Dan Carpenter [Sat, 30 May 2020 14:02:24 +0000 (17:02 +0300)]
IB/hfi1: Fix hfi1_netdev_rx_init() error handling
The hfi1_vnic_up() function doesn't check whether hfi1_netdev_rx_init()
returns errors. In hfi1_vnic_init() we need to change the code to
preserve the error code instead of returning success.
Fixes: 2280740f01ae ("IB/hfi1: Virtual Network Interface Controller (VNIC) HW support") Fixes: 4730f4a6c6b2 ("IB/hfi1: Activate the dummy netdev") Link: https://lore.kernel.org/r/20200530140224.GA1330098@mwanda Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Yamin Friedman [Wed, 27 May 2020 08:34:53 +0000 (11:34 +0300)]
RDMA/core: Introduce shared CQ pool API
Allow a ULP to ask the core to provide a completion queue based on a
least-used search on a per-device CQ pools. The device CQ pools grow in a
lazy fashion when more CQs are requested.
This feature reduces the amount of interrupts when using many QPs. Using
shared CQs allows for more effcient completion handling. It also reduces
the amount of overhead needed for CQ contexts.
Test setup:
Intel(R) Xeon(R) Platinum 8176M CPU @ 2.10GHz servers.
Running NVMeoF 4KB read IOs over ConnectX-5EX across Spectrum switch.
TX-depth = 32. The patch was applied in the nvme driver on both the target
and initiator. Four controllers are accessed from each core. In the
current test case we have exposed sixteen NVMe namespaces using four
different subsystems (four namespaces per subsystem) from one NVM port.
Each controller allocated X queues (RDMA QPs) and attached to Y CQs.
Before this series we had X == Y, i.e for four controllers we've created
total of 4X QPs and 4X CQs. In the shared case, we've created 4X QPs and
only X CQs which means that we have four controllers that share a
completion queue per core. Until fourteen cores there is no significant
change in performance and the number of interrupts per second is less than
a million in the current case.
==================================================
|Cores|Current KIOPs |Shared KIOPs |improvement|
|-----|---------------|--------------|-----------|
|14 |2332 |2723 |16.7% |
|-----|---------------|--------------|-----------|
|20 |2086 |2712 |30% |
|-----|---------------|--------------|-----------|
|28 |1971 |2669 |35.4% |
|=================================================
|Cores|Current avg lat|Shared avg lat|improvement|
|-----|---------------|--------------|-----------|
|14 |767us |657us |14.3% |
|-----|---------------|--------------|-----------|
|20 |1225us |943us |23% |
|-----|---------------|--------------|-----------|
|28 |1816us |1341us |26.1% |
========================================================
|Cores|Current interrupts|Shared interrupts|improvement|
|-----|------------------|-----------------|-----------|
|14 |1.6M/sec |0.4M/sec |72% |
|-----|------------------|-----------------|-----------|
|20 |2.8M/sec |0.6M/sec |72.4% |
|-----|------------------|-----------------|-----------|
|28 |2.9M/sec |0.8M/sec |63.4% |
====================================================================
|Cores|Current 99.99th PCTL lat|Shared 99.99th PCTL lat|improvement|
|-----|------------------------|-----------------------|-----------|
|14 |67ms |6ms |90.9% |
|-----|------------------------|-----------------------|-----------|
|20 |5ms |6ms |-10% |
|-----|------------------------|-----------------------|-----------|
|28 |8.7ms |6ms |25.9% |
|===================================================================
Performance improvement with sixteen disks (sixteen CQs per core) is
comparable.
Link: https://lore.kernel.org/r/1590568495-101621-3-git-send-email-yaminf@mellanox.com Signed-off-by: Yamin Friedman <yaminf@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Reviewed-by: Max Gurtovoy <maxg@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Yamin Friedman [Wed, 27 May 2020 08:34:52 +0000 (11:34 +0300)]
RDMA/core: Add protection for shared CQs used by ULPs
A pre-step for adding shared CQs. Add the infrastructure to prevent shared
CQ users from altering the CQ configurations. For now all cqs are marked
as private (non-shared). The core driver should use the new force
functions to perform resize/destroy/moderation changes that are not
allowed for users of shared CQs.
Link: https://lore.kernel.org/r/1590568495-101621-2-git-send-email-yaminf@mellanox.com Signed-off-by: Yamin Friedman <yaminf@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Reviewed-by: Max Gurtovoy <maxg@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Qiushi Wu [Thu, 28 May 2020 03:02:30 +0000 (22:02 -0500)]
RDMA/core: Fix several reference count leaks.
kobject_init_and_add() takes reference even when it fails. If this
function returns an error, kobject_put() must be called to properly clean
up the memory associated with the object. Previous
commit b8eb718348b8 ("net-sysfs: Fix reference count leak in
rx|netdev_queue_add_kobject") fixed a similar problem.
Colin Ian King [Thu, 28 May 2020 15:04:27 +0000 (16:04 +0100)]
RDMA/hns: remove duplicate assignment to pointer raq
The pointer raq is being assigned twice. Fix this by removing one of the
redundant assignments.
Fixes: 14ba87304bf9 ("RDMA/hns: Remove redundant type cast for general pointers") Link: https://lore.kernel.org/r/20200528150427.420624-1-colin.king@canonical.com
Addressses-Coverity: ("Evaluation order violation") Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Mark Zhang [Wed, 27 May 2020 05:50:14 +0000 (08:50 +0300)]
RDMA/mlx5: Support TX port affinity for VF drivers in LAG mode
The mlx5 VF driver doesn't set QP tx port affinity because it doesn't know
if the lag is active or not, since the "lag_active" works only for PF
interfaces. In this case for VF interfaces only one lag is used which
brings performance issue.
Add a lag_tx_port_affinity CAP bit; When it is enabled and
"num_lag_ports > 1", then driver always set QP tx affinity, regardless
of lag state.
Link: https://lore.kernel.org/r/20200527055014.355093-1-leon@kernel.org Signed-off-by: Mark Zhang <markz@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Bart Van Assche [Mon, 25 May 2020 17:22:12 +0000 (10:22 -0700)]
RDMA/srpt: Increase max_send_sge
The ib_srpt driver limits max_send_sge to 16. Since that is a workaround
for an mlx4 bug that has been fixed, increase max_send_sge. See also
commit f95ccffc715b ("IB/mlx4: Use 4K pages for kernel QP's WQE buffer").
Bart Van Assche [Mon, 25 May 2020 17:22:10 +0000 (10:22 -0700)]
RDMA/srpt: Make debug output more detailed
Since the session name by itself is not sufficient to uniquely identify a
queue pair, include the queue pair number. Show the ASCII channel state
name instead of the numeric value. This change makes the ib_srpt debug
output more consistent.
Bart Van Assche [Mon, 25 May 2020 17:22:09 +0000 (10:22 -0700)]
RDMA/srp: Make the channel count configurable per target
Increase the flexibility of the SRP initiator driver by making the channel
count configurable per target instead of only providing a kernel module
parameter for configuring the channel count.
Leon Romanovsky [Tue, 26 May 2020 11:54:40 +0000 (14:54 +0300)]
RDMA/mlx5: Return ECE data after modify QP
After users sets the ECE option, FW will return the agreed/supported bits
through an output structures of modify QP stages for regular QPs or
through create QP for the DCT.
Leon Romanovsky [Tue, 26 May 2020 11:54:39 +0000 (14:54 +0300)]
RDMA/mlx5: Set ECE options during modify QP
The most common way to set ECE option will be during modify QP command in
INIT2RTR, RTR2RTS and RTS2RTS stages, so update mlx5 to support it.
The new bit in the comp_mask is needed to mark that kernel supports ECE
and can receive data instead of "reserved" field in the struct
mlx5_ib_modify_qp.
Leon Romanovsky [Tue, 26 May 2020 11:54:38 +0000 (14:54 +0300)]
RDMA/mlx5: Convert modify QP to use MLX5_SET macros
Instead of hand crafted mlx5_qp_context and mlx5_qp_path use common
MLX5_SET() macros.
Link: https://lore.kernel.org/r/20200526115440.205922-7-leon@kernel.org Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Reviewed-by: Mark Zhang <markz@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Leon Romanovsky [Tue, 26 May 2020 11:54:35 +0000 (14:54 +0300)]
RDMA/mlx5: Set ECE options during QP create
Allow users to ask creation of QPs with specific ECE options. Such early
set even before RDMA-CM connection is established is useful if user knows
exactly which option he needs.
Leon Romanovsky [Tue, 26 May 2020 11:54:34 +0000 (14:54 +0300)]
RDMA/mlx5: Get ECE options from FW during create QP
Supported ECE options are returned from FW in the create_qp phase and zero
means that field is not valid. Such default value allows us to reuse
reserved field without worries about comp_mask.
Leon Romanovsky [Tue, 26 May 2020 10:33:03 +0000 (13:33 +0300)]
RDMA/cma: Connect ECE to rdma_accept
The rdma_accept() is called by both passive and active sides of CMID
connection to mark readiness to start data transfer. For passive side,
this is called explicitly, for active side, it is called implicitly while
receiving REP message.
Provide ECE data to rdma_accept function needed for passive side to send
that REP message.
Leon Romanovsky [Tue, 26 May 2020 10:33:01 +0000 (13:33 +0300)]
RDMA/ucma: Deliver ECE parameters through UCMA events
Passive side of CMID connection receives ECE request through REQ message
and needs to respond with relevant REP message which will be forwarded to
active side.
The UCMA events interface is responsible for such communication with the
user space (librdmacm). Extend it to provide ECE wire data.
Leon Romanovsky [Tue, 26 May 2020 10:33:00 +0000 (13:33 +0300)]
RDMA/ucma: Extend ucma_connect to receive ECE parameters
Active side of CMID initiates connection through librdmacm's
rdma_connect() and kernel's ucma_connect(). Extend UCMA interface to
handle those new parameters.
Extend REQ (request for communications), REP (reply to request for
communication), rejected reason and SIDR_REP (service ID resolution
response) structures with hardware vendor ID bits according to IBTA v1.4.
Jason Gunthorpe [Wed, 27 May 2020 19:01:17 +0000 (16:01 -0300)]
Merge branch 'mellanox/mlx5-next' into rdma.git for/next
From the mlx5-next branch at
git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux
Required for dependencies in following patches
* branch 'mellanox/mlx5-next':
net/mlx5: Add ability to read and write ECE options
net/mlx5: Add support for RDMA TX FT headers modifying
net/mlx5: Move iseg access helper routines close to mlx5_core driver
net/mlx5: Cleanup mlx5_ifc_fte_match_set_misc2_bits
Mark Zhang [Wed, 27 May 2020 13:57:03 +0000 (16:57 +0300)]
IB/mlx5: Fix DEVX support for MLX5_CMD_OP_INIT2INIT_QP command
The commit citied in the Fixes line wasn't complete and solved
only part of the problems. Update the mlx5_ib to properly support
MLX5_CMD_OP_INIT2INIT_QP command in the DEVX, that is required when
modify the QP tx_port_affinity.
Fixes: 819f7427bafd ("RDMA/mlx5: Add init2init as a modify command") Link: https://lore.kernel.org/r/20200527135703.482501-1-leon@kernel.org Signed-off-by: Mark Zhang <markz@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Leon Romanovsky [Mon, 9 Mar 2020 14:44:25 +0000 (16:44 +0200)]
net/mlx5: Add ability to read and write ECE options
The end result of RDMA-CM ECE handshake is ECE options, which is
needed to be used while configuring data QPs. Such options can
come in any QP state, so add in/out fields to set and query
ECE options.
OUT fields:
* create_qp() - default ECE options for that type of QP.
* modify_qp() - enabled ECE options after QP state transition.
IN fields:
* create_qp() - create QP with this ECE option.
* modify_qp() - requested options. For unconnected QPs, the FW
will return an error if ECE is already configured with any options
that not equal to previously set.
Reviewed-by: Mark Zhang <markz@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Yixian Liu [Fri, 22 May 2020 13:02:59 +0000 (21:02 +0800)]
RDMA/hns: Make the end of sge process more clear
Instead of i with the sge number of wr will make the comparision more
clear, that is, when the sge number in wr is small than the maximum
supported sge number in the queue, then a stop sge needed to be filled at
the end of sges in wr.
Weihang Li [Fri, 22 May 2020 13:02:56 +0000 (21:02 +0800)]
RDMA/hns: Remove redundant type cast for general pointers
There is no need to do a type cast on genernal pointers, they could be
assigned to any type of variables. In addition, optimize initialization of
some variables and adjust order of them.
Xi Wang [Wed, 20 May 2020 13:53:19 +0000 (21:53 +0800)]
RDMA/hns: Optimize the usage of MTR
Currently, the MTR region is configed before hns_roce_mtr_map() is
invoked, but in some scenarios, the region is configed at MTR creation,
the caller need to store this config and call hns_roce_mtr_map() later. So
optimize the usage by wrapping the MTR region config into MTR.
Xi Wang [Wed, 20 May 2020 13:53:18 +0000 (21:53 +0800)]
RDMA/hns: Refactor the QP context filling process related to WQE buffer configure
Split the code related to WQE buffer configure from the QPC filling
process into two functions: config_qp_sq_buf() and config_qp_rq_buf(),
this will make the code more readable.
Lang Cheng [Wed, 20 May 2020 13:53:12 +0000 (21:53 +0800)]
RDMA/hns: Add CQ flag instead of independent enable flag
It's easier to understand and maintain enable flags of cq using a single
field in type of u32 than defining a field for every flags in the
structure hns_roce_cq, and we can add new flags for features more
conveniently in the future.
Lang Cheng [Wed, 20 May 2020 13:53:11 +0000 (21:53 +0800)]
RDMA/hns: Let software PI/CI grow naturally
The hardware can truncate PI/CI when posting or polling, the driver does
not need to do truncation. Therefore keep the software's PI/CI consistent
with it in the hardware.
Danil Kipnis [Thu, 21 May 2020 18:59:09 +0000 (20:59 +0200)]
RDMA/rnbd: Fix compilation error when CONFIG_MODULES is disabled
module_is_live function is only defined when CONFIG_MODULES is enabled.
Use try_module_get instead to check whether the module is being removed.
When module unload and manuall unmapping is happening in parallel, we can
try removing the symlink twice: rnbd_client_exit
vs. rnbd_clt_unmap_dev_store.
This is probably not the best way to deal with this race in general, but
for now this fixes the compilation issue when CONFIG_MODULES is disabled
and has no functional impact. Regression tests passed.
Fixes: 1eb54f8f5dd8 ("block/rnbd: client: sysfs interface functions") Link: https://lore.kernel.org/r/20200521185909.457245-1-danil.kipnis@cloud.ionos.com Reported-by: Randy Dunlap <rdunlap@infradead.org> Suggested-by: Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by: Danil Kipnis <danil.kipnis@cloud.ionos.com> Acked-by: Randy Dunlap <rdunlap@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Maor Gottlieb [Thu, 21 May 2020 07:26:50 +0000 (10:26 +0300)]
IB/cma: Fix ports memory leak in cma_configfs
The allocated ports structure in never freed. The free function should be
called by release_cma_ports_group, but the group is never released since
we don't remove its default group.
Remove default groups when device group is deleted.
Fixes: 045959db65c6 ("IB/cma: Add configfs for rdma_cm") Link: https://lore.kernel.org/r/20200521072650.567908-1-leon@kernel.org Signed-off-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Dan Carpenter [Tue, 19 May 2020 12:03:47 +0000 (15:03 +0300)]
block/rnbd: Fix an IS_ERR() vs NULL check in find_or_create_sess()
The alloc_sess() function returns error pointers, it never returns NULL.
Fixes: f7a7a5c228d4 ("block/rnbd: client: main functionality") Link: https://lore.kernel.org/r/20200519120347.GD42765@mwanda Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Jack Wang <jinpu.wang@cloud.ionos.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Yishai Hadas [Tue, 19 May 2020 07:27:08 +0000 (10:27 +0300)]
IB/uverbs: Move QP, SRQ, WQ type and flags to UAPI
These constants are going to be used in the ioctl interface in coming
patches so they are part of the UAPI, place them in the correct header
for clarity.
Jason Gunthorpe [Tue, 19 May 2020 07:27:05 +0000 (10:27 +0300)]
RDMA/core: Allow the ioctl layer to abort a fully created uobject
While creating a uobject every create reaches a point where the uobject is
fully initialized. For ioctls that go on to copy_to_user this means they
need to open code the destruction of a fully created uobject - ie the
RDMA_REMOVE_DESTROY sort of flow.
Open coding this creates bugs, eg the CQ does not properly flush the
events list when it does its error unwind.
Provide a uverbs_finalize_uobj_create() function which indicates that the
uobject is fully initialized and that abort should call to destroy_hw to
destroy the uobj->object and related.
Methods can call this function if they go on to have error cases after
setting uobj->object. Once done those error cases can simply do return,
without an error unwind.
IB/hfi1: Enable the transmit side of the datagram ipoib netdev
This patch hooks the transmit side of the datagram netdev with
ipoib by setting the rdma_netdev_get_params function for the
hfi1 ib_device_ops structue. It also enables the receiving side
by adding the AIP capability into the default capabilities.
Link: https://lore.kernel.org/r/20200511160712.173205.65700.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Piotr Stankiewicz <piotr.stankiewicz@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Gary Leshner [Mon, 11 May 2020 16:07:06 +0000 (12:07 -0400)]
IB/ipoib: Add capability to switch between datagram and connected mode
This is the prerequisite modification to the ipoib ulp to allow a
rdma netdev to obtain the default ndo ops for init/uninit/open/close.
This is accomplished by setting the netdev ops field within the
callback function passed to the netdev allocation routine which
in turn was passed into the rdma netdev allocation routine.
This allows the rdma netdev to call back into the ulp to create the
resources required for connected mode operation.
Additionally as the ulp is not re-entrant, when switching modes,
the number of real tx queues is set to 1 for the connected mode.
For datagram mode the number of real tx queues is set to the
actual number of tx queues specified at the netdev's allocation.
For the internal ulp netdev the number of tx queues defaults to 1.
It is up to the rdma netdev to specify the actual number it can support.
When the driver does not support a rdma netdev for acceleration,
(-ENOTSUPPORTED return code or the verbs function for allocation is
NULL) the ipoib ulp functions are unaffected by using the internal
netdev allocated by the ipoib ulp.
Link: https://lore.kernel.org/r/20200511160706.173205.19086.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Gary Leshner <Gary.S.Leshner@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Add a simple trace event taking context number and building simple
histogram to print packets distribution between contexts.
Link: https://lore.kernel.org/r/20200511160700.173205.84270.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Gary Leshner [Mon, 11 May 2020 16:06:55 +0000 (12:06 -0400)]
IB/{hfi1, ipoib, rdma}: Broadcast ping sent packets which exceeded mtu size
When in connected mode ipoib sent broadcast pings which exceeded the mtu
size for broadcast addresses.
Add an mtu attribute to the rdma_netdev structure which ipoib sets to its
mcast mtu size.
The RDMA netdev uses this value to determine if the skb length is too long
for the mtu specified and if it is, drops the packet and logs an error
about the errant packet.
Link: https://lore.kernel.org/r/20200511160655.173205.14546.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Gary Leshner <Gary.S.Leshner@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
As described in earlier patches, ipoib netdev will share receive
contexts with existing VNIC netdev through a dummy netdev. The
following changes are made to achieve that:
- Set up netdev receive contexts after user contexts. A function is
added to count the available netdev receive contexts.
- Add functions to set/get receive map table free index.
- Rename NUM_VNIC_MAP_ENTRIES as NUM_NETDEV_MAP_ENTRIES.
- Let the dummy netdev own the receive contexts instead of VNIC.
- Allocate the dummy netdev when the hfi1 device is added and free it
when the device is removed.
- Initialize AIP RSM rules when the IpoIb rxq is initialized and
remove the rules when it is de-initialized.
- Convert VNIC to use the dummy netdev.
Link: https://lore.kernel.org/r/20200511160649.173205.4626.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
This patch adds the rx functions for the dummy netdev:
- Functions to allocate/free the dummy netdev.
- Functions to allocate/free receiving contexts for the netdev.
- Functions to initialize/de-initialize the receive queue.
- Functions to enable/disable the receive queue.
Link: https://lore.kernel.org/r/20200511160643.173205.75087.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
IB/hfi1: Add interrupt handler functions for accelerated ipoib
This patch adds the interrupt handler function, the NAPI poll
function, and its associated helper functions for receiving
accelerated ipoib packets. While we are here, fix the formats
of two error printouts.
Link: https://lore.kernel.org/r/20200511160637.173205.64890.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Kaike Wan [Mon, 11 May 2020 16:06:31 +0000 (12:06 -0400)]
IB/hfi1: Add functions to receive accelerated ipoib packets
Ipoib netdev will share receive contexts with existing VNIC netdev.
To achieve that, a dummy netdev is allocated with hfi1_devdata to
own the receive contexts, and ipoib and VNIC netdevs will be put
on top of it. Each receive context is associated with a single
NAPI object.
This patch adds the functions to receive incoming packets for
accelerated ipoib.
Link: https://lore.kernel.org/r/20200511160631.173205.54184.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
IB/hfi1: Rename num_vnic_contexts as num_netdev_contexts
Rename num_vnic_contexts as num_ndetdev_contexts since VNIC and ipoib
will share the same set of receive contexts.
Link: https://lore.kernel.org/r/20200511160625.173205.53306.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Currently the ipoib UD mtu is restricted to 4K bytes. Remove this
limitation so that the IPOIB module can potentially use an MTU (in UD
mode) that is bounded by the MTU of the underlying device. A field is
added to the ib_port_attr structure to indicate the maximum physical
MTU the underlying device supports.
Link: https://lore.kernel.org/r/20200511160618.173205.23053.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Sadanand Warrier <sadanand.warrier@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
This is implementation of RSM rule for AIP packets.
AIP rule will use rule RSM2 and will match standard
Infiniband packet containg BTH (LNH==BTH) and
having Dest QPN prefixed with value 0x81. Spread between
receive contexts will be done using source QPN bits.
VNIC and AIP will share receive contexts, so their rules
will point to the same RMT entries and their shared
code is moved to separate functions.
If any of the rules is active RMT mapping will be skipped
for latter.
Changed function hfi1_vnic_is_rsm_full to be more general
and moved it from main header to chip.c.
Changed the order of RSM rules because AIP rule as
more specific one is needed to be placed before more
general QOS rule. Rules are occupying two last RSM
registers.
Link: https://lore.kernel.org/r/20200511160612.173205.73002.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Gary Leshner [Mon, 11 May 2020 16:06:07 +0000 (12:06 -0400)]
IB/{rdmavt, hfi1}: Implement creation of accelerated UD QPs
Adds capability to create a qpn to be recognized as an accelerated
UD QP for ipoib.
This is accomplished by reserving 0x81 in byte[0] of the qpn as the
prefix for these qp types and reserving qpns between 0x810000 and
0x81ffff.
The hfi1 capability mask already contained a flag for the VNIC netdev.
This has been renamed and extended to include both VNIC and ipoib.
The rvt code to allocate qps now recognizes this flag and sets 0x81
into byte[0] of the qpn.
The code to allocate qpns is modified to reset the qpn numbering when it
is detected that a value is located in byte[0] for a UD QP and it is a
qpn being requested for net dev use. If it is a regular UD QP then it is
allowable to have bits set in byte[0] of the qpn and provide the
previously normal behavior.
The code to free the qpn now checks for the AIP prefix value of 0x81 and
removes it from the qpn before being freed so that the lower 16 bit
number can be reused.
This patch requires minor changes in the IB core and ipoib to facilitate
the creation of accelerated UP QPs.
Link: https://lore.kernel.org/r/20200511160607.173205.11757.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Gary Leshner <Gary.S.Leshner@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Gary Leshner [Mon, 11 May 2020 16:06:00 +0000 (12:06 -0400)]
IB/hfi1: Remove module parameter for KDETH qpns
The module parameter for KDETH qpns is being removed in favor
of always using the default value of 0x80 as the qpn prefix.
Defines have been added for various KDETH values including
the prefix of 0x80.
The reserved range now starts at the base value for KDETH
qpns (0x80) and extends up to and including the last qpn for
other reserved QP prefixed types.
Adjust other QP prefixed define names to match KDETH defined
names.
Link: https://lore.kernel.org/r/20200511160600.173205.27508.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Gary Leshner <Gary.S.Leshner@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Gary Leshner [Mon, 11 May 2020 16:05:54 +0000 (12:05 -0400)]
IB/hfi1: Add the transmit side of a datagram ipoib RDMA netdev
This implements the transmit side of the multiple transmit queue RDMA
netdev used to accelerate ipoib. The receive side remains the ipoib
internal implementation.
The init/unint/open/stop netdev operations are saved off and called by the
versions within the hfi1 netdev in order to initialize the connected mode
resources present in ipoib thus allowing us to switch modes between
datagram and connected.
The datagram queue pair instantiated by the ipoib ulp is used by this
implementation for its queue pair number and to register with multicast.
The above queue pair is not used on transmit other than its qpn as the
verbs layer is skipped and packets are directly submitted to the sdma
engines.
Link: https://lore.kernel.org/r/20200511160554.173205.1369.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Gary Leshner <Gary.S.Leshner@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Gary Leshner [Mon, 11 May 2020 16:05:48 +0000 (12:05 -0400)]
IB/hfi1: Add functions to transmit datagram ipoib packets
This patch implements the mechanism to accelerate the transmit side of
a multiple transmit queue RDMA netdev by submitting the packets to
the SDMA engine directly instead of sending through the verbs layer.
This patch also changes the UD/SEND_ONLY op to output the entropy value
in byte 0 of deth[1]. UD/SEND_ONLY_WITH_IMMEDIATE uses the previous
behavior with no entropy value being output.
The code in the ipoib rdma netdev which submits tx requests upon
successful submission will call trace_sdma_output_ibhdr to output
the ibhdr to the trace buffer.
Link: https://lore.kernel.org/r/20200511160548.173205.45616.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Gary Leshner <Gary.S.Leshner@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Kaike Wan [Mon, 11 May 2020 16:05:41 +0000 (12:05 -0400)]
IB/hfi1: Add accelerated IP capability bit
The accelerated IP capability bit is added to allow users to control
which feature is enabled and disabled.
Link: https://lore.kernel.org/r/20200511160541.173205.96870.stgit@awfm-01.aw.intel.com Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> 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: Jason Gunthorpe <jgg@mellanox.com>
Gal Pressman [Tue, 12 May 2020 15:22:04 +0000 (18:22 +0300)]
RDMA/efa: Report host information to the device
The host info feature allows the driver to infrom the EFA device
firmware with system configuration for debugging and troubleshooting
purposes.
The host info buffer is passed as an admin command DMA mapped control
buffer, and is unmapped and freed once the command CQE is consumed.
Currently, the setting of host info is done for each device on its
probe. Failing to set the host info for the device shall not disturb the
probe flow, any errors will be discarded.
Gal Pressman [Tue, 12 May 2020 15:22:03 +0000 (18:22 +0300)]
RDMA/efa: Fix setting of wrong bit in get/set_feature commands
When using a control buffer the ctrl_data bit should be set in order to
indicate the control buffer address is valid, not ctrl_data_indirect
which is used when the control buffer itself is indirect.
Fixes: e9c6c5373088 ("RDMA/efa: Add common command handlers") Link: https://lore.kernel.org/r/20200512152204.93091-2-galpress@amazon.com Reviewed-by: Firas JahJah <firasj@amazon.com> Reviewed-by: Yossi Leybovich <sleybo@amazon.com> Signed-off-by: Gal Pressman <galpress@amazon.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Aharon Landau [Wed, 13 May 2020 09:55:50 +0000 (12:55 +0300)]
RDMA/mlx5: Add init2init as a modify command
Missing INIT2INIT entry in the list of modify commands caused DEVX
applications to be unable to modify_qp for this transition state. Add the
MLX5_CMD_OP_INIT2INIT_QP opcode to the list of allowed DEVX opcodes.
Fixes: e662e14d801b ("IB/mlx5: Add DEVX support for modify and query commands") Link: https://lore.kernel.org/r/20200513095550.211345-1-leon@kernel.org Signed-off-by: Aharon Landau <aharonl@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Lijun Ou [Fri, 8 May 2020 09:45:59 +0000 (17:45 +0800)]
RDMA/hns: Reserve one sge in order to avoid local length error
When rq/srq sge length is smaller than sq sge length, it will produce a
local length error and may cause the bus to hang. Therefore, for rq wqe
and srq wqe, one reserved sge pointing to a reserved mr is used to avoid
this error.
Weihang Li [Fri, 8 May 2020 09:45:57 +0000 (17:45 +0800)]
RDMA/hns: Remove redundant memcpy()
srq_context is a local variables and is only used to get some fields from
buffer of mailbox. It's meaningless to copy mailbox's buffer's contents
back to it.
Lang Cheng [Fri, 8 May 2020 09:45:52 +0000 (17:45 +0800)]
RDMA/hns: Fix cmdq parameter of querying pf timer resource
The firmware has reduced the number of descriptions of command
HNS_ROCE_OPC_QUERY_PF_TIMER_RES to 1. The driver needs to adapt, otherwise
the hardware will report error 4(CMD_NEXT_ERR).
but the preferred mechanism to declare variable-length types such as
these ones is a flexible array member[1][2], introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on. So, replace
the one-element array with a flexible-array member.
Also, make use of the new struct_size() helper to properly calculate the
size of struct siw_pbl.
This issue was found with the help of Coccinelle and, audited and fixed
_manually_.
Danil Kipnis [Tue, 19 May 2020 11:14:19 +0000 (13:14 +0200)]
rnbd/rtrs: Pass max segment size from blk user to the rdma library
When Block Device Layer is disabled, BLK_MAX_SEGMENT_SIZE is undefined.
The rtrs is a transport library and should compile independently of the
block layer. The desired max segment size should be passed down by the
user.
Introduce max_segment_size parameter for the rtrs_clt_open() call.
Fixes: f7a7a5c228d4 ("block/rnbd: client: main functionality") Fixes: 6a98d71daea1 ("RDMA/rtrs: client: main functionality") Fixes: cb80329c9434 ("RDMA/rtrs: client: private header with client structs and functions") Fixes: b5c27cdb094e ("RDMA/rtrs: public interface header to establish RDMA connections") Link: https://lore.kernel.org/r/20200519111419.924170-1-danil.kipnis@cloud.ionos.com Signed-off-by: Danil Kipnis <danil.kipnis@cloud.ionos.com> Reported-by: Randy Dunlap <rdunlap@infradead.org> Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
projects / thirdparty / linux.git / log
