Mellanox's SmartNIC combines embedded CPU(e.g, ARM) processing power
with advanced network offloads to accelerate a multitude of security,
networking and storage applications.
With the introduction of the SmartNIC, there is a new PCI function
called Embedded CPU Physical Function(ECPF). And it's possible for a
PF to get its ICM pages from the ECPF PCI function. Driver shall
identify if it is running on such a function by reading a bit in
the initialization segment.
When firmware asks for pages, it would issue a page request event
specifying how many pages it requests and for which function. That
driver responds with a manage_pages command providing the requested
pages along with an indication for which function it is providing these
pages.
The encoding before this patch was as follows:
function_id == 0: pages are requested for the function receiving
the EQE.
function_id != 0: pages are requested for VF identified by the
function_id value
A new one bit field in the EQE identifies that pages are requested for
the ECPF.
The notion of page_supplier can be introduced here and to support that,
manage pages and query pages were modified so firmware can distinguish
the following cases:
1. Function provides pages for itself
2. PF provides pages for its VF
3. ECPF provides pages to itself
4. ECPF provides pages for another function
This distinction is possible through the introduction of the bit
"embedded_cpu_function" in query_pages, manage_pages and page request
EQE.
Signed-off-by: Bodong Wang <bodong@mellanox.com>
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Use u16 for vport number, which matches how hardware refers to this
argument throughout commands.
This patch doesn't change any functionality.
Signed-off-by: Bodong Wang <bodong@mellanox.com>
Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
From
git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux
For dependencies needed in the next ODP patches.
* branch 'mlx5-next':
net/mlx5: Set ODP SRQ support in firmware
net/mlx5: Add XRC transport to ODP device capabilities layout
To avoid compatibility issue with older kernels the firmware doesn't
allow SRQ to work with ODP unless kernel asks for it.
Signed-off-by: Moni Shoua <monis@mellanox.com>
Reviewed-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
The device capabilities for ODP structure was missing the field for XRC
transport so add it here.
Signed-off-by: Moni Shoua <monis@mellanox.com>
Reviewed-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
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>
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>
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>
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>
The implementation of mlx5_core_page_fault_resume() was removed in commit
d5d284b829 ("{net,IB}/mlx5: Move Page fault EQ and ODP logic to
RDMA"). This patch removes declaration too.
Fixes: d5d284b829 ("{net,IB}/mlx5: Move Page fault EQ and ODP logic to RDMA")
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Pull rdma updates from Jason Gunthorpe:
"This has been a fairly typical cycle, with the usual sorts of driver
updates. Several series continue to come through which improve and
modernize various parts of the core code, and we finally are starting
to get the uAPI command interface cleaned up.
- Various driver fixes for bnxt_re, cxgb3/4, hfi1, hns, i40iw, mlx4,
mlx5, qib, rxe, usnic
- Rework the entire syscall flow for uverbs to be able to run over
ioctl(). Finally getting past the historic bad choice to use
write() for command execution
- More functional coverage with the mlx5 'devx' user API
- Start of the HFI1 series for 'TID RDMA'
- SRQ support in the hns driver
- Support for new IBTA defined 2x lane widths
- A big series to consolidate all the driver function pointers into a
big struct and have drivers provide a 'static const' version of the
struct instead of open coding initialization
- New 'advise_mr' uAPI to control device caching/loading of page
tables
- Support for inline data in SRPT
- Modernize how umad uses the driver core and creates cdev's and
sysfs files
- First steps toward removing 'uobject' from the view of the drivers"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (193 commits)
RDMA/srpt: Use kmem_cache_free() instead of kfree()
RDMA/mlx5: Signedness bug in UVERBS_HANDLER()
IB/uverbs: Signedness bug in UVERBS_HANDLER()
IB/mlx5: Allocate the per-port Q counter shared when DEVX is supported
IB/umad: Start using dev_groups of class
IB/umad: Use class_groups and let core create class file
IB/umad: Refactor code to use cdev_device_add()
IB/umad: Avoid destroying device while it is accessed
IB/umad: Simplify and avoid dynamic allocation of class
IB/mlx5: Fix wrong error unwind
IB/mlx4: Remove set but not used variable 'pd'
RDMA/iwcm: Don't copy past the end of dev_name() string
IB/mlx5: Fix long EEH recover time with NVMe offloads
IB/mlx5: Simplify netdev unbinding
IB/core: Move query port to ioctl
RDMA/nldev: Expose port_cap_flags2
IB/core: uverbs copy to struct or zero helper
IB/rxe: Reuse code which sets port state
IB/rxe: Make counters thread safe
IB/mlx5: Use the correct commands for UMEM and UCTX allocation
...
Add support for the HW feature of multi-packet WQE in XDP
xmit flow.
The conventional TX descriptor (WQE, Work Queue Element) serves
a single packet. Our HW has support for multi-packet WQE (MPWQE)
in which a single descriptor serves multiple TX packets.
This reduces both the PCI overhead and the CPU cycles wasted on
writing them.
In this patch we add support for the HW feature, which is supported
starting from ConnectX-5.
Performance:
Tested packet rate for UDP 64Byte multi-stream over ConnectX-5 NICs.
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz
XDP_TX:
We see a huge gain on single port ConnectX-5, and reach the 100 Mpps
milestone.
* Single-port HCA:
Before: 70 Mpps
After: 100 Mpps (+42.8%)
* Dual-port HCA:
Before: 51.7 Mpps
After: 57.3 Mpps (+10.8%)
* In both cases we tested traffic on one port and for now On Dual-port HCAs
we see only small gain, we are working to overcome this bottleneck, but
for the moment only with experimental firmware on dual port HCAs we can
reach the wanted numbers as seen on Single-port HCAs.
XDP_REDIRECT:
Redirect from (A) ConnectX-5 to (B) ConnectX-5.
Due to a setup limitation, (A) and (B) are on different NUMA nodes,
so absolute performance numbers are not optimal.
Note:
Below is the transmit rate of (B), not the redirect rate of (A)
which is in some cases higher.
* (B) is single-port:
Before: 77 Mpps
After: 90 Mpps (+16.8%)
* (B) is dual-port:
Before: 61 Mpps
After: 72 Mpps (+18%)
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
During testing the command format was changed to close a security
hole. Revise the driver to use the command format that will actually be
supported in GA firmware.
Both the UMEM and UCTX are intended only for use by the kernel and cannot
be executed using a general command.
Since the UMEM and CTX are not part of the general object the caps bits
were moved to be some log_xxx location in the general HCA caps.
The firmware code was adapted as well to match the above.
Fixes: a8b92ca1b0 ("IB/mlx5: Introduce DEVX")
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
From git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux
mlx5 updates taken for dependencies on following patches.
* branche 'mlx5-next': (23 commits)
IB/mlx5: Introduce uid as part of alloc/dealloc transport domain
net/mlx5: Add shared Q counter bits
net/mlx5: Continue driver initialization despite debugfs failure
net/mlx5: Fold the modify lag code into function
net/mlx5: Add lag affinity info to log
net/mlx5: Split the activate lag function into two routines
net/mlx5: E-Switch, Introduce flow counter affinity
IB/mlx5: Unify e-switch representors load approach between uplink and VFs
net/mlx5: Use lowercase 'X' for hex values
net/mlx5: Remove duplicated include from eswitch.c
net/mlx5: Remove the get protocol device interface entry
net/mlx5: Support extended destination format in flow steering command
net/mlx5: E-Switch, Change vhca id valid bool field to bit flag
net/mlx5: Introduce extended destination fields
net/mlx5: Revise gre and nvgre key formats
net/mlx5: Add monitor commands layout and event data
net/mlx5: Add support for plugged-disabled cable status in PME
net/mlx5: Add support for PCIe power slot exceeded error in PME
net/mlx5: Rework handling of port module events
net/mlx5: Move flow counters data structures from flow steering header
...
Lots of conflicts, by happily all cases of overlapping
changes, parallel adds, things of that nature.
Thanks to Stephen Rothwell, Saeed Mahameed, and others
for their guidance in these resolutions.
Signed-off-by: David S. Miller <davem@davemloft.net>
Introduce uid as part of alloc/dealloc transport domain to match the
device specification.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Updated HW specification file with needed bits to allow
sharing of Q counters between DEVX contexts and kernel.
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
With the introduction of SR-IOV LAG, checking whether LAG is active
is no longer good enough, since RoCE and SR-IOV LAG each entails
different behavior by both the core and infiniband drivers.
This patch introduces facilities to discern LAG type, in addition to
mlx5_lag_is_active(). These are implemented in such a way as to allow
more complex mode combinations in the future.
Signed-off-by: Aviv Heller <avivh@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
This introduces devcom, a generic mechanism for performing operations
on both physical functions of the same Connect-X card.
The first user of this API is merged eswitch, which will be introduced
in subsequent patches.
Signed-off-by: Aviv Heller <avivh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
mlx5-next shared branch with rdma subtree to avoid mlx5 rdma v.s. netdev
conflicts.
Highlights:
1) Lag refactroing and flow counter affinity bits.
2) mlx5 core cleanups
By Roi Dayan (2) and others
* 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux:
net/mlx5: Fold the modify lag code into function
net/mlx5: Add lag affinity info to log
net/mlx5: Split the activate lag function into two routines
net/mlx5: E-Switch, Introduce flow counter affinity
IB/mlx5: Unify e-switch representors load approach between uplink and VFs
net/mlx5: Use lowercase 'X' for hex values
net/mlx5: Remove duplicated include from eswitch.c
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
This dictates the device affinity for eswitch flow counters, set by the FW
according to the HW device capabilities.
Under "source eswitch" affinity, the counter should be allocated on the
device related to the source vport in the match. This covers both non
merged e-switch mode as well as old FW that does not advertise this cap.
Under "flow eswitch" affinity, the counter should be allocated on the
device where the eswitch rule is set.
Signed-off-by: Shahar Klein <shahark@mellanox.com>
Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Apparently gcc is cool with upper case '0X' but it is not commonly used.
Replace '0X' with lowercase '0x' in mlx5_ifc.h file.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The cap bits locations for the fdb caps of multi path to table (used for
local mirroring) and multi encap (used for prio/chains) were wrongly used
in swapped locations. This went unnoted so far b/c we tested the offending
patch with CX5 FW that supports both of them. On different environments where
not both caps are supported, we will be messed up, fix that.
Fixes: b9aa0ba17a ('net/mlx5: Add cap bits for multi fdb encap')
Signed-off-by: Vu Pham <vu@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Tested-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
PPCNT is not supported if PCAM access reg is supported and ppcnt bit is clear.
Signed-off-by: Eyal Davidovich <eyald@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Writing 64B CQEs to 128B cache lines results in a RMW operation. Padding
the CQEs to 128B if possible improves performance on 128B cache line
systems like PPC.
Testing on PPC showed up to a 24% improvement in small packet throughput
vs the default behavior, depending on the workload and system topology.
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
CapabilityMask2 exists when IB_PORT_CAP_MASK2_SUP is set in the original
capability mask. In such cases, query its value and report it in query
port.
Signed-off-by: Michael Guralnik <michaelgur@mellanox.com>
Reviewed-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Update the flow steering command formatting according to the extended
destination API.
Note that the FW dictates that multi destination FTEs that involve at
least one encap must use the extended destination format, while single
destination ones must use the legacy format.
Using extended destination format requires FW support. Check for its
capabilities and return error if not supported.
Signed-off-by: Eli Britstein <elibr@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Oz Shlomo <ozsh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Change the driver flow destination struct to use bit flags with the vhca
id valid being the 1st one. The flags field is more extendable and will
be used in downstream patch.
Signed-off-by: Eli Britstein <elibr@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Oz Shlomo <ozsh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Extended destinations provide the ability to configure different
encapsulation properties per destination on a single FTE. This is
needed for use-cases such as remote mirroring over tunneled networks.
Signed-off-by: Eli Britstein <elibr@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Oz Shlomo <ozsh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
GRE RFC defines a 32 bit key field. NVGRE RFC splits the 32 bit
key field to 24 bit VSID (gre_key_h) and 8 bit flow entropy (gre_key_l).
Define the two key parsing alternatives in a union, thus enabling both
access methods.
Signed-off-by: Oz Shlomo <ozsh@mellanox.com>
Reviewed-by: Eli Britstein <elibr@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Will be used in downstream patch to monitor counter changes
by the HCA and report it to the driver by an event.
The driver will update its counters cached data accordingly.
Signed-off-by: Eyal Davidovich <eyald@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Introduce and use a helper that extracts the opcode
from a CQE (completion queue entry) structure.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Danit Goldberg says:
Packet based credit mode
Packet based credit mode is an alternative end-to-end credit mode for QPs
set during their creation. Credits are transported from the responder to
the requester to optimize the use of its receive resources. In
packet-based credit mode, credits are issued on a per packet basis.
The advantage of this feature comes while sending large RDMA messages
through switches that are short in memory.
The first commit exposes QP creation flag and the HCA capability. The
second commit adds support for a new DV QP creation flag. The last commit
report packet based credit mode capability via the MLX5DV device
capabilities.
* branch 'mlx5-packet-credit-fc':
IB/mlx5: Report packet based credit mode device capability
IB/mlx5: Add packet based credit mode support
net/mlx5: Expose packet based credit mode
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Packet based credit mode bit determines whether the credit mode
is done per message or packet. Expose the QP creation flag and
the HCA capability.
Signed-off-by: Danit Goldberg <danitg@mellanox.com>
Reviewed-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Update the supported DEVX commands, it includes adding to the
query/modify command's list and to the encoding handling.
In addition, a valid range for general commands was added to be used for
future commands.
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: Doug Ledford <dledford@redhat.com>
Expose device capabilities for DEVX user context, it includes which caps
the device is supported and a matching bit to set as part of user
context creation.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Reviewed-by: Artemy Kovalyov <artemyko@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Transfer initialization and cleanup from mlx5_priv struct of
mlx5_core_dev to be part of mlx5_ib_dev. This completes removal
of SRQ from mlx5_core.
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
There is no need to keep SRQ which is RDMA object in mlx5_core.
In this patch, we partially move the execution code, while next patches
will move table initialization/release logic too.
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Delete functions which are not called and not needed.
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Ensure that both RDMA and netdev parts of SRQ implementation
has same copyright and license information annotated by SPDX
tags.
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Allow forwarding of SRQ events to mlx5_core interfaces, e.g. mlx5_ib.
Use mlx5_notifier_register/unregister in srq.c in order to allow seamless
transition of srq.c to infiniband subsystem.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Allow forwarding QP and WQ events to mlx5_core interfaces, e.g. mlx5_ib
Use mlx5_notifier_register/unregister in qp.c in order to allow seamless
transition of qp.c to infiniband subsystem.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Before the new mlx5 event notification infrastructure and API,
mlx5_core used to process all events before forwarding them to mlx5
interfaces (mlx5e/mlx5_ib) and used to translate the event type enum
to a software defined enum, this is not needed anymore since it is ok
for mlx5e and mlx5_ib to receive FW events as is, at least the few ones
mlx5 core allows.
mlx5e and mlx5_ib already moved to use the new API and they only handle FW
events types, it is now safe to remove all equivalent software defined
events and the logic around them.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
The mlx5_interface->event callback is not used by mlx5e/mlx5_ib anymore.
We totally remove the delayed events logic work around, since with
the dynamic notifier registration API it is not needed anymore, mlx5_ib
can register its notifier and start receiving events exactly at the moment
it is ready to handle them.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
This to allow seamless migration to the new notifier chain API, and to
eventually deprecate interfaces dev->event callback.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Use atomic notifier chain to fire events to mlx5 core driver
consumers (mlx5e/mlx5_ib) and provide mlx5 register/unregister notifier
API.
This API will replace the current mlx5_interface->event callback and all
the logic around it, especially the delayed events logic introduced by
commit 97834eba7c ("net/mlx5: Delay events till ib registration ends")
Which is not needed anymore with this new API where the mlx5 interface
can dynamically register/unregister its notifier.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Move all the generic async events handling into new specific events
handling file events.c to keep eq.c file clean from concrete event logic
handling.
Use new API to register for NOTIFY_ANY to handle generic events and
dispatch allowed events to mlx5_core consumers (mlx5_ib and mlx5e)
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Remove the explicit call to QP/SRQ resources events handlers on several FW
events and let resources logic register resources events notifiers via the
new API.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
