This patch defines various PCI config space access APIs
in order to read and find IOV specific PCI capabilities.
With these definitions implemented, it enables the base
driver to do SR-IOV specific initialization and HW specific
configuration required from PF-PMD driver instance.
Signed-off-by: Manish Chopra <manishc@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: Rasesh Mody <rmody@marvell.com>
On ungraceful termination of DPDK application, PMD VF driver
fails to re-load due to PF seeing the VF in unexpected state
during VF acquisition handshake.
This patch fixes it by allowing VF to request the PF for soft
FLR during the load in such cases so that it can get cleanly
re-loaded.
Fixes: 2ea6f76aff ("qede: add core driver")
Cc: stable@dpdk.org
Signed-off-by: Manish Chopra <manishc@marvell.com>
Signed-off-by: Shahed Shaikh <shshaikh@marvell.com>
Signed-off-by: Rasesh Mody <rmody@marvell.com>
Acked-by: Rasesh Mody <rmody@marvell.com>
This patch updates the FW to 8.40.25.0 and corresponding base driver
changes. It also updates the PMD version to 2.11.0.1. The FW updates
consists of enhancements and fixes as described below.
- VF RX queue start ramrod can get stuck due to completion error.
Return EQ completion with error, when fail to load VF data. Use VF
FID in RX queue start ramrod
- Fix big receive buffer initialization for 100G to address failure
leading to BRB hardware assertion
- GRE tunnel traffic doesn't run when non-L2 ethernet protocol is
enabled, fix FW to not forward tunneled SYN packets to LL2.
- Fix the FW assert that is caused during vport_update when
tx-switching is enabled
- Add initial FW support for VF Representors
- Add ecore_get_hsi_def_val() API to get default HSI values
- Move following from .c to .h files:
TSTORM_QZONE_START and MSTORM_QZONE_START
enum ilt_clients
renamed struct ecore_dma_mem to phys_mem_desc and moved
- Add ecore_cxt_set_cli() and ecore_cxt_set_blk() APIs to set client
config and block details
- Use SET_FIELD() macro where appropriate
- Address spell check and code alignment issues
Signed-off-by: Rasesh Mody <rmody@marvell.com>
Move DMA engine (DMAE) structures from base driver to HSI module.
Use DMAE_PARAMS_* in place of ECORE_DMAE_FLAG_*.
Enforce SET_FIELD() macro where appropriate.
Signed-off-by: Rasesh Mody <rmody@marvell.com>
Some global variables are defined with generic names, add component name
as prefix to variables to prevent collusion with application variables.
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>
Acked-by: Tianfei Zhang <tianfei.zhang@intel.com>
Change details:
- Get engine affinity from the management FW and configure accordingly
- Add an LLH filter with the primary MAC address in QPAR/NPAR
- Move some of the LLH APIs around
- Add PPFID APIs
- Update all allocated ppfids with the same value for the
following PORT_PF registers:
NIG_REG_DSCP_TO_TC_MAP_ENABLE
- Add port_id, src_pfid and dst_pfid to DMA engine params
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
When trust mode is set to ON, VF can change it's MAC address
inspite PF has set a forced MAC for that VF from HV.
Earlier similar functionality is provided by module parameter
"allow_vf_mac_change_mode" of qed.
This change makes few changes in behavior of VF shadow config -
- Let driver track the VF mac in shadow config as long as trust
mode is OFF.
- Once trust mode is ON, we should not care about MACs in shadow
config (because we never intend to fall back because of lack of restore
implementation).
- Delete existing shadow MAC (this helps when trust mode is turned OFF,
and VF tries to add new MAC – it won’t fail that time since we have
a clean slate).
- Skip addition and deletion of MACs in shadow configs.
Signed-off-by: Shahed Shaikh <shahed.shaikh@cavium.com>
We were using LICENSE.qede_pmd to reference inclusion of SPDX licensing
tag from all the source file. Remove the LICENSE.qede_pmd file and
directly include SPDX tags in source files.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
This patch fixes VF MTU update to work without having to restart the
vport and there by not requiring port re-configuration. It adds a
VF MTU Update TLV to achieve the same. Firmware can handle VF MTU update
by just pausing the vport.
Fixes: dd28bc8c6e ("net/qede: fix VF port creation sequence")
Cc: stable@dpdk.org
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
This change aligns the PF behavior with that of the L2-fw, so that flows
(to remove an incorrect MAC) for PF and VF could remain the same.
Fixes: 86a2265e59 ("qede: add SRIOV support")
Cc: stable@dpdk.org
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
IOV regression testing led to discovery of a minor issue + possibly race
in IOV flows:
a. Malicious indications in VF-database on PF-side get cleared during
FLR flows - but not when disabling SRIOV. At least in Linux if you
disable IOV while having a malicious VF you wouldn't be able to
clear the indication as driver would prevent from initializing it.
b. Possible race during PF response to VF - the channel is made ready
only after sending the rc via dmae to VF. It's possible due to
context switch at end of DMAE [when releasing Mutex] that VF would
start running and send another message prior to PF clearing the
channel, making the FW consider that VF to be malicious.
This patch fixes that by
- clearing the indication even if we're only going to disable VF
- resetting the channel to ready before PF copies the rc to the VF, PF
can then continue and send an additional message
Fixes: 47b302d646 ("net/qede/base: add handling of malicious VF")
Fixes: 86a2265e59 ("qede: add SRIOV support")
Cc: stable@dpdk.org
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Changes include:
- adding new log to print Dcbx version
- reformatting log in ecore_dmae_operation_wait() in case of DMA engine
failure
- changing verbosity of some log messages such as:
VFs incorrect behavior should be logged on PF with DP_VERBOSE(), not
DP_NOTICE(). In general keep IOV-related logs at low verbosity.
Log the critical issues from VF perspective, like message to PF
times-out or the PF rejects the VF configuration, as NOTICE than
VERBOSE
- Add a printout of some MCP CPU info in case of no MFW response
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
- Introduce OSAL_IOV_VF_VPORT_STOP to allow VF to carry out required
operations and prevent a potential assert before closing vport
- Add OSAL_DIV_S64() for 64-bit division on 32-bit platforms.
- Add OSAL for transceiver update OSAL_TRANSCEIVER_UPDATE()
- Add OSAL for MFW command preemption OSAL_MFW_CMD_PREEMPT() within the
spinning loops while sending a mailbox command to the MFW
- Implement OSAL_SPIN_LOCK_IRQSAVE macro
- Rename OSAL_NUM_ACTIVE_CPU() to OSAL_NUM_CPUS()
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Introduce 2 new API functions, one for the VF and the other for
PF [per-VF] which allows to decide whether to use HW/SW channel
for PF<->VF communication(a per-VF configuration). A HyperV might
have different VMs with different requirements.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Determines whether VF utilizes doorbells via limited register bar or via
the doorbell bar and return the size of the HW doorbell bar via acquire
response. By doing that limit the VF CIDs to an amount that would make sure
doorbells for all CIDs fall within the bar.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Upgrade QEDE PMD FW to version 8.30.12.0.
The firmware upgrade change details are as:
- Add support for steering by IP and UDP destination port.
- Add source QP field for GSI offload.
- Add UFP support.
- Add support for outer IPv4 TX CSO with unknown tunnel type (in addition
to inner header CSO).
- Support flow ID in accelerated RFS flow.
- Allow Doorbell on empty SPQ and LL2 TX queue (for doorbell recovery).
- Enable PCI Relaxed Ordering for L2 RX data placement.
- Additional enhancements and bug fixes
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Change verbosity of events related to malicious VFs to DP_NOTICE
from DP_INFO. They are not just informational events.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
- Add ECORE_IS_CMT macro (CMT: couple mode teaming) and use that in all the
places where there are checks for number of HW functions per device > 1.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Add base driver API to read per queue coalescing from hardware.
Move ecore_set_rxq|txq_coalesce() declarations to ecore_l2.h.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
This is an attempt to clean up many unused API parameters across the base
code. Most of the changes are related to removing unused p_hwfn or p_ptt
handlers. The warnings are generated using 'unused-parameter' cflags.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
PF checks upon stop-vport from VF whether it's legal, but if it's not it
would STILL send the request to FW, which might cause it to assert.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
For chip variant CHIP_NUM_AH_xxx, MSI-x configuration for VFs is controlled
per-PF [for all of its child VFs] instead of on a per-VF basis. A flag
called "dont_override_vf_msix" is added that allows the caller/client to
specify the mode they want to operate. If dont_override_vf_msix is false as
in the case of VF of CHIP_NUM_AH_xxx, first a check is made as to what is
currently configured number. Management FW will be asked to configure the
requested number only if its bigger than the currently configured value.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
There's a possible race in multiple VF scenarios for base driver users
that use the optional APIs ecore_iov_pf_get_and_clear_pending_events,
ecore_iov_pf_add_pending_events. If the client doesn't synchronize the two
calls, it's possible for the PF to clear a VF pending message indication
without ever getting it [as 'get & clear' isn't atomic], leading to VF
timeout on the command.
The solution is to switch into a per-VF indication rather than having a
bitfield for the various VFs with pending events. As part of the solution,
the setting/clearing of the indications is done internally by base driver.
As a result, ecore_iov_pf_add_pending_events is no longer needed and
ecore_iov_pf_get_and_clear_pending_events loses the 'and_clear' from its
name as its now a proper getter.
A VF would be considered 'pending' [I.e., get_pending_events() should
have '1' for it in its bitfield] beginning with the PF's base driver
recognizing a message sent by that VF [in SP_DPC] and ending only when
that VF message is processed.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Use the ptt[PF translation table] handler that is passed rather than using
main ptt from the HW function.
In ecore_hw_get_resc()'s error flow, release the MFW generic resource lock
only if needed.
Change the verbosity level of GRC timeout from DP_INFO() to DP_NOTICE().
Reduce verbosity of print in ecore_hw_bar_size().
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Status Block reallocation - allow a PF and its child VF to change SB
between them using new base driver APIs.
The changes that are inside base driver flows are:
New APIs ecore_int_igu_reset_cam() and ecore_int_igu_reset_cam_default()
added to reset IGU CAM.
a. During hw_prepare(), driver would re-initialize the IGU CAM.
b. During hw_stop(), driver would initialize the IGU CAM to default.
Use igu_sb_id instead of sb_idx [protocol index] to allow setting of
the timer-resolution in CAU[coalescing algorithm unit] for all SBs,
sb_idx could limit SBs 0-11 only to be able change their timer-resolution.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Allow opening Multiple Tx queues on a single qzone for VFs.
This is supported by Rx/Tx TLVs now having an additional extended TLV that
passes the `qid_usage_idx', a unique number per each queue-cid that was
opened for a given queue-zone.
Fix to overcome TX timeout issue due to more than 16 CIDs by adding an
additional VF legacy mode. This will detach the CIDs from the original
only-existing legacy mode suited for older releases.
Following this change, only VFs that would publish VFPF_ACQUIRE_CAP_QIDS
would have the new CIDs scheme applied. I.e., the new 'legacy' mode is
actually whether this capability is published or not.
Changed the logic to clear doorbells for legacy and non-legacy VFs, so
the PF is cleaning the doorbells for both cases.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Move ecore_crc32() macro to within base driver to qede_crc32() and use
OSAL_CRC32() where required.
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
A step toward having multi-Txq support on same queue-zone for VFs.
This change takes care of:
- VFs assume a single CID per-queue, where queue X receives CID X.
Switch to a model similar to that of PF - I.e., Use different CIDs
for Rx/Tx, and use mapping to acquire/release those. Each VF
currently will have 32 CIDs available for it [for its possible 16
Rx & 16 Tx queues].
- To retain the same interface for PFs/VFs when initializing queues,
the base driver would have to retain a unique number per-each queue
that would be communicated in some extended TLV [current TLV
interface allows the PF to send only the queue-id]. The new TLV isn't
part of the current change but base driver would now start adding
such unique keys internally to queue_cids. This would also force
us to start having alloc/setup/free for L2 [we've refrained from
doing so until now]
The limit would be no-more than 64 queues per qzone [This could be
changed if needed, but hopefully no one needs so many queues]
- In IOV, Add infrastructure for up to 64 qids per-qzone, although
at the moment hard-code '0' for Rx and '1' for Tx [Since VF still
isn't communicating via new TLV which index to associate with a
given queue in its queue-zone].
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
L2 handler changes:
This is change to remove the queue-id/qzone difference for Tx queues.
It does that by mainly doing:
a. VFs queues are no longer determined by the SBs they're using.
Instead, the ecore-client needs to maintain those and choose the values
to be used by VF when initializing it.
b. Eliminate the HW-cid array in the hw-function.
To do that, have all the rx/tx functionality turn into 'handle' base -
when queue would be started the caller would get a (void*) handle,
which it would later use with ecore for configuring various
queue-related stop [update, close].
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
