Patrick Fu [Tue, 7 Jul 2020 05:07:08 +0000 (13:07 +0800)]
vhost: introduce async enqueue registration API
Performing large memory copies usually takes up a major part of CPU
cycles and becomes the hot spot in vhost-user enqueue operation. To
offload the large copies from CPU to the DMA devices, asynchronous
APIs are introduced, with which the CPU just submits copy jobs to
the DMA but without waiting for its copy completion. Thus, there is
no CPU intervention during data transfer. We can save precious CPU
cycles and improve the overall throughput for vhost-user based
applications. This patch introduces registration/un-registration
APIs for vhost async data enqueue operation. Together with the
registration APIs implementations, data structures and the prototype
of the async callback functions required for async enqueue data path
are also defined.
Signed-off-by: Patrick Fu <patrick.fu@intel.com> Reviewed-by: Chenbo Xia <chenbo.xia@intel.com>
Jeff Guo [Tue, 7 Jul 2020 05:25:45 +0000 (13:25 +0800)]
net/ice: enable additional input set for RSS hash
PF could add or delete a RSS rule base on the PF's hash capability.
Some new rss input set will be supported, the protocols as below:
eth/vlan/l2tpv3/esp/ah/pfcp.
Signed-off-by: Jeff Guo <jia.guo@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Jeff Guo [Tue, 7 Jul 2020 05:14:06 +0000 (13:14 +0800)]
net/iavf: enable additional hash flow
Some new hash flow will be supported to expend the flow hash
capability, the input set are the session id for NAT-T ESP protocol,
the l3 src/dst and the teid for GTPU_IP protocol.
Signed-off-by: Jeff Guo <jia.guo@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Jeff Guo [Tue, 7 Jul 2020 05:14:05 +0000 (13:14 +0800)]
net/iavf: enable 5 tuple RSS hash
Previous iavf not support 5 tuple hash, this patch aims to enable it
for regular ip pattern and also GTPU inner ip pattern, the 5 tuple
involves ip src and ip dst, tcp sport and tcp dport, udp sport and
udp dport and protocol id.
Signed-off-by: Jeff Guo <jia.guo@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Jeff Guo [Tue, 7 Jul 2020 05:14:04 +0000 (13:14 +0800)]
net/iavf: refactor hash flow
Refactor hash flow by change the process of the pattern parser and the
action parser, and refine the lookup table for regular IP and GTPU_EH,
ETH, and also VLAN.
Signed-off-by: Jeff Guo <jia.guo@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Guinan Sun [Mon, 6 Jul 2020 08:12:14 +0000 (08:12 +0000)]
net/e1000/base: modify flow control setup
Customers had a problem with large pings after connected standby.
This is due to the requirement of maintaining link after CS - the driver
blocks resets during "AdapterStart" and skips flow control setup.
Added condition in e1000_setup_link_ich8lan.c function that always setup
flow control, and setup physical interface only when no need to block
resets.
Guinan Sun [Mon, 6 Jul 2020 08:12:12 +0000 (08:12 +0000)]
net/e1000/base: remove useless statement
As it stands now, this fix will not change the current functionality
of the code. In addition, we remove the comment that seemed to be
a copy/paste from a separate implementation.
Signed-off-by: Jeb Cramer <jeb.j.cramer@intel.com> Signed-off-by: Guinan Sun <guinanx.sun@intel.com> Reviewed-by: Wei Zhao <wei.zhao1@intel.com>
Guinan Sun [Mon, 6 Jul 2020 08:12:09 +0000 (08:12 +0000)]
net/e1000/base: increase timeout for ME ULP exit
Due timing issues in WHL and since recovery by host is
not always supported, increased timeout for Manageability Engine(ME)
to finish Ultra Low Power(ULP) exit flow for Nahum before timer expiration.
Guinan Sun [Mon, 6 Jul 2020 08:12:08 +0000 (08:12 +0000)]
net/e1000/base: add missing register defines
Added defines for the EEC, SHADOWINF and FLFWUPDATE registers needed for
the nvmupd_validate_offset function to correctly validate the NVM update
offset.
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Guinan Sun <guinanx.sun@intel.com> Reviewed-by: Wei Zhao <wei.zhao1@intel.com>
Guinan Sun [Mon, 6 Jul 2020 08:12:05 +0000 (08:12 +0000)]
net/e1000/base: remove duplicated codes
Add two files base.c and base.h to reduce the redundancy
in the silicon family code.
Remove the code duplication from e1000_82575 files.
Clean family specific functions from base.
Fix up a stray and duplicate function declaration.
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Sasha Neftin <sasha.neftin@intel.com> Signed-off-by: Guinan Sun <guinanx.sun@intel.com> Reviewed-by: Wei Zhao <wei.zhao1@intel.com>
Guinan Sun [Mon, 6 Jul 2020 08:12:03 +0000 (08:12 +0000)]
net/e1000/base: fix minor issues and improve code style
Fix typo in piece of code of NVM access for SPT.
And cleans up the remaining instances in the shared code
where it was not adhering to the Linux code standard.
Wrong description was found in the mentioned file, so fix them.
Remove shadowing variable declarations.
Relating to operands in bitwise operations having different sizes.
Unreachable code since *clock_in_i2c_* always return success.
Don't return unused s32 and don't check for constants.
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Sasha Neftin <sasha.neftin@intel.com> Signed-off-by: Vitaly Lifshits <vitaly.lifshits@intel.com> Signed-off-by: Robert Konklewski <robertx.konklewski@intel.com> Signed-off-by: Doug Dziggel <douglas.a.dziggel@intel.com> Signed-off-by: Todd Fujinaka <todd.fujinaka@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Signed-off-by: Guinan Sun <guinanx.sun@intel.com> Reviewed-by: Wei Zhao <wei.zhao1@intel.com>
Simei Su [Tue, 7 Jul 2020 01:38:18 +0000 (09:38 +0800)]
net/ice: fix GTPU/PPPoE packets with no hash value
When RSS init, because of profile overlap, the GTPU_IPV4 packets
don't hit GTPU_INNER_IPV4 profile which causes no hash value. Because
of no PPPoE profile, the PPPoE packets also has no hash value. This
patch solves this issue by pulling GTPU_IPV4 profile into inner ipv4
group and creating related PPPoE profile at the same time.
Fixes: 4717a12cfaf1 ("net/ice: initialize and update RSS based on user config") Signed-off-by: Simei Su <simei.su@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Shougang Wang [Mon, 6 Jul 2020 09:37:05 +0000 (09:37 +0000)]
net/ice: fix error log in generic flow
When create a RSS rule with void action, the error log is "Invalid
input set". This patch fix the issue by adding check for the type of
first actions item.
Fixes: 7615a6895009 ("net/ice: rework for generic flow enabling") Cc: stable@dpdk.org Signed-off-by: Shougang Wang <shougangx.wang@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Simei Su [Thu, 2 Jul 2020 05:11:48 +0000 (13:11 +0800)]
ethdev: add PPPoE RSS offload types
This patch defines new RSS offload types for PPPoE. Typically,
session id would be the RSS input set for a PPPoE packet, but
as a hint, each driver may have different default behaviors.
The Flow counter manager allocates memory to hold the software view
of the counters where the on-chip counter data will be accumulated
along with another memory block that will be shadowing the on-chip
counter data i.e where the raw counter data will be DMAed into from
the chip.
It also keeps track of the first HW counter ID as that will be needed
to retrieve the counter data in bulk using a TF API. It issues this cmd
in an rte_alarm thread that keeps running every second.
Currently, only vfrep transmit requires cfa_action to be filled
in the tx buffer descriptor. However with truflow, dpdk(non vfrep)
to port also requires cfa_action to be filled in the tx buffer
descriptor.
This patch uses the correct cfa_action pointer while transmitting
the packet. Based on whether the packet is transmitted on non-vfrep
or vfrep, tx_cfa_action or vfr_tx_cfa_action inside txq will be
filled in the tx buffer descriptor.
net/bnxt: create default flow rules for port reprentor
Invoked 3 new APIs for the default flow create/destroy and to get
the action ptr for a default flow.
Changed ulp_intf_update() to accept rte_eth_dev as input and invoke
the same from the VF rep start function.
ULP Mark Manager will indicate if the cfa_code returned in the
Rx completion descriptor was for one of the default flow rules
created for the VF representor conduit. The mark_id returned
in such a case would be the VF rep's DPDK Port id, which can be
used to get the corresponding rte_eth_dev struct in bnxt_vf_recv
net/bnxt: fill mapper parameters with default rules
Default rules are needed for the packets to be punted between the
following entities in the non-offloaded path
1. Device PORT to DPDK App
2. DPDK App to Device PORT
3. VF Representor to VF
4. VF to VF Representor
This patch fills all the relevant information in the computed fields
& the act_prop fields for the flow mapper to create the necessary
tables in the hardware to enable the default rules.
When VF-VFR conduits are created, a mark is added to the mark database.
mark_flag indicates whether the mark is valid and has VFR information
(VFR_ID bit in mark_flag). Rx path was checking for this VFR_ID bit.
However, while adding the mark to the mark database, VFR_ID bit is not
set in mark_flag.
1. Add "enum bnxt_ulp_intf_type” as the second parameter for the
port & func helper functions
2. Return vfrep related port & func information in the helper functions
3. Allocate phy_port_list dynamically based on port count
4. Introduce ulp_func_id_tbl array for book keeping func related
information indexed by func_id
net/bnxt: add index opcode and operand to mapper table
Extended the regfile and computed field operations to a common
index opcode operation and added globlal resource operations are
also part of the index opcode operation.
Mike Baucom [Thu, 2 Jul 2020 23:28:20 +0000 (16:28 -0700)]
net/bnxt: support internal encap records
Modifications to allow internal encap records to be supported:
- Modified the mapper index table processing to handle encap without an
action record
- Modified the session open code to reserve some 64 Byte internal encap
records on tx
- Modified the blob encap swap to support encap without action record
- Select EEM Host or System memory via config parameter
- Add EEM system memory support for kernel memory
- Dependent on DPDK changes that add support for the HWRM_OEM_CMD.
Signed-off-by: Peter Spreadborough <peter.spreadborough@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Jay Ding [Thu, 2 Jul 2020 23:28:17 +0000 (16:28 -0700)]
net/bnxt: add global config set and get functions
- Add support to update global configuration for ACT_TECT
and ACT_ABCR.
- Add support to allow Tunnel and Action global configuration.
- Remove register read and write operations.
- Remove the register read and write support.
Signed-off-by: Jay Ding <jay.ding@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
- Remove table scope data from session. Added to EEM.
- Complete move to RM of table scope base and range.
- Fix some err messaging strings.
- Fix the tcam logging message.
Signed-off-by: Peter Spreadborough <peter.spreadborough@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
- For the EM Module there is a need to only allocate the EM Records in
HCAPI RM but the storage control is requested to be outside of the RM
DB.
- Add TF_RM_ELEM_CFG_HCAPI_BA.
- Return error when the number of reserved entries for wc tcam is odd
number in tf_tcam_bind.
- Remove em_pool from session
- Use RM provided start offset and size
- HCAPI returns entry index instead of row index for WC TCAM.
- Move resource type conversion to hrwm set/free tcam functions.
Signed-off-by: Peter Spreadborough <peter.spreadborough@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Michael Wildt [Thu, 2 Jul 2020 23:28:10 +0000 (16:28 -0700)]
net/bnxt: update table get to use new design
- Move bulk table get implementation to new Tbl Module design.
- Update messages for bulk table get
- Retrieve specified table element using bulk mechanism
- Remove deprecated resource definitions
- Update device type configuration for P4.
- Update RM DB HCAPI count check and fix EM internal and host
code such that EM DBs can be created correctly.
- Update error logging to be info on unbind in the different modules.
- Move RTE RSVD out of tf_resources.h
Signed-off-by: Michael Wildt <michael.wildt@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
net/bnxt: support EM and TCAM lookup with table scope
- Support for table scope within the EM module
- Support for host and system memory
- Update TCAM set/free.
- Replace TF device type by HCAPI RM type.
- Update TCAM set and free for HCAPI RM type
Signed-off-by: Peter Spreadborough <peter.spreadborough@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Jay Ding [Thu, 2 Jul 2020 23:28:08 +0000 (16:28 -0700)]
net/bnxt: support two-level priority for TCAMs
Allow TCAM indexes to be allocated from top or bottom.
If the priority is set to 0, allocate from the
lowest tcam indexes i.e. from top. Any other value,
allocate it from the highest tcam indexes i.e. from
bottom.
Signed-off-by: Jay Ding <jay.ding@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Michael Wildt [Thu, 2 Jul 2020 23:28:07 +0000 (16:28 -0700)]
net/bnxt: update RM with residual checker
- Add residual checker to the TF Host RM as well as new RM APIs. On
close it will scan the DB and check of any remaining elements. If
found they will be logged and FW msg sent for FW to scrub that
specific type of resources.
- Update the module bind to be aware of the module type, for each of
the modules.
- Add additional type 2 string util functions.
- Fix the device naming to be in compliance with TF.
- Update the device unbind order as to assure TCAMs gets flushed
first.
- Update the close functionality such that the session gets
closed after the device is unbound.
Signed-off-by: Michael Wildt <michael.wildt@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Michael Wildt [Thu, 2 Jul 2020 23:28:05 +0000 (16:28 -0700)]
net/bnxt: support multiple device
Implement the Identifier, Table Type and the Resource Manager
modules.
Integrate Resource Manager with HCAPI.
Update open/close session.
Move to direct msgs for qcaps and resv messages.
Signed-off-by: Michael Wildt <michael.wildt@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Allow TCAM indexes to be allocated from top or bottom.
If the priority is set to 0, allocate from the
lowest tcam indexes i.e. from top. Any other value,
allocate it from the highest tcam indexes i.e. from
bottom.
Michael Wildt [Thu, 2 Jul 2020 23:28:00 +0000 (16:28 -0700)]
net/bnxt: update multi device design
- Implement the modules RM, Device (WH+), Identifier.
- Update Session module.
- Implement new HWRMs for RM direct messaging.
- Add new parameter check macro's and clean up the header includes for
i.e. tfp such that bnxt.h is not directly included in the new modules.
- Add cfa_resource_types, required for RM design.
Signed-off-by: Michael Wildt <michael.wildt@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
- Add new bulk table type get using FW
to DMA the data back to host.
- Add flag to allow records to be cleared if possible
- Set mirror using tf_alloc_tbl_entry
- Add Exact Match support
- Create EM table pool of memory indices
- Insert exact match internal entry API
- Sends EM internal insert and delete request to firmware
Signed-off-by: Peter Spreadborough <peter.spreadborough@broadcom.com> Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com> Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Apart from func_svif, func_id & vnic, port_db now stores and
retrieves func_spif, func_parif, phy_port_id, port_svif, port_spif,
port_parif, port_vport. New helper functions have been added to
support the same.
Use 'first_vf_id' and the 'vf_id' that is input as part of adding
a representor to obtain the PCI function ID(FID) of the VF(VFR endpoint).
Use the FID as an input to FUNC_QCFG HWRM cmd to obtain the default
vnic ID of the VF.
Along with getting the default vNIC ID by supplying the FW FID of
the VF-rep endpoint to HWRM_FUNC_QCFG, obtain and store it's
function svif.
Added code to support Tx/Rx from a VF representor port.
The VF-reps use the RX/TX rings of the Trusted VF/PF.
For each VF-rep, the Trusted VF/PF driver issues a VFR_ALLOC FW cmd that
returns "cfa_code" and "cfa_action" values.
The FW sets up the filter tables in such a way that VF traffic by
default (in absence of other rules) gets punted to the parent function
i.e. either the Trusted VF or the PF.
The cfa_code value in the RX-compl informs the driver of the source VF.
For traffic being transmitted from the VF-rep, the TX BD is tagged with
a cfa_action value that informs the HW to punt it to the corresponding
VF.
Defines data structures and code to init/uninit
VF representors during pci_probe and pci_remove
respectively.
Most of the dev_ops for the VF representor are just
stubs for now and will be will be filled out in next patch.
To create a representor using testpmd:
testpmd -c 0xff -wB:D.F,representor=1 -- -i
testpmd -c 0xff -w05:02.0,representor=[1] -- -i
To create a representor using ovs-dpdk:
1. First add the trusted VF port to a bridge
ovs-vsctl add-port ovsbr0 vf_rep1 -- set Interface vf_rep1 type=dpdk
options:dpdk-devargs=0000:06:02.0
2. Add the representor port to the bridge
ovs-vsctl add-port ovsbr0 vf_rep1 -- set Interface vf_rep1 type=dpdk
options:dpdk-devargs=0000:06:02.0,representor=1
Shy Shyman [Sun, 28 Jun 2020 09:36:31 +0000 (12:36 +0300)]
net/mlx5: fix flow META item validation
When flow is inserted with meta match item it requires a certain
register support.
As part of the flow validation of such flows, the validation
function is missing a check that the mlx5 driver is not in
legacy mode in terms of extended meta data support
(MLX5_XMETA_MODE_LEGACY flag).
If the driver is in legacy mode it will cause downstream
function that allocates needed register for meta data.
The fix checks explicitly the conditions for support of
meta data in FDB mode. If the conditions are not met
an error message will be issued.
This patch add more specific tunnel type for ipv4/ipv6 packet,
it enable tcp/udp layer of ipv4/ipv6 as L4 payload but without
L4 dst/src port number as input set for the switch filter rule.
git.droids-corp.org / dpdk.git / log