src/sys/dev/ixl, branch release/12.3.0

iflib: ensure that tx interrupts enabled and cleanups

2021-10-03T02:03:11Z

Doing a 'dd' over iscsi will reliably cause stalls. Tx cleaning _should_ reliably happen as data is sent. However, currently if the transmit queue fills it will wait until the iflib timer (hz/2) runs. This change causes the the tx taskq thread to be run if there are completed descriptors. While here: - make timer interrupt delay a sysctl - simplify txd_db_check handling - comment on INTR types Background on the change: Initially doorbell updates were minimized by only writing to the register on every fourth packet. If txq_drain would return without writing to the doorbell it scheduled a callout on the next tick to do the doorbell write to ensure that the write otherwise happened "soon". At that time a sysctl was added for users to avoid the potential added latency by simply writing to the doorbell register on every packet. This worked perfectly well for e1000 and ixgbe ... and appeared to work well on ixl. However, as it turned out there was a race to this approach that would lockup the ixl MAC. It was possible for a lower producer index to be written after a higher one. On e1000 and ixgbe this was harmless - on ixl it was fatal. My initial response was to add a lock around doorbell writes - fixing the problem but adding an unacceptable amount of lock contention. The next iteration was to use transmit interrupts to drive delayed doorbell writes. If there were no packets in the queue all doorbell writes would be immediate as the queue started to fill up we could delay doorbell writes further and further. At the start of drain if we've cleaned any packets we know we've moved the state machine along and we write the doorbell (an obvious missing optimization was to skip that doorbell write if db_pending is zero). This change required that tx interrupts be scheduled periodically as opposed to just when the hardware txq was full. However, that just leads to our next problem. Initially dedicated msix vectors were used for both tx and rx. However, it was often possible to use up all available vectors before we set up all the queues we wanted. By having rx and tx share a vector for a given queue we could halve the number of vectors used by a given configuration. The problem here is that with this change only e1000 passed the necessary value to have the fast interrupt drive tx when appropriate. Reported by: mav@ Tested by: mav@ Reviewed by: gallatin@ MFC after: 1 month Sponsored by: iXsystems Differential Revision: https://reviews.freebsd.org/D27683 (cherry picked from commit 81be655266fac2b333e25f386f32c9bcd17f523d)

iflib: Make if_shared_ctx_t a pointer to const

2021-10-03T01:55:49Z

This structure is shared among multiple instances of a driver, so we should ensure that it doesn't somehow get treated as if there's a separate instance per interface. This is especially important for software-only drivers like wg. DEVICE_REGISTER() still returns a void * and so the per-driver sctx structures are not yet defined with the const qualifier. Reviewed by: gallatin, erj MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D29102 (cherry picked from commit ffe3def903a5f239c319e5fe12450659658974a5)

iflib: Free resources in a consistent order during detach

2021-09-28T17:21:52Z

Memory and PCI resources are freed with no particular order. This could cause use-after-frees when detaching following a failed attach. For instance, iflib_tx_structures_free() frees ctx->ifc_txqs[] but iflib_tqg_detach() attempts to access this array. Similarly, adapter queues gets freed by IFDI_QUEUES_FREE() but IFDI_DETACH() attempts to access adapter queues to free PCI resources. MFC after: 2 weeks Sponsored by: NetApp, Inc. Differential Revision: https://reviews.freebsd.org/D27634 (cherry picked from commit 38bfc6dee33bedb290e1ea2540f97a86fe3caee0)

ixl(4): Fix 2.5 and 5G speeds reporting and update shared code

2021-09-20T22:12:32Z

Fix 2.5 and 5G speeds reporting and update shared code with recent changes: - Update expected FW API versions for X710 and X722 adapters - Define pointers related to Preservation Rules Module - Add definitions for Shadow RAM pointers to new modules: 5th and 6th FPA, and Preservation Rules Module. - Add I40E_RX_PTYPE_PARSER_ABORTED definition, so the driver will know opcode for parser aborted packets. - Add the new filter types needed for custom cloud filters. - Add support for Minimum Rollback Revision - Fix RX_ONLY mode for unicast promiscuous on VLAN - Add EEE LPI status check for X722 adapters - Fix PHY type identifiers for 2.5G and 5G adapters - Fix update link data for X722 - Increase the timeout value for PF reset to give PF more time to finish reset if it is loaded with filters. - Added support for Min Rollback Revision for 4 more X722 modules - Fix reporting of Active Optical Cable media type - Add flags and fields for double VLAN processing - Fix potentially uninitialized variables in NVM code Reviewed by: kbowling@, mike.jakubik@gmail.com Tested by: gowtham.kumar.ks@intel.com Sponsored by: Intel Corporation Differential Revision: https://reviews.freebsd.org/D31565 (cherry picked from commit abf774528d7e497460510b0026db85e30f054142)

ixl(4): Fix reporting of unqualified transceivers

2021-09-20T22:12:05Z

When link_active_on_if_down flag is disabled and link is brought down with ifconfig, FW reports a false positive link event about an unqualified transceiver. The condition used in the driver to filter out those false positive events was incorrect and caused that unqualified module event to also not be reported when the event was valid. Change the condition to rely on IFF_UP flag instead of link_active_on_if_down and bump driver version to 2.3.1-k. Signed-off-by: Krzysztof Galazka Signed-off-by: Eric Joyner Reviewed by: stallamr@netapp.com, erj@ Tested by: gowtham.kumar.ks@intel.com Sponsored by: Intel Corporation Differential Revision: https://reviews.freebsd.org/D30733 (cherry picked from commit c4622b01d2f12b889b57ff7d0b03a38dfcb00fd8)

ixl(4): Add tunable to override Flow Control settings

2021-09-20T22:11:14Z

Add flow_control to hw.ixl tunables tree to let override initial flow control configuration for all interfaces. Keep using configuration set by NVM by default. Reviewed by: erj@, gallatin@ Tested by: gowtham.kumar.ks_intel.com Sponsored by: Intel Corporation Differential Revision: https://reviews.freebsd.org/D29338 (cherry picked from commit 20a52706c814ccfd91c65586404abd2a1563a330)

ixl(4): Report RX errors as sum of all RX error counters

2021-09-20T21:31:03Z

HW keeps track of RX errors using several counters, each for specific type of errors. Report RX errors to OS as sum of all those counters: CRC errors, illegal bytes, checksum, length, undersize, fragment, oversize and jabber errors. There is no HW counter for frames with invalid L3/L4 checksums so add a SW one. Also add a "rx_errors" sysctl with a copy of netstat IERRORS counter value to make it easier accessible from scripts. Reviewed By: erj Tested By: gowtham.kumar.ks@intel.com Sponsored By: Intel Corporation Differential Revision: https://reviews.freebsd.org/D27639 (cherry picked from commit 9f99061ef9c95b171fc92d34026222bb5e052337)

ixl(4): Add ability to control link state on ifconfig down

2021-09-20T21:30:13Z

Add sysctl link_active_on_if_down, which allows user to control if interface is kept in active state when it is brought down with ifconfig. Set it to enabled by default to preserve backwards compatibility. Reviewed by: erj Tested by: gowtham.kumar.ks@intel.com Sponsored by: Intel Corporation Differential Revision: https://reviews.freebsd.org/D28028 (cherry picked from commit 21802a127d83e8a8c721d69b697e7ddb223d7797)

intel ethernet: Use ether_gen_addr(9)

2021-08-27T01:38:09Z

Use ether_gen_addr(9) for VF MAC generation Reviewed by: Intel Networking (erj), kevans MFC after: 1 week Differential Revision: https://reviews.freebsd.org/D31616 (cherry picked from commit 0e5811a2a9a1b4a7e0c79b73e3b9b9d15d481577)

ixl: Permit 802.1ad frames to pass though the chip

2021-02-02T14:42:38Z

This patch is a quick hack to change the internal Ethertype used within the chip. All frames with this type are dropped silently. This patch allows you to overwrite the factory default 0x88a8, which is used by IEEE 802.1ad VLAN stacking. Reviewed by: kp, philip, brueffer Approved by: philip (mentor) Differential Revision: https://reviews.freebsd.org/D24179 (cherry picked from commit fa6662b3689eeb71cb63c2b230ca08e7342cabf0)