: Enable TID RDMA READ protocol

This patch enables TID RDMA READ protocol by converting a qualified
RDMA READ request into a TID RDMA READ request internally:
(1) The TID RDMA capability must be enabled;
(2) The request must start on a 4K page boundary and all receiving
 buffers must start on 4K page boundaries;
(3) The request length must be a multiple of 4K and must be larger or
equal to 256K. Each receiving buffer length must be a multiple of 4K.

Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add interlock between a TID RDMA request and other requests
Kaike Wan [Thu, 24 Jan 2019 14:36:48 +0000 (06:36 -0800)]
IB/hfi1: Add interlock between a TID RDMA request and other requests

This locking mechanism is designed to provent vavious memory corruption
scenarios from occurring when requests are pipelined, especially when
RDMA READ/WRITE requests are interleaved with TID RDMA READ/WRITE
requests:
1. READ-AFTER-READ;
2. READ-AFTER-WRITE;
3. WRITE-AFTER-READ;
When memory corruption is likely, a request will be held back until
previous requests have been completed.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Integrate TID RDMA READ protocol into RC protocol
Kaike Wan [Thu, 24 Jan 2019 03:32:09 +0000 (19:32 -0800)]
IB/hfi1: Integrate TID RDMA READ protocol into RC protocol

This patch integrates the TID RDMA READ protocol into the IB RC protocol.
This protocol is an end-to-end protocol between the hfi1 drivers on two
OPA nodes that converts a qualified RDMA READ request into a TID RDMA
READ request to avoid data copying on the requester side. The following
codes are added in this patch:
- Send the TID RDMA READ request;
- Complete the TID RDMA READ send request;
- Send the TID RDMA READ response;
- Complete the TID RDMA READ request;

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Increment the retry timeout value for TID RDMA READ request
Kaike Wan [Thu, 24 Jan 2019 03:31:57 +0000 (19:31 -0800)]
IB/hfi1: Increment the retry timeout value for TID RDMA READ request

The RC retry timeout value is based on the estimated time for the
response packet to come back. However, for TID RDMA READ request, due
to the use of header suppression, the driver is normally not notified
for each incoming response packet until the last TID RDMA READ response
packet. Consequently, the retry timeout value should be extended to
cover the transaction time for the entire length of a segment (default
256K) instead of that for a single packet. This patch addresses the
issue by introducing new retry timer functions to account for multiple
packets and wrapper functions for backward compatibility.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add functions for restarting TID RDMA READ request
Kaike Wan [Thu, 24 Jan 2019 03:31:46 +0000 (19:31 -0800)]
IB/hfi1: Add functions for restarting TID RDMA READ request

This patch adds functions to retry TID RDMA READ request. Since TID RDMA
READ request could be retried from any segment boundary, it requires
a number of tracking fields in various structures and those fields
should be reset properly. The qp->s_num_rd_atomic field is reset before
retry and therefore should be incremented for each new or retried
RDMA READ or atomic request.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add TID RDMA handlers
Kaike Wan [Thu, 24 Jan 2019 14:36:34 +0000 (06:36 -0800)]
IB/hfi1: Add TID RDMA handlers

This commit adds the TID RDMA READ pointers to the receiving opcode
handlers. It also adds TID RDMA READ header sizes to header size table.
A function to print the RHF EFLAGS errors is created so that it can be
shared by both IB and TID RDMA receiving functions.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add functions to receive TID RDMA READ response
Kaike Wan [Tue, 5 Feb 2019 22:13:30 +0000 (14:13 -0800)]
IB/hfi1: Add functions to receive TID RDMA READ response

This patch adds the functions to receive TID RDMA READ response. The TID
resource information in the KDETH packet header will direct the hardware
to deliver the packet payload to the user buffer automatically and the
software will handle the packet header for the last packet of a segment
as all other packet headers are suppressed by default. The TID entries
will be freed when all packets for a segment have been received. This
patch also adds the functions to handle KDETH eflag errors, including
flow sequence and generation errors, when a TID RDMA READ response
packet is received . The flow sequence error can be recovered by software
checking of the flow sequence and will disappear when the hardware flow
is programmed with a new generation number.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add a function to build TID RDMA READ response
Kaike Wan [Thu, 24 Jan 2019 03:31:12 +0000 (19:31 -0800)]
IB/hfi1: Add a function to build TID RDMA READ response

This patch adds the function to build TID RDMA READ response packet.
The previously received TID resource information will be used to
build the KDETH packet, which will direct the delivery of packet payload
by hardware.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add functions to receive TID RDMA READ request
Kaike Wan [Thu, 24 Jan 2019 03:31:02 +0000 (19:31 -0800)]
IB/hfi1: Add functions to receive TID RDMA READ request

This patch adds the functions to receive TID RDMA READ request. The TID
resource information will be stored and tracked. Duplicate request
will also be handled properly.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Set PbcInsertHcrc for TID RDMA packets
Kaike Wan [Thu, 24 Jan 2019 03:30:51 +0000 (19:30 -0800)]
IB/hfi1: Set PbcInsertHcrc for TID RDMA packets

All TID RDMA packets are in KDETH packet format and therefore the
PbcInsertHcrc must be set properly before sending the packet to
hardware. Otherwise, the packets will be dropped by the receiver.
By default, HCRC is not inserted for 9B packets without KDETH, and
this patch adds that back for TID RDMA packets.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add functions to build TID RDMA READ request
Kaike Wan [Thu, 24 Jan 2019 03:30:40 +0000 (19:30 -0800)]
IB/hfi1: Add functions to build TID RDMA READ request

This patch adds the helper functions to build the TID RDMA READ request
on the requester side. The key is to allocate TID resources (TID flow
and TID entries) and send the resource information to the responder side
along with the read request. Since the TID resources are limited, each
TID RDMA READ request has to be split into segments with a default
segment size of 256K. A software flow is allocated to track the data
transaction for each segment. The work request opcode, packet opcode, and
packet formats for TID RDMA READ protocol are also defined in this patch.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add static trace for flow and TID management functions
Kaike Wan [Thu, 24 Jan 2019 03:30:28 +0000 (19:30 -0800)]
IB/hfi1: Add static trace for flow and TID management functions

This patch adds the static trace for the flow and TID management
functions to help debugging in the filed.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add the counter n_tidwait
Kaike Wan [Thu, 24 Jan 2019 03:30:18 +0000 (19:30 -0800)]
IB/hfi1: Add the counter n_tidwait

This patch adds the counter n_tidwait to count the number of times the
TID resource allocator has to wait for TID resources.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: TID RDMA RcvArray programming and TID allocation
Kaike Wan [Thu, 24 Jan 2019 03:30:07 +0000 (19:30 -0800)]
IB/hfi1: TID RDMA RcvArray programming and TID allocation

TID entries are used by hfi1 hardware to receive data payload from
incoming packets directly into a user buffer and thus avoid data copying
by software. This patch implements the functions for TID allocation,
freeing, and programming TID RcvArray entries in hardware for kernel
clients. TID entries are managed via lists of TID groups similar to PSM.
Furthermore, to track TID resource allocation for each request, software
flows are also allocated and freed as needed. Since software flows
consume large amount of memory for tracking TID allocation and freeing,
it is generally desirable to allocate them dynamically in the send queue
and only for TID RDMA requests, but pre-allocate them for receive queue
because the send queue could have thousands of entries while the receive
queue has only a limited number of entries.

Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: TID RDMA flow allocation
Kaike Wan [Tue, 5 Feb 2019 22:13:13 +0000 (14:13 -0800)]
IB/hfi1: TID RDMA flow allocation

The hfi1 hardware flow is a hardware flow-control mechanism for a KDETH
data packet that is received on a hfi1 port. It validates the packet by
checking both the generation and sequence. Each QP that uses the TID RDMA
mechanism will allocate a hardware flow from its receiving context for
any incoming KDETH data packets.

This patch implements:
(1) a function to allocate hardware flow
(2) a function to free hardware flow
(3) a function to initialize hardware flow generation for a receiving
    context
(4) a wait mechanism if the hardware flow is not available
(4) a function to remove the qp from the wait queue for hardware flow
    when the qp is reset or destroyed.

Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi: Move RC functions into a header file
Kaike Wan [Thu, 24 Jan 2019 03:29:44 +0000 (19:29 -0800)]
IB/hfi: Move RC functions into a header file

This patch moves some RC helper functions into a header file so that
they can be called from both RC and  TID RDMA functions. In addition,
a common function for rewinding a request is created in rdmavt so that
it can be shared between qib and hfi1 driver.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoMerge branch 'opfn' into hfi1-tid
Doug Ledford [Thu, 31 Jan 2019 16:56:25 +0000 (11:56 -0500)]
Merge branch 'opfn' into hfi1-tid

This series adds the OPFN feature, which is used as the negotiation
protocol by TID RDMA. This adds a totally hidden, in-band negotiation
transfer that happens on the consumer's queue pair but without the
consumer's knowledge.  For that reason, things like completions for OPFN
transfers must be filtered out of the completion queue and not sent to
the consumer.  This feature does not impact any consumer APIs, but does
impact the driver/driver wire API.

At a high level OPFN enables exchanging parameters between two hosts
using IB compare and swap requests to a special virtual address. The
request uses a reserved IB work request opcode (see patch 3).

* opfn:
  IB/hfi1: Add static trace for OPFN
  IB/hfi1: Integrate OPFN into RC transactions
  IB/hfi1, IB/rdmavt: Allow for extending of QP's s_ack_queue
  IB/hfi1: OPFN interface
  IB/hfi1: Add OPFN helper functions for TID RDMA feature
  IB/hfi1: OPFN support discovery

Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add static trace for OPFN
Kaike Wan [Thu, 24 Jan 2019 14:10:09 +0000 (06:10 -0800)]
IB/hfi1: Add static trace for OPFN

This patch adds the static trace to the OPFN code and moves tid related
static trace code into a new header file.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Integrate OPFN into RC transactions
Kaike Wan [Thu, 24 Jan 2019 03:21:11 +0000 (19:21 -0800)]
IB/hfi1: Integrate OPFN into RC transactions

OPFN parameter negotiation allows a pair of connected RC QPs to exchange
a set of parameters in succession. This negotiation does not commence
till the first ULP request. Because OPFN operations are operations
private to the driver, they do not generate user completions or put the
QP into error when they run out of retries. This patch integrates the
OPFN protocol into the transactions of an RC QP.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1, IB/rdmavt: Allow for extending of QP's s_ack_queue
Kaike Wan [Thu, 24 Jan 2019 03:21:01 +0000 (19:21 -0800)]
IB/hfi1, IB/rdmavt: Allow for extending of QP's s_ack_queue

The OPFN protocol uses the COMPARE_SWAP request to exchange data
between the requester and the responder and therefore needs to
be stored in the QP's s_ack_queue when the request is received
on the responder side. However, because the user does not know
anything about the OPFN protocol, this extra entry in the
queue cannot be advertised to the user. This patch adds an extra
entry in a QP's s_ack_queue.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: OPFN interface
Kaike Wan [Thu, 24 Jan 2019 03:20:52 +0000 (19:20 -0800)]
IB/hfi1: OPFN interface

OPFN allows a pair of connected RC QPs to exchange a set of parameters
in succession. The parameter exchange itself is done using the IB compare
and swap request with a special virtual address. The request is triggered
using a reserved IB work request opcode. This patch implements the OPFN
interface to initialize, start, process, and reset the OPFN request.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: Add OPFN helper functions for TID RDMA feature
Kaike Wan [Thu, 24 Jan 2019 03:20:42 +0000 (19:20 -0800)]
IB/hfi1: Add OPFN helper functions for TID RDMA feature

This patch adds the OPFN helper functions to initialize, encode, decode,
and reset OPFN parameters for the TID RDMA feature.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/hfi1: OPFN support discovery
Mitko Haralanov [Thu, 24 Jan 2019 14:09:46 +0000 (06:09 -0800)]
IB/hfi1: OPFN support discovery

OPFN (Omni Path Feature Negotiation) support discovery allows a RC QP to
announce that it supports OPFN and also discover if OPFN is supported by
the peer QP. OPFN parameter negotiation is skipped unless OPFN support is
first discovered. OPFN support is announced by claiming what was
the reserved bit in dword 1 of OmniPath modified base transport header
in requests and responses.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Mitko Haralanov <mitko.haralanov@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/{hfi1, qib, rvt} Cleanup open coded sge usage
Michael J. Ruhl [Thu, 24 Jan 2019 03:08:29 +0000 (19:08 -0800)]
IB/{hfi1, qib, rvt} Cleanup open coded sge usage

Several locations for manipulating sges use an open coded sequence
that is covered by helper functions.

Use the appropriate helper functions.

Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/{hfi1,qib}: Cleanup open coded sge sizing
Michael J. Ruhl [Thu, 24 Jan 2019 03:08:19 +0000 (19:08 -0800)]
IB/{hfi1,qib}: Cleanup open coded sge sizing

Sge sizing is done in several places using an open coded method.

This can cause maintenance issues.  The open coded method is
encapsulated in a helper routine.  The helper was introduced with
commit:

1198fcea8a78 ("IB/hfi1, rdmavt: Move SGE state helper routines into
rdmavt")

Update all call sites that have the open coded path with the helper
routine.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoIB/ipoib: Make ipoib_intercept_dev_id_attr() static
Kamal Heib [Tue, 29 Jan 2019 11:55:48 +0000 (13:55 +0200)]
IB/ipoib: Make ipoib_intercept_dev_id_attr() static

The function ipoib_intercept_dev_id_attr() is only used in ipoib_main.c

Fixes: f6350da41dc7 ("IB/ipoib: Log sysfs 'dev_id' accesses from userspace")
Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/vmw_pvrdma: Support upto 64-bit PFNs
Adit Ranadive [Sat, 26 Jan 2019 05:09:36 +0000 (05:09 +0000)]
RDMA/vmw_pvrdma: Support upto 64-bit PFNs

Update the driver to use the new device capability to report 64-bit UAR
PFNs.

Reviewed-by: Jorgen Hansen <jhansen@vmware.com>
Signed-off-by: Adit Ranadive <aditr@vmware.com>
Reviewed-by: Vishnu Dasa <vdasa@vmware.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoMerge branch 'devx-async' into k.o/for-next
Jason Gunthorpe [Tue, 29 Jan 2019 20:49:31 +0000 (13:49 -0700)]
Merge branch 'devx-async' into k.o/for-next

Yishai Hadas says:

Enable DEVX asynchronous query commands

This series enables querying a DEVX object in an asynchronous mode.

The userspace application won't block when calling the firmware and it will be
able to get the response back once that it will be ready.

To enable the above functionality:

- DEVX asynchronous command completion FD object was introduced.
- The applicable file operations were implemented to enable using it by
  the user application.
- Query asynchronous method was added to the DEVX object, it will call the
  firmware asynchronously and manages the response on the given input FD.
- Hot unplug support was added for the FD to work properly upon
  unbind/disassociate.
- mlx5 core fence for asynchronous commands was implemented and used to
  prevent racing upon unbind/disassociate.

This branch is based on mlx5-next & v5.0-rc2 due to dependencies, from
git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux

* branch 'devx-async':
  IB/mlx5: Implement DEVX hot unplug for async command FD
  IB/mlx5: Implement the file ops of DEVX async command FD
  IB/mlx5: Introduce async DEVX obj query API
  IB/mlx5: Introduce MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Implement DEVX hot unplug for async command FD
Yishai Hadas [Tue, 22 Jan 2019 06:29:59 +0000 (08:29 +0200)]
IB/mlx5: Implement DEVX hot unplug for async command FD

Implement DEVX hot unplug for the async command FD.

This is done by managing a list of the inflight commands and wait until
all launched work is completed as part of
devx_hot_unplug_async_cmd_event_file.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Implement the file ops of DEVX async command FD
Yishai Hadas [Tue, 22 Jan 2019 06:29:58 +0000 (08:29 +0200)]
IB/mlx5: Implement the file ops of DEVX async command FD

Implement the file ops of the DEVX async command FD, this enables using
the FD for reading the events and manage other options on the FD.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Introduce async DEVX obj query API
Yishai Hadas [Tue, 22 Jan 2019 06:29:57 +0000 (08:29 +0200)]
IB/mlx5: Introduce async DEVX obj query API

Introduce async DEVX obj query API to get the command response back to
user space once it's ready without blocking when calling the firmware.

The event's data includes a header with some meta data then the firmware
output command data.

The header includes:
- The input 'wr_id' to let application recognizing the response.

The input FD attribute is used to have the event data ready on.
Downstream patches from this series will implement the file ops to let
application read it.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Introduce MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD
Yishai Hadas [Tue, 22 Jan 2019 06:29:56 +0000 (08:29 +0200)]
IB/mlx5: Introduce MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD

Introduce MLX5_IB_OBJECT_DEVX_ASYNC_CMD_FD and its initial implementation.

This object is from type class FD and will be used to read DEVX async
commands completion.

The core layer should allow the driver to set object from type FD in a
safe mode, this option was added with a matching comment in place.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: prefix header search paths with $(srctree)/
Masahiro Yamada [Fri, 25 Jan 2019 04:39:31 +0000 (13:39 +0900)]
infiniband: prefix header search paths with $(srctree)/

Currently, the Kbuild core manipulates header search paths in a crazy
way [1].

To fix this mess, I want all Makefiles to add explicit $(srctree)/ to
the search paths in the srctree. Some Makefiles are already written in
that way, but not all. The goal of this work is to make the notation
consistent, and finally get rid of the gross hacks.

Having whitespaces after -I does not matter since commit 48f6e3cf5bc6
("kbuild: do not drop -I without parameter").

[1]: https://patchwork.kernel.org/patch/9632347/

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Parvi Kaustubhi <pkaustub@cisco.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: remove unneeded header search paths
Masahiro Yamada [Fri, 25 Jan 2019 04:39:30 +0000 (13:39 +0900)]
infiniband: remove unneeded header search paths

The included headers are located in include/target/. I was able to
build these drivers without the extra header search paths.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/core: Destroy QP if XRC QP fails
Yuval Avnery [Tue, 22 Jan 2019 07:02:05 +0000 (09:02 +0200)]
IB/core: Destroy QP if XRC QP fails

The open-coded variant missed destroy of SELinux created QP, reuse already
existing ib_detroy_qp() call and use this opportunity to clean
ib_create_qp() from double prints and unclear exit paths.

Reported-by: Parav Pandit <parav@mellanox.com>
Fixes: d291f1a65232 ("IB/core: Enforce PKey security on QPs")
Signed-off-by: Yuval Avnery <yuvalav@mellanox.com>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/qib: Use GUP longterm for PSM page pining
Ira Weiny [Wed, 23 Jan 2019 19:02:24 +0000 (14:02 -0500)]
RDMA/qib: Use GUP longterm for PSM page pining

Similar to the core change commit 5f1d43de5416 ("IB/core: disable memory
registration of filesystem-dax vmas")

PSM should be prevented from using filesystem DAX pages.

Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/hns: Add timer allocation support for hip08
Yangyang Li [Tue, 18 Dec 2018 13:21:55 +0000 (21:21 +0800)]
RDMA/hns: Add timer allocation support for hip08

This patch adds qpc timer and cqc timer allocation support for hardware
timeout retransmission in kernel space driver.

Signed-off-by: Yangyang Li <liyangyang20@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/hns: Add SCC context clr support for hip08
Yangyang Li [Tue, 18 Dec 2018 13:21:54 +0000 (21:21 +0800)]
RDMA/hns: Add SCC context clr support for hip08

This patch adds SCC context clear support for DCQCN in kernel space
driver.

Signed-off-by: Yangyang Li <liyangyang20@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/hns: Add SCC context allocation support for hip08
Yangyang Li [Tue, 18 Dec 2018 13:21:53 +0000 (21:21 +0800)]
RDMA/hns: Add SCC context allocation support for hip08

This patch adds SCC context allocation and initialization support for
DCQCN in kernel space driver.

Signed-off-by: Yangyang Li <liyangyang20@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Remove dead code
Moni Shoua [Tue, 22 Jan 2019 07:16:09 +0000 (09:16 +0200)]
IB/mlx5: Remove dead code

When CONFIG_INFINIBAND_ON_DEMAND_PAGING is not set there is no caller to
ib_alloc_odp_umem() so let's remove it.

Signed-off-by: Moni Shoua <monis@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Ranges in implicit ODP MR inherit its write access
Moni Shoua [Tue, 22 Jan 2019 07:16:08 +0000 (09:16 +0200)]
IB/mlx5: Ranges in implicit ODP MR inherit its write access

A sub-range in ODP implicit MR should take its write permission from the
MR and not be set always to allow.

Fixes: d07d1d70ce1a ("IB/umem: Update on demand page (ODP) support")
Signed-off-by: Moni Shoua <monis@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/iw_cxgb4: Drop __GFP_NOFAIL
Jason Gunthorpe [Sun, 20 Jan 2019 01:27:13 +0000 (02:27 +0100)]
RDMA/iw_cxgb4: Drop __GFP_NOFAIL

There is no reason for this __GFP_NOFAIL, none of the other routines in
this file use it, and there is an error unwind here. NOFAIL should be
reserved for special cases, not used by network drivers.

Fixes: 6a0b6174d35a ("rdma/cxgb4: Add support for kernel mode SRQ's")
Reported-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Declare local functions 'static'
Bart Van Assche [Tue, 22 Jan 2019 18:24:39 +0000 (10:24 -0800)]
IB/mlx5: Declare local functions 'static'

This patch avoids that sparse complains about missing function
declarations.

Fixes: c9990ab39b6e ("RDMA/umem: Move all the ODP related stuff out of ucontext and into per_mm")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/core: Declare local functions 'static'
Bart Van Assche [Tue, 22 Jan 2019 18:24:23 +0000 (10:24 -0800)]
IB/core: Declare local functions 'static'

This patch avoids that sparse complains about missing function
declarations.

Fixes: f27a0d50a4bc ("RDMA/umem: Use umem->owning_mm inside ODP")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: ipoib: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:18:00 +0000 (16:18 +0100)]
infiniband: ipoib: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: usnic: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:17:59 +0000 (16:17 +0100)]
infiniband: usnic: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Parvi Kaustubhi <pkaustub@cisco.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: ocrdma: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:17:58 +0000 (16:17 +0100)]
infiniband: ocrdma: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: mlx5: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:17:57 +0000 (16:17 +0100)]
infiniband: mlx5: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: qib: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:17:56 +0000 (16:17 +0100)]
infiniband: qib: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: hfi1: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:17:55 +0000 (16:17 +0100)]
infiniband: hfi1: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: hfi1: drop crazy DEBUGFS_SEQ_FILE_CREATE() macro
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:17:54 +0000 (16:17 +0100)]
infiniband: hfi1: drop crazy DEBUGFS_SEQ_FILE_CREATE() macro

The macro was just making things harder to follow, and audit, so remove
it and call debugfs_create_file() directly.  Also, the macro did not
need to warn about the call failing as no one should ever care about any
debugfs functions failing.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoinfiniband: cxgb4: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Tue, 22 Jan 2019 15:17:53 +0000 (16:17 +0100)]
infiniband: cxgb4: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/umad: Do not check status of nonseekable_open()
Parav Pandit [Tue, 22 Jan 2019 06:31:20 +0000 (08:31 +0200)]
IB/umad: Do not check status of nonseekable_open()

As the comment block of nonseekable_open() describes, nonseekable_open()
can never fail. Several places in kernel depend on this behavior.
Therefore, simplify the umad module to depend on this basic kernel
functionality.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agonet/mlx5: Add pci AtomicOps request
Michael Guralnik [Sat, 19 Jan 2019 00:33:11 +0000 (16:33 -0800)]
net/mlx5: Add pci AtomicOps request

Calling pci_enable_atomic_ops_to_root enables AtomicOp requests to pci
root port.

AtomicOp requests will be enabled only if the completer and all
intermediate pci bridges support PCI atomic operations.
This, together with appropriate settings in the NVCONFIG should enable
PCI atomic operations on the device.

PCI atomic operations were first introduced in PCI Express Base Specification
2.1. The Supported operations are Swap (Unconditional Swap), CAS (Compare and
Swap) and FetchAdd (Fetch and Add).

Unlike other atomic operation modes PCI atomic operations gives the user
the option to do atomic operations on local memory, without involving verbs
api, without it compromising the operation's atomicity.

Signed-off-by: Michael Guralnik <michaelgur@mellanox.com>
Reviewed-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
5 years agonet/mlx5: Make mlx5_cmd_exec_cb() a safe API
Jason Gunthorpe [Sat, 19 Jan 2019 00:33:10 +0000 (16:33 -0800)]
net/mlx5: Make mlx5_cmd_exec_cb() a safe API

APIs that have deferred callbacks should have some kind of cleanup
function that callers can use to fence the callbacks. Otherwise things
like module unloading can lead to dangling function pointers, or worse.

The IB MR code is the only place that calls this function and had a
really poor attempt at creating this fence. Provide a good version in
the core code as future patches will add more places that need this
fence.

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
5 years agoIB/umad: Avoid additional device reference during open()/close()
Parav Pandit [Tue, 22 Jan 2019 06:33:00 +0000 (08:33 +0200)]
IB/umad: Avoid additional device reference during open()/close()

ib_umad_init_port_dev() holds the reference of a ib_umad_device instance.
ib_umad_device contains standard core device and cdev.  cdev holds the
reference of its parent core device.  file ops holds the reference to cdev
using core kernel.

Therefore, there is no need to hold additional reference while opening
umad related char devices.

While at it, add comments to bring clarity on releasing references to
ib_umd_device.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Don't override existing ip_protocol
Maor Gottlieb [Thu, 17 Jan 2019 18:08:15 +0000 (20:08 +0200)]
IB/mlx5: Don't override existing ip_protocol

Two flow specifications can set the ip protocol field in
the flow table entry:

1) IB_FLOW_SPEC_TCP/UDP/GRE - set the ip protocol accordingly.
2) IB_FLOW_SPEC_IPV4/6 - has ip_protocol field for users
who want to receive specific L4 packets.

We need to avoid overriding of the ip_protocol with zeros,
in case that the user first put the L4 specification and
only then the L3.

Fixes: ca0d47538528b ('IB/mlx5: Add support in TOS and protocol to flow steering')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Add support for ODP for DEVX indirection mkey
Yishai Hadas [Sun, 13 Jan 2019 14:01:18 +0000 (16:01 +0200)]
IB/mlx5: Add support for ODP for DEVX indirection mkey

Add support for ODP for DEVX indirection mkey, it includes:
- Recognizing its type as part of the radix tree lookup.
- Use similar flow as done for the MW MKEY type.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: Manage indirection mkey upon DEVX flow for ODP
Yishai Hadas [Sun, 13 Jan 2019 14:01:17 +0000 (16:01 +0200)]
IB/mlx5: Manage indirection mkey upon DEVX flow for ODP

Manage indirection mkey upon DEVX flow to support ODP.

To support a page fault event on the indirection mkey it needs to be part
of the device mkey radix tree.

Both the creation and the deletion flows for a DEVX object which is
indirection mkey were adapted to handle that.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/mlx5: DEVX handling for indirection MKEY
Yishai Hadas [Sun, 13 Jan 2019 14:01:16 +0000 (16:01 +0200)]
IB/mlx5: DEVX handling for indirection MKEY

Once an indirection MKEY is created umem valid bit shouldn't be set as
this MKEY doesn't really hold a umem.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/hns: Add the process of AEQ overflow for hip08
Xiaofei Tan [Sat, 19 Jan 2019 06:23:29 +0000 (14:23 +0800)]
RDMA/hns: Add the process of AEQ overflow for hip08

AEQ overflow will be reported by hardware when too many asynchronous
events occurred but not be handled in time.  Normally, AEQ overflow error
is not easy to occur. Once happened, we have to do physical function reset
to recover.  PF reset is implemented in two steps. Firstly, set reset
level with ae_dev->ops->set_default_reset_request.  Secondly, run reset
with ae_dev->ops->reset_event.

Signed-off-by: Xiaofei Tan <tanxiaofei@huawei.com>
Signed-off-by: Yixian Liu <liuyixian@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/rxe: Remove unnecessary rxe variable
Zhu Yanjun [Sun, 20 Jan 2019 13:21:40 +0000 (08:21 -0500)]
IB/rxe: Remove unnecessary rxe variable

The variable rxe in the function is not used. So it is removed.

Signed-off-by: Zhu Yanjun <yanjun.zhu@oracle.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/iser: Pass the correct number of entries for dma mapped SGL
Israel Rukshin [Thu, 17 Jan 2019 15:45:45 +0000 (15:45 +0000)]
IB/iser: Pass the correct number of entries for dma mapped SGL

ib_dma_map_sg() augments the SGL into a 'dma mapped SGL'. This process may
change the number of entries and the lengths of each entry.

Code that touches dma_address is iterating over the 'dma mapped SGL' and
must use dma_nents which returned from ib_dma_map_sg().

ib_sg_to_pages() and ib_map_mr_sg() are using dma_address so they must use
dma_nents.

Fixes: 39405885005a ("IB/iser: Port to new fast registration API")
Fixes: bfe066e256d5 ("IB/iser: Reuse ib_sg_to_pages")
Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Acked-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/hw: Remove unneeded semicolons
YueHaibing [Fri, 18 Jan 2019 03:09:00 +0000 (11:09 +0800)]
IB/hw: Remove unneeded semicolons

Remove unneeded semicolons.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/rdmavt: Add wc_flags and wc_immdata to cq entry trace
Mike Marciniszyn [Thu, 17 Jan 2019 20:41:43 +0000 (12:41 -0800)]
IB/rdmavt: Add wc_flags and wc_immdata to cq entry trace

These fields were missing from the trace.  Add them.

Fixes: c6ad9482fcb8 ("IB/rdmavt: Add tracing for cq entry and poll")
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoIB/core: Simplify rdma cgroup registration
Parav Pandit [Thu, 17 Jan 2019 18:14:15 +0000 (20:14 +0200)]
IB/core: Simplify rdma cgroup registration

RDMA cgroup registration routine always returns success, so simplify
function to be void and run clang formatter over whole CONFIG_CGROUP_RDMA
art of core_priv.h.

This reduces unwinding error path for regular registration and future net
namespace change functionality for rdma device.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/qedr: remove set but not used variable 'ib_ctx'
YueHaibing [Thu, 17 Jan 2019 03:21:47 +0000 (03:21 +0000)]
RDMA/qedr: remove set but not used variable 'ib_ctx'

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/infiniband/hw/qedr/verbs.c: In function 'qedr_create_srq':
drivers/infiniband/hw/qedr/verbs.c:1436:22: warning:
 variable 'ib_ctx' set but not used [-Wunused-but-set-variable]

drivers/infiniband/hw/qedr/verbs.c: In function 'qedr_create_user_qp':
drivers/infiniband/hw/qedr/verbs.c:1701:22: warning:
 variable 'ib_ctx' set but not used [-Wunused-but-set-variable]

Fixes: b0ea0fa5435f ("IB/{core,hw}: Have ib_umem_get extract the ib_ucontext from ib_udata")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/device: Use __ib_device_get_by_name() in ib_device_rename()
Jason Gunthorpe [Sat, 12 Jan 2019 02:42:42 +0000 (02:42 +0000)]
RDMA/device: Use __ib_device_get_by_name() in ib_device_rename()

No reason to open code this loop.

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
5 years agoRDMA/hns: RDMA/hns: Assign rq head pointer when enable rq record db
Lijun Ou [Sat, 12 Jan 2019 10:36:29 +0000 (18:36 +0800)]
RDMA/hns: RDMA/hns: Assign rq head pointer when enable rq record db

When flush cqe, it needs to get the pointer of rq and sq from db address
space of user and update it into qp context by modified qp. if rq does not
exist, it will not get the value from db address space of user.

Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/mlx5: Replace kzalloc with kcalloc
Gustavo A. R. Silva [Tue, 15 Jan 2019 06:00:48 +0000 (00:00 -0600)]
RDMA/mlx5: Replace kzalloc with kcalloc

Replace kzalloc() function with its 2-factor argument form, kcalloc().

This patch replaces cases of:

kzalloc(a * b, gfp)

with:
kcalloc(a, b, gfp)

This code was detected with the help of Coccinelle.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/iw_cxgb4: Fix the unchecked ep dereference
Raju Rangoju [Fri, 11 Jan 2019 14:57:18 +0000 (20:27 +0530)]
RDMA/iw_cxgb4: Fix the unchecked ep dereference

The patch 944661dd97f4: "RDMA/iw_cxgb4: atomically lookup ep and get a
reference" from May 6, 2016, leads to the following Smatch complaint:

    drivers/infiniband/hw/cxgb4/cm.c:2953 terminate()
    error: we previously assumed 'ep' could be null (see line 2945)

Fixes: 944661dd97f4 ("RDMA/iw_cxgb4: atomically lookup ep and get a reference")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Raju Rangoju <rajur@chelsio.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/mad: Reduce MAD scope to mlx5_ib only
Leon Romanovsky [Sun, 13 Jan 2019 13:57:04 +0000 (15:57 +0200)]
RDMA/mad: Reduce MAD scope to mlx5_ib only

Management Datagram Interface (MAD) is applicable
only when physical port is Infiniband. It makes MAD
command logic to be completely unrelated to eth/core
parts of mlx5.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Acked-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/cma: Rollback source IP address if failing to acquire device
Myungho Jung [Thu, 10 Jan 2019 06:27:31 +0000 (22:27 -0800)]
RDMA/cma: Rollback source IP address if failing to acquire device

If cma_acquire_dev_by_src_ip() returns error in addr_handler(), the
device state changes back to RDMA_CM_ADDR_BOUND but the resolved source
IP address is still left. After that, if rdma_destroy_id() is called
after rdma_listen(), the device is freed without removed from
listen_any_list in cma_cancel_operation(). Revert to the previous IP
address if acquiring device fails.

Reported-by: syzbot+f3ce716af730c8f96637@syzkaller.appspotmail.com
Signed-off-by: Myungho Jung <mhjungk@gmail.com>
Reviewed-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA/bnxt_re: fix a size calculation
Dan Carpenter [Thu, 10 Jan 2019 13:00:19 +0000 (16:00 +0300)]
RDMA/bnxt_re: fix a size calculation

This is from static analysis not from testing.  Depending on the value
of rcfw->cmdq_depth, then this might not cause an issue at runtime.

The BITS_TO_LONGS() macro tells us how many longs it take to hold a
bitmap.  In other words, it divides by the number if bits per long and
rounds up.  Then we want to take that number and multiple by
sizeof(long) to get the number of bytes to allocate.

The code here does the multiplication first so the rounding up is done
in the wrong place.  So imagine we want to allocate 1 bit, then
"(1 * 8) / 64 = 1" when we round up.  But it should be
"(1 / 64) * 8 = 8".  In other words, because of the rounding difference
we might allocate up to "sizeof(long) - 1" bytes fewer than intended.

Fixes: 1ac5a4047975 ("RDMA/bnxt_re: Add bnxt_re RoCE driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-By: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA: Introduce and use rdma_device_to_ibdev()
Parav Pandit [Tue, 18 Dec 2018 12:15:56 +0000 (14:15 +0200)]
RDMA: Introduce and use rdma_device_to_ibdev()

Introduce and use rdma_device_to_ibdev() API for those drivers which are
registering one sysfs group and also use in ib_core.

In subsequent patch, device->provider_ibdev one-to-one mapping is no
longer holds true during accessing sysfs entries.
Therefore, introduce an API rdma_device_to_ibdev() that provides such
information.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoRDMA: Rename port_callback to init_port
Parav Pandit [Tue, 18 Dec 2018 12:28:30 +0000 (14:28 +0200)]
RDMA: Rename port_callback to init_port

Most provider routines are callback routines which ib core invokes.
_callback suffix doesn't convey information about when such callback is
invoked. Therefore, rename port_callback to init_port.

Additionally, store the init_port function pointer in ib_device_ops, so
that it can be accessed in subsequent patches when binding rdma device to
net namespace.

Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agoLinux 5.0-rc2 v5.0-rc2
Linus Torvalds [Sun, 13 Jan 2019 22:41:12 +0000 (10:41 +1200)]
Linux 5.0-rc2

5 years agokernel/sys.c: Clarify that UNAME26 does not generate unique versions anymore
Jonathan Neuschäfer [Sat, 12 Jan 2019 17:14:30 +0000 (18:14 +0100)]
kernel/sys.c: Clarify that UNAME26 does not generate unique versions anymore

UNAME26 is a mechanism to report Linux's version as 2.6.x, for
compatibility with old/broken software.  Due to the way it is
implemented, it would have to be updated after 5.0, to keep the
resulting versions unique.  Linus Torvalds argued:

 "Do we actually need this?

  I'd rather let it bitrot, and just let it return random versions. It
  will just start again at 2.4.60, won't it?

  Anybody who uses UNAME26 for a 5.x kernel might as well think it's
  still 4.x. The user space is so old that it can't possibly care about
  differences between 4.x and 5.x, can it?

  The only thing that matters is that it shows "2.4.<largeenough>",
  which it will do regardless"

Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMerge tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Sun, 13 Jan 2019 22:34:14 +0000 (10:34 +1200)]
Merge tag 'armsoc-fixes' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC fixes from Olof Johansson:
 "A bigger batch than I anticipated this week, for two reasons:

   - Some fallout on Davinci from board file -> DTB conversion, that
     also includes a few longer-standing fixes (i.e. not recent
     regressions).

   - drivers/reset material that has been in linux-next for a while, but
     didn't get sent to us until now for a variety of reasons
     (maintainer out sick, holidays, etc). There's a functional
     dependency in there such that one platform (Altera's SoCFPGA) won't
     boot without one of the patches; instead of reverting the patch
     that got merged, I looked at this set and decided it was small
     enough that I'll pick it up anyway. If you disagree I can revisit
     with a smaller set.

  That being said, there's also a handful of the usual stuff:

   - Fix for a crash on Armada 7K/8K when the kernel touches
     PSCI-reserved memory

   - Fix for PCIe reset on Macchiatobin (Armada 8K development board,
     what this email is sent from in fact :)

   - Enable a few new-merged modules for Amlogic in arm64 defconfig

   - Error path fixes on Integrator

   - Build fix for Renesas and Qualcomm

   - Initialization fix for Renesas RZ/G2E

  .. plus a few more fixlets"

* tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (28 commits)
  ARM: integrator: impd1: use struct_size() in devm_kzalloc()
  qcom-scm: Include <linux/err.h> header
  gpio: pl061: handle failed allocations
  ARM: dts: kirkwood: Fix polarity of GPIO fan lines
  arm64: dts: marvell: mcbin: fix PCIe reset signal
  arm64: dts: marvell: armada-ap806: reserve PSCI area
  ARM: dts: da850-lcdk: Correct the sound card name
  ARM: dts: da850-lcdk: Correct the audio codec regulators
  ARM: dts: da850-evm: Correct the sound card name
  ARM: dts: da850-evm: Correct the audio codec regulators
  ARM: davinci: omapl138-hawk: fix label names in GPIO lookup entries
  ARM: davinci: dm644x-evm: fix label names in GPIO lookup entries
  ARM: davinci: dm355-evm: fix label names in GPIO lookup entries
  ARM: davinci: da850-evm: fix label names in GPIO lookup entries
  ARM: davinci: da830-evm: fix label names in GPIO lookup entries
  arm64: defconfig: enable modules for amlogic s400 sound card
  reset: uniphier-glue: Add AHCI reset control support in glue layer
  dt-bindings: reset: uniphier: Add AHCI core reset description
  reset: uniphier-usb3: Rename to reset-uniphier-glue
  dt-bindings: reset: uniphier: Replace the expression of USB3 with generic peripherals
  ...

5 years agoMerge tag 'for-5.0-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave...
Linus Torvalds [Sun, 13 Jan 2019 17:55:51 +0000 (05:55 +1200)]
Merge tag 'for-5.0-rc1-tag' of git://git./linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:

 - two regression fixes in clone/dedupe ioctls, the generic check
   callback needs to lock extents properly and wait for io to avoid
   problems with writeback and relocation

 - fix deadlock when using free space tree due to block group creation

 - a recently added check refuses a valid fileystem with seeding device,
   make that work again with a quickfix, proper solution needs more
   intrusive changes

* tag 'for-5.0-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: Use real device structure to verify dev extent
  Btrfs: fix deadlock when using free space tree due to block group creation
  Btrfs: fix race between reflink/dedupe and relocation
  Btrfs: fix race between cloning range ending at eof and writeback

5 years agoMerge tag 'driver-core-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 13 Jan 2019 17:51:08 +0000 (05:51 +1200)]
Merge tag 'driver-core-5.0-rc2' of git://git./linux/kernel/git/gregkh/driver-core

Pull driver core fixes from Greg KH:
 "Here is one small sysfs change, and a documentation update for 5.0-rc2

  The sysfs change moves from using BUG_ON to WARN_ON, as discussed in
  an email thread on lkml while trying to track down another driver bug.
  sysfs should not be crashing and preventing people from seeing where
  they went wrong. Now it properly recovers and warns the developer.

  The documentation update removes the use of BUS_ATTR() as the kernel
  is moving away from this to use the specific BUS_ATTR_RW() and friends
  instead. There are pending patches in all of the different subsystems
  to remove the last users of this macro, but for now, don't advertise
  it should be used anymore to keep new ones from being introduced.

  Both have been in linux-next with no reported issues"

* tag 'driver-core-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core:
  Documentation: driver core: remove use of BUS_ATTR
  sysfs: convert BUG_ON to WARN_ON

5 years agoMerge tag 'staging-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 13 Jan 2019 17:49:35 +0000 (05:49 +1200)]
Merge tag 'staging-5.0-rc2' of git://git./linux/kernel/git/gregkh/staging

Pull staging driver fixes from Greg KH:
 "Here are some small staging driver fixes for some reported issues.

  One reverts a patch that was made to the rtl8723bs driver that turned
  out to not be needed at all as it was a bug in clang. The others fix
  up some reported issues in the rtl8188eu driver and update the
  MAINTAINERS file to point to Larry for this driver so he can get the
  bug reports easier.

  All have been in linux-next with no reported issues"

* tag 'staging-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  Revert "staging: rtl8723bs: Mark ACPI table declaration as used"
  staging: rtl8188eu: Fix module loading from tasklet for WEP encryption
  staging: rtl8188eu: Fix module loading from tasklet for CCMP encryption
  MAINTAINERS: Add entry for staging driver r8188eu

5 years agoMerge tag 'tty-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 13 Jan 2019 17:47:48 +0000 (05:47 +1200)]
Merge tag 'tty-5.0-rc2' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial fixes from Greg KH:
 "Here are 2 tty and serial fixes for 5.0-rc2 that resolve some reported
  issues.

  The first is a simple serial driver fix for a regression that showed
  up in 5.0-rc1. The second one resolves a number of reported issues
  with the recent tty locking fixes that went into 5.0-rc1. Lots of
  people have tested the second one and say it resolves their issues.

  Both have been in linux-next with no reported issues"

* tag 'tty-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  tty: Don't hold ldisc lock in tty_reopen() if ldisc present
  serial: lantiq: Do not swap register read/writes

5 years agoMerge tag 'usb-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sun, 13 Jan 2019 17:45:28 +0000 (05:45 +1200)]
Merge tag 'usb-5.0-rc2' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB driver fixes and quirk updates for 5.0-rc2.

  The majority here are some quirks for some storage devices to get them
  to work properly. There's also a fix here to resolve the reported
  issues with some audio devices that say they are UAC3 compliant, but
  really are not.

  And a fix up for the MAINTAINERS file to remove a dead url.

  All have been in linux-next with no reported issues"

* tag 'usb-5.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: storage: Remove outdated URL from MAINTAINERS
  USB: Add USB_QUIRK_DELAY_CTRL_MSG quirk for Corsair K70 RGB
  usbcore: Select only first configuration for non-UAC3 compliant devices
  USB: storage: add quirk for SMI SM3350
  USB: storage: don't insert sane sense for SPC3+ when bad sense specified
  usb: cdc-acm: send ZLP for Telit 3G Intel based modems

5 years agoMerge tag '5.0-rc1-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sun, 13 Jan 2019 17:43:40 +0000 (05:43 +1200)]
Merge tag '5.0-rc1-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fixes from Steve French:
 "A set of cifs/smb3 fixes, 4 for stable, most from Pavel. His patches
  fix an important set of crediting (flow control) problems, and also
  two problems in cifs_writepages, ddressing some large i/o and also
  compounding issues"

* tag '5.0-rc1-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: update internal module version number
  CIFS: Fix error paths in writeback code
  CIFS: Move credit processing to mid callbacks for SMB3
  CIFS: Fix credits calculation for cancelled requests
  cifs: Fix potential OOB access of lock element array
  cifs: Limit memory used by lock request calls to a page
  cifs: move large array from stack to heap
  CIFS: Do not hide EINTR after sending network packets
  CIFS: Fix credit computation for compounded requests
  CIFS: Do not set credits to 1 if the server didn't grant anything
  CIFS: Fix adjustment of credits for MTU requests
  cifs: Fix a tiny potential memory leak
  cifs: Fix a debug message

5 years agoMerge tag 'reset-for-5.0-rc2' of git://git.pengutronix.de/git/pza/linux into fixes
Olof Johansson [Sun, 13 Jan 2019 06:06:54 +0000 (22:06 -0800)]
Merge tag 'reset-for-5.0-rc2' of git://git.pengutronix.de/git/pza/linux into fixes

Late reset controller changes for v5.0

This adds missing deassert functionality to the ARC HSDK reset driver,
fixes some indentation and grammar issues in the kernel docs, adds a
helper to count the number of resets on a device for the non-DT case
as well, adds an early reset driver for SoCFPGA and simple reset driver
support for Stratix10, and generalizes the uniphier USB3 glue layer
reset to also cover AHCI.

* tag 'reset-for-5.0-rc2' of git://git.pengutronix.de/git/pza/linux:
  reset: uniphier-glue: Add AHCI reset control support in glue layer
  dt-bindings: reset: uniphier: Add AHCI core reset description
  reset: uniphier-usb3: Rename to reset-uniphier-glue
  dt-bindings: reset: uniphier: Replace the expression of USB3 with generic peripherals
  ARM: socfpga: dts: document "altr,stratix10-rst-mgr" binding
  reset: socfpga: add an early reset driver for SoCFPGA
  reset: fix null pointer dereference on dev by dev_name
  reset: Add reset_control_get_count()
  reset: Improve reset controller kernel docs
  ARC: HSDK: improve reset driver

Signed-off-by: Olof Johansson <olof@lixom.net>
5 years agoMerge tag 'mvebu-fixes-5.0-1' of git://git.infradead.org/linux-mvebu into fixes
Olof Johansson [Sun, 13 Jan 2019 06:03:59 +0000 (22:03 -0800)]
Merge tag 'mvebu-fixes-5.0-1' of git://git.infradead.org/linux-mvebu into fixes

mvebu fixes for 5.0

They are all device tree fixes which also worth being in stable:

 - Reserve PSCI area on Armada 7K/8K preventing the kernel accessing
   this area and crashing while doing it.

 - Use correct PCIe reset signal on MACCHIATOBin  (Armada 8040 based)

 - Fix polarity of GPIO fan line D-Link DNS NASes(kikwood based)

* tag 'mvebu-fixes-5.0-1' of git://git.infradead.org/linux-mvebu:
  ARM: dts: kirkwood: Fix polarity of GPIO fan lines
  arm64: dts: marvell: mcbin: fix PCIe reset signal
  arm64: dts: marvell: armada-ap806: reserve PSCI area

Signed-off-by: Olof Johansson <olof@lixom.net>
5 years agoMerge tag 'integrator-fixes-armsoc' of git://git.kernel.org/pub/scm/linux/kernel...
Olof Johansson [Sun, 13 Jan 2019 06:03:18 +0000 (22:03 -0800)]
Merge tag 'integrator-fixes-armsoc' of git://git./linux/kernel/git/linusw/linux-integrator into fixes

Fixes for the Integrator:
- Handle failed allocations in the IM/PC bus attachment.
- Use struct_size() for allocation.

* tag 'integrator-fixes-armsoc' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-integrator:
  ARM: integrator: impd1: use struct_size() in devm_kzalloc()
  gpio: pl061: handle failed allocations

Signed-off-by: Olof Johansson <olof@lixom.net>
5 years agoMerge tag 'amlogic-fixes' of https://git.kernel.org/pub/scm/linux/kernel/git/khilman...
Olof Johansson [Sun, 13 Jan 2019 06:02:28 +0000 (22:02 -0800)]
Merge tag 'amlogic-fixes' of https://git./linux/kernel/git/khilman/linux-amlogic into fixes

Amlogic DT fixes for v5.0-rc
- arm64: defconfig: enable modules for amlogic s400 sound card

* tag 'amlogic-fixes' of https://git.kernel.org/pub/scm/linux/kernel/git/khilman/linux-amlogic:
  arm64: defconfig: enable modules for amlogic s400 sound card

Signed-off-by: Olof Johansson <olof@lixom.net>
5 years agoMerge tag 'qcom-fixes-for-5.0-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Olof Johansson [Sun, 13 Jan 2019 06:01:49 +0000 (22:01 -0800)]
Merge tag 'qcom-fixes-for-5.0-rc1' of git://git./linux/kernel/git/agross/linux into fixes

Qualcomm Driver Fixes for 5.0-rc1

* Add required includes into qcom_scm.h

* tag 'qcom-fixes-for-5.0-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/agross/linux:
  qcom-scm: Include <linux/err.h> header

Signed-off-by: Olof Johansson <olof@lixom.net>
5 years agoMerge tag 'davinci-fixes-for-v5.0' of git://git.kernel.org/pub/scm/linux/kernel/git...
Olof Johansson [Sun, 13 Jan 2019 06:00:36 +0000 (22:00 -0800)]
Merge tag 'davinci-fixes-for-v5.0' of git://git./linux/kernel/git/nsekhar/linux-davinci into fixes

This pull request fixes some more regressions on legacy
DaVinci board support due to GPIO driver clean-up introduced
in v4.20 kernel. These are marked for stable.

Also has fixes for some long standing Audio issues on DA850
boards.

* tag 'davinci-fixes-for-v5.0' of git://git.kernel.org/pub/scm/linux/kernel/git/nsekhar/linux-davinci:
  ARM: dts: da850-lcdk: Correct the sound card name
  ARM: dts: da850-lcdk: Correct the audio codec regulators
  ARM: dts: da850-evm: Correct the sound card name
  ARM: dts: da850-evm: Correct the audio codec regulators
  ARM: davinci: omapl138-hawk: fix label names in GPIO lookup entries
  ARM: davinci: dm644x-evm: fix label names in GPIO lookup entries
  ARM: davinci: dm355-evm: fix label names in GPIO lookup entries
  ARM: davinci: da850-evm: fix label names in GPIO lookup entries
  ARM: davinci: da830-evm: fix label names in GPIO lookup entries

Signed-off-by: Olof Johansson <olof@lixom.net>
5 years agoMerge tag 'renesas-fixes-for-v5.0' of https://git.kernel.org/pub/scm/linux/kernel...
Olof Johansson [Sun, 13 Jan 2019 05:59:42 +0000 (21:59 -0800)]
Merge tag 'renesas-fixes-for-v5.0' of https://git./linux/kernel/git/horms/renesas into fixes

Renesas ARM Based SoC Fixes for v5.0

Renesas SoCs:
* Fix build regressions caused by move of Kconfig symbols

RZ/G2E (r8a774c0) SoC:
* Correct initialization order of 3DG-{A,B} in SYSC driver

* tag 'renesas-fixes-for-v5.0' of https://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas:
  soc: renesas: r8a774c0-sysc: Fix initialization order of 3DG-{A,B}
  ARM: shmobile: fix build regressions

Signed-off-by: Olof Johansson <olof@lixom.net>
5 years agophy: fix build breakage: add PHY_MODE_SATA
John Hubbard [Sun, 13 Jan 2019 01:29:09 +0000 (17:29 -0800)]
phy: fix build breakage: add PHY_MODE_SATA

Commit 49e54187ae0b ("ata: libahci_platform: comply to PHY framework") uses
the PHY_MODE_SATA, but that enum had not yet been added. This caused a
build failure for me, with today's linux.git.

Also, there is a potentially conflicting (mis-named) PHY_MODE_SATA, hiding
in the Marvell Berlin SATA PHY driver.

Fix the build by:

    1) Renaming Marvell's defined value to a more scoped name,
       in order to avoid any potential conflicts: PHY_BERLIN_MODE_SATA.

    2) Adding the missing enum, which was going to be added anyway as part
       of [1].

[1] https://lkml.kernel.org/r/20190108163124.6409-3-miquel.raynal@bootlin.com

Fixes: 49e54187ae0b ("ata: libahci_platform: comply to PHY framework")

Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Acked-by: Olof Johansson <olof@lixom.net>
Cc: Grzegorz Jaszczyk <jaz@semihalf.com>
Cc: Miquel Raynal <miquel.raynal@bootlin.com>
Cc: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMerge tag 'for-linus-20190112' of git://git.kernel.dk/linux-block
Linus Torvalds [Sat, 12 Jan 2019 21:40:51 +0000 (13:40 -0800)]
Merge tag 'for-linus-20190112' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:

 - NVMe pull request from Christoph, with little fixes all over the map

 - Loop caching fix for offset/bs change (Jaegeuk Kim)

 - Block documentation tweaks (Jeff, Jon, Weiping, John)

 - null_blk zoned tweak (John)

 - ahch mvebu suspend/resume support. Should have gone into the merge
   window, but there was some confusion on which tree had it. (Miquel)

* tag 'for-linus-20190112' of git://git.kernel.dk/linux-block: (22 commits)
  ata: ahci: mvebu: request PHY suspend/resume for Armada 3700
  ata: ahci: mvebu: add Armada 3700 initialization needed for S2RAM
  ata: ahci: mvebu: do Armada 38x configuration only on relevant SoCs
  ata: ahci: mvebu: remove stale comment
  ata: libahci_platform: comply to PHY framework
  loop: drop caches if offset or block_size are changed
  block: fix kerneldoc comment for blk_attempt_plug_merge()
  nvme: don't initlialize ctrl->cntlid twice
  nvme: introduce NVME_QUIRK_IGNORE_DEV_SUBNQN
  nvme: pad fake subsys NQN vid and ssvid with zeros
  nvme-multipath: zero out ANA log buffer
  nvme-fabrics: unset write/poll queues for discovery controllers
  nvme-tcp: don't ask if controller is fabrics
  nvme-tcp: remove dead code
  nvme-pci: fix out of bounds access in nvme_cqe_pending
  nvme-pci: rerun irq setup on IO queue init errors
  nvme-pci: use the same attributes when freeing host_mem_desc_bufs.
  nvme-pci: fix the wrong setting of nr_maps
  block: doc: add slice_idle_us to bfq documentation
  block: clarify documentation for blk_{start|finish}_plug
  ...

5 years agoMerge tag 'remove-dma_zalloc_coherent-5.0' of git://git.infradead.org/users/hch/dma...
Linus Torvalds [Sat, 12 Jan 2019 18:52:40 +0000 (10:52 -0800)]
Merge tag 'remove-dma_zalloc_coherent-5.0' of git://git.infradead.org/users/hch/dma-mapping

Pull dma_zalloc_coherent() removal from Christoph Hellwig:
 "We've always had a weird situation around dma_zalloc_coherent. To
  safely support mapping the allocations to userspace major
  architectures like x86 and arm have always zeroed allocations from
  dma_alloc_coherent, but a couple other architectures were missing that
  zeroing either always or in corner cases.

  Then later we grew anothe dma_zalloc_coherent interface to explicitly
  request zeroing, but that just added __GFP_ZERO to the allocation
  flags, which for some allocators that didn't end up using the page
  allocator ended up being a no-op and still not zeroing the
  allocations.

  So for this merge window I fixed up all remaining architectures to
  zero the memory in dma_alloc_coherent, and made dma_zalloc_coherent a
  no-op wrapper around dma_alloc_coherent, which fixes all of the above
  issues.

  dma_zalloc_coherent is now pointless and can go away, and Luis helped
  me writing a cocchinelle script and patch series to kill it, which I
  think we should apply now just after -rc1 to finally settle these
  issue"

* tag 'remove-dma_zalloc_coherent-5.0' of git://git.infradead.org/users/hch/dma-mapping:
  dma-mapping: remove dma_zalloc_coherent()
  cross-tree: phase out dma_zalloc_coherent() on headers
  cross-tree: phase out dma_zalloc_coherent()

5 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sat, 12 Jan 2019 18:39:43 +0000 (10:39 -0800)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM fixes from Radim Krčmář:
 "Minor fixes for new code, corner cases, and documentation"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  x86/kvm/nVMX: don't skip emulated instruction twice when vmptr address is not backed
  Documentation/virtual/kvm: Update URL for AMD SEV API specification
  KVM/VMX: Avoid return error when flush tlb successfully in the hv_remote_flush_tlb_with_range()
  kvm: sev: Fail KVM_SEV_INIT if already initialized
  KVM: validate userspace input in kvm_clear_dirty_log_protect()
  KVM: x86: Fix bit shifting in update_intel_pt_cfg