Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-nextgrafted

Pull networking updates from Jakub Kicinski:
 "Core:

   - Add dedicated kmem_cache for typical/small skb->head, avoid having
     to access struct page at kfree time, and improve memory use.

   - Introduce sysctl to set default RPS configuration for new netdevs.

   - Define Netlink protocol specification format which can be used to
     describe messages used by each family and auto-generate parsers.
     Add tools for generating kernel data structures and uAPI headers.

   - Expose all net/core sysctls inside netns.

   - Remove 4s sleep in netpoll if carrier is instantly detected on
     boot.

   - Add configurable limit of MDB entries per port, and port-vlan.

   - Continue populating drop reasons throughout the stack.

   - Retire a handful of legacy Qdiscs and classifiers.

  Protocols:

   - Support IPv4 big TCP (TSO frames larger than 64kB).

   - Add IP_LOCAL_PORT_RANGE socket option, to control local port range
     on socket by socket basis.

   - Track and report in procfs number of MPTCP sockets used.

   - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path
     manager.

   - IPv6: don't check net.ipv6.route.max_size and rely on garbage
     collection to free memory (similarly to IPv4).

   - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986).

   - ICMP: add per-rate limit counters.

   - Add support for user scanning requests in ieee802154.

   - Remove static WEP support.

   - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate
     reporting.

   - WiFi 7 EHT channel puncturing support (client & AP).

  BPF:

   - Add a rbtree data structure following the "next-gen data structure"
     precedent set by recently added linked list, that is, by using
     kfunc + kptr instead of adding a new BPF map type.

   - Expose XDP hints via kfuncs with initial support for RX hash and
     timestamp metadata.

   - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to
     better support decap on GRE tunnel devices not operating in collect
     metadata.

   - Improve x86 JIT's codegen for PROBE_MEM runtime error checks.

   - Remove the need for trace_printk_lock for bpf_trace_printk and
     bpf_trace_vprintk helpers.

   - Extend libbpf's bpf_tracing.h support for tracing arguments of
     kprobes/uprobes and syscall as a special case.

   - Significantly reduce the search time for module symbols by
     livepatch and BPF.

   - Enable cpumasks to be used as kptrs, which is useful for tracing
     programs tracking which tasks end up running on which CPUs in
     different time intervals.

   - Add support for BPF trampoline on s390x and riscv64.

   - Add capability to export the XDP features supported by the NIC.

   - Add __bpf_kfunc tag for marking kernel functions as kfuncs.

   - Add cgroup.memory=nobpf kernel parameter option to disable BPF
     memory accounting for container environments.

  Netfilter:

   - Remove the CLUSTERIP target. It has been marked as obsolete for
     years, and we still have WARN splats wrt races of the out-of-band
     /proc interface installed by this target.

   - Add 'destroy' commands to nf_tables. They are identical to the
     existing 'delete' commands, but do not return an error if the
     referenced object (set, chain, rule...) did not exist.

  Driver API:

   - Improve cpumask_local_spread() locality to help NICs set the right
     IRQ affinity on AMD platforms.

   - Separate C22 and C45 MDIO bus transactions more clearly.

   - Introduce new DCB table to control DSCP rewrite on egress.

   - Support configuration of Physical Layer Collision Avoidance (PLCA)
     Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of
     shared medium Ethernet.

   - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing
     preemption of low priority frames by high priority frames.

   - Add support for controlling MACSec offload using netlink SET.

   - Rework devlink instance refcounts to allow registration and
     de-registration under the instance lock. Split the code into
     multiple files, drop some of the unnecessarily granular locks and
     factor out common parts of netlink operation handling.

   - Add TX frame aggregation parameters (for USB drivers).

   - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning
     messages with notifications for debug.

   - Allow offloading of UDP NEW connections via act_ct.

   - Add support for per action HW stats in TC.

   - Support hardware miss to TC action (continue processing in SW from
     a specific point in the action chain).

   - Warn if old Wireless Extension user space interface is used with
     modern cfg80211/mac80211 drivers. Do not support Wireless
     Extensions for Wi-Fi 7 devices at all. Everyone should switch to
     using nl80211 interface instead.

   - Improve the CAN bit timing configuration. Use extack to return
     error messages directly to user space, update the SJW handling,
     including the definition of a new default value that will benefit
     CAN-FD controllers, by increasing their oscillator tolerance.

  New hardware / drivers:

   - Ethernet:
      - nVidia BlueField-3 support (control traffic driver)
      - Ethernet support for imx93 SoCs
      - Motorcomm yt8531 gigabit Ethernet PHY
      - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA)
      - Microchip LAN8841 PHY (incl. cable diagnostics and PTP)
      - Amlogic gxl MDIO mux

   - WiFi:
      - RealTek RTL8188EU (rtl8xxxu)
      - Qualcomm Wi-Fi 7 devices (ath12k)

   - CAN:
      - Renesas R-Car V4H

  Drivers:

   - Bluetooth:
      - Set Per Platform Antenna Gain (PPAG) for Intel controllers.

   - Ethernet NICs:
      - Intel (1G, igc):
         - support TSN / Qbv / packet scheduling features of i226 model
      - Intel (100G, ice):
         - use GNSS subsystem instead of TTY
         - multi-buffer XDP support
         - extend support for GPIO pins to E823 devices
      - nVidia/Mellanox:
         - update the shared buffer configuration on PFC commands
         - implement PTP adjphase function for HW offset control
         - TC support for Geneve and GRE with VF tunnel offload
         - more efficient crypto key management method
         - multi-port eswitch support
      - Netronome/Corigine:
         - add DCB IEEE support
         - support IPsec offloading for NFP3800
      - Freescale/NXP (enetc):
         - support XDP_REDIRECT for XDP non-linear buffers
         - improve reconfig, avoid link flap and waiting for idle
         - support MAC Merge layer
      - Other NICs:
         - sfc/ef100: add basic devlink support for ef100
         - ionic: rx_push mode operation (writing descriptors via MMIO)
         - bnxt: use the auxiliary bus abstraction for RDMA
         - r8169: disable ASPM and reset bus in case of tx timeout
         - cpsw: support QSGMII mode for J721e CPSW9G
         - cpts: support pulse-per-second output
         - ngbe: add an mdio bus driver
         - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing
         - r8152: handle devices with FW with NCM support
         - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation
         - virtio-net: support multi buffer XDP
         - virtio/vsock: replace virtio_vsock_pkt with sk_buff
         - tsnep: XDP support

   - Ethernet high-speed switches:
      - nVidia/Mellanox (mlxsw):
         - add support for latency TLV (in FW control messages)
      - Microchip (sparx5):
         - separate explicit and implicit traffic forwarding rules, make
           the implicit rules always active
         - add support for egress DSCP rewrite
         - IS0 VCAP support (Ingress Classification)
         - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS
           etc.)
         - ES2 VCAP support (Egress Access Control)
         - support for Per-Stream Filtering and Policing (802.1Q,
           8.6.5.1)

   - Ethernet embedded switches:
      - Marvell (mv88e6xxx):
         - add MAB (port auth) offload support
         - enable PTP receive for mv88e6390
      - NXP (ocelot):
         - support MAC Merge layer
         - support for the the vsc7512 internal copper phys
      - Microchip:
         - lan9303: convert to PHYLINK
         - lan966x: support TC flower filter statistics
         - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x
         - lan937x: support Credit Based Shaper configuration
         - ksz9477: support Energy Efficient Ethernet
      - other:
         - qca8k: convert to regmap read/write API, use bulk operations
         - rswitch: Improve TX timestamp accuracy

   - Intel WiFi (iwlwifi):
      - EHT (Wi-Fi 7) rate reporting
      - STEP equalizer support: transfer some STEP (connection to radio
        on platforms with integrated wifi) related parameters from the
        BIOS to the firmware.

   - Qualcomm 802.11ax WiFi (ath11k):
      - IPQ5018 support
      - Fine Timing Measurement (FTM) responder role support
      - channel 177 support

   - MediaTek WiFi (mt76):
      - per-PHY LED support
      - mt7996: EHT (Wi-Fi 7) support
      - Wireless Ethernet Dispatch (WED) reset support
      - switch to using page pool allocator

   - RealTek WiFi (rtw89):
      - support new version of Bluetooth co-existance

   - Mobile:
      - rmnet: support TX aggregation"

* tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits)
  page_pool: add a comment explaining the fragment counter usage
  net: ethtool: fix __ethtool_dev_mm_supported() implementation
  ethtool: pse-pd: Fix double word in comments
  xsk: add linux/vmalloc.h to xsk.c
  sefltests: netdevsim: wait for devlink instance after netns removal
  selftest: fib_tests: Always cleanup before exit
  net/mlx5e: Align IPsec ASO result memory to be as required by hardware
  net/mlx5e: TC, Set CT miss to the specific ct action instance
  net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG
  net/mlx5: Refactor tc miss handling to a single function
  net/mlx5: Kconfig: Make tc offload depend on tc skb extension
  net/sched: flower: Support hardware miss to tc action
  net/sched: flower: Move filter handle initialization earlier
  net/sched: cls_api: Support hardware miss to tc action
  net/sched: Rename user cookie and act cookie
  sfc: fix builds without CONFIG_RTC_LIB
  sfc: clean up some inconsistent indentings
  net/mlx4_en: Introduce flexible array to silence overflow warning
  net: lan966x: Fix possible deadlock inside PTP
  net/ulp: Remove redundant ->clone() test in inet_clone_ulp().
  ...

1 files changed, 346 insertions, 0 deletions

diff --git a/Documentation/driver-api/surface_aggregator/ssh.rst b/Documentation/driver-api/surface_aggregator/ssh.rst
new file mode 100644
index 000000000..18fd0f0ae
--- /dev/null
+++ b/Documentation/driver-api/surface_aggregator/ssh.rst
@@ -0,0 +1,346 @@
+.. SPDX-License-Identifier: GPL-2.0+
+
+.. |u8| replace:: :c:type:`u8 <u8>`
+.. |u16| replace:: :c:type:`u16 <u16>`
+.. |TYPE| replace:: ``TYPE``
+.. |LEN| replace:: ``LEN``
+.. |SEQ| replace:: ``SEQ``
+.. |SYN| replace:: ``SYN``
+.. |NAK| replace:: ``NAK``
+.. |ACK| replace:: ``ACK``
+.. |DATA| replace:: ``DATA``
+.. |DATA_SEQ| replace:: ``DATA_SEQ``
+.. |DATA_NSQ| replace:: ``DATA_NSQ``
+.. |TC| replace:: ``TC``
+.. |TID| replace:: ``TID``
+.. |SID| replace:: ``SID``
+.. |IID| replace:: ``IID``
+.. |RQID| replace:: ``RQID``
+.. |CID| replace:: ``CID``
+
+===========================
+Surface Serial Hub Protocol
+===========================
+
+The Surface Serial Hub (SSH) is the central communication interface for the
+embedded Surface Aggregator Module controller (SAM or EC), found on newer
+Surface generations. We will refer to this protocol and interface as
+SAM-over-SSH, as opposed to SAM-over-HID for the older generations.
+
+On Surface devices with SAM-over-SSH, SAM is connected to the host via UART
+and defined in ACPI as device with ID ``MSHW0084``. On these devices,
+significant functionality is provided via SAM, including access to battery
+and power information and events, thermal read-outs and events, and many
+more. For Surface Laptops, keyboard input is handled via HID directed
+through SAM, on the Surface Laptop 3 and Surface Book 3 this also includes
+touchpad input.
+
+Note that the standard disclaimer for this subsystem also applies to this
+document: All of this has been reverse-engineered and may thus be erroneous
+and/or incomplete.
+
+All CRCs used in the following are two-byte ``crc_ccitt_false(0xffff, ...)``.
+All multi-byte values are little-endian, there is no implicit padding between
+values.
+
+
+SSH Packet Protocol: Definitions
+================================
+
+The fundamental communication unit of the SSH protocol is a frame
+(:c:type:`struct ssh_frame <ssh_frame>`). A frame consists of the following
+fields, packed together and in order:
+
+.. flat-table:: SSH Frame
+   :widths: 1 1 4
+   :header-rows: 1
+
+   * - Field
+     - Type
+     - Description
+
+   * - |TYPE|
+     - |u8|
+     - Type identifier of the frame.
+
+   * - |LEN|
+     - |u16|
+     - Length of the payload associated with the frame.
+
+   * - |SEQ|
+     - |u8|
+     - Sequence ID (see explanation below).
+
+Each frame structure is followed by a CRC over this structure. The CRC over
+the frame structure (|TYPE|, |LEN|, and |SEQ| fields) is placed directly
+after the frame structure and before the payload. The payload is followed by
+its own CRC (over all payload bytes). If the payload is not present (i.e.
+the frame has ``LEN=0``), the CRC of the payload is still present and will
+evaluate to ``0xffff``. The |LEN| field does not include any of the CRCs, it
+equals the number of bytes inbetween the CRC of the frame and the CRC of the
+payload.
+
+Additionally, the following fixed two-byte sequences are used:
+
+.. flat-table:: SSH Byte Sequences
+   :widths: 1 1 4
+   :header-rows: 1
+
+   * - Name
+     - Value
+     - Description
+
+   * - |SYN|
+     - ``[0xAA, 0x55]``
+     - Synchronization bytes.
+
+A message consists of |SYN|, followed by the frame (|TYPE|, |LEN|, |SEQ| and
+CRC) and, if specified in the frame (i.e. ``LEN > 0``), payload bytes,
+followed finally, regardless if the payload is present, the payload CRC. The
+messages corresponding to an exchange are, in part, identified by having the
+same sequence ID (|SEQ|), stored inside the frame (more on this in the next
+section). The sequence ID is a wrapping counter.
+
+A frame can have the following types
+(:c:type:`enum ssh_frame_type <ssh_frame_type>`):
+
+.. flat-table:: SSH Frame Types
+   :widths: 1 1 4
+   :header-rows: 1
+
+   * - Name
+     - Value
+     - Short Description
+
+   * - |NAK|
+     - ``0x04``
+     - Sent on error in previously received message.
+
+   * - |ACK|
+     - ``0x40``
+     - Sent to acknowledge receival of |DATA| frame.
+
+   * - |DATA_SEQ|
+     - ``0x80``
+     - Sent to transfer data. Sequenced.
+
+   * - |DATA_NSQ|
+     - ``0x00``
+     - Same as |DATA_SEQ|, but does not need to be ACKed.
+
+Both |NAK|- and |ACK|-type frames are used to control flow of messages and
+thus do not carry a payload. |DATA_SEQ|- and |DATA_NSQ|-type frames on the
+other hand must carry a payload. The flow sequence and interaction of
+different frame types will be described in more depth in the next section.
+
+
+SSH Packet Protocol: Flow Sequence
+==================================
+
+Each exchange begins with |SYN|, followed by a |DATA_SEQ|- or
+|DATA_NSQ|-type frame, followed by its CRC, payload, and payload CRC. In
+case of a |DATA_NSQ|-type frame, the exchange is then finished. In case of a
+|DATA_SEQ|-type frame, the receiving party has to acknowledge receival of
+the frame by responding with a message containing an |ACK|-type frame with
+the same sequence ID of the |DATA| frame. In other words, the sequence ID of
+the |ACK| frame specifies the |DATA| frame to be acknowledged. In case of an
+error, e.g. an invalid CRC, the receiving party responds with a message
+containing an |NAK|-type frame. As the sequence ID of the previous data
+frame, for which an error is indicated via the |NAK| frame, cannot be relied
+upon, the sequence ID of the |NAK| frame should not be used and is set to
+zero. After receival of an |NAK| frame, the sending party should re-send all
+outstanding (non-ACKed) messages.
+
+Sequence IDs are not synchronized between the two parties, meaning that they
+are managed independently for each party. Identifying the messages
+corresponding to a single exchange thus relies on the sequence ID as well as
+the type of the message, and the context. Specifically, the sequence ID is
+used to associate an ``ACK`` with its ``DATA_SEQ``-type frame, but not
+``DATA_SEQ``- or ``DATA_NSQ``-type frames with other ``DATA``- type frames.
+
+An example exchange might look like this:
+
+::
+
+    tx: -- SYN FRAME(D) CRC(F) PAYLOAD CRC(P) -----------------------------
+    rx: ------------------------------------- SYN FRAME(A) CRC(F) CRC(P) --
+
+where both frames have the same sequence ID (``SEQ``). Here, ``FRAME(D)``
+indicates a |DATA_SEQ|-type frame, ``FRAME(A)`` an ``ACK``-type frame,
+``CRC(F)`` the CRC over the previous frame, ``CRC(P)`` the CRC over the
+previous payload. In case of an error, the exchange would look like this:
+
+::
+
+    tx: -- SYN FRAME(D) CRC(F) PAYLOAD CRC(P) -----------------------------
+    rx: ------------------------------------- SYN FRAME(N) CRC(F) CRC(P) --
+
+upon which the sender should re-send the message. ``FRAME(N)`` indicates an
+|NAK|-type frame. Note that the sequence ID of the |NAK|-type frame is fixed
+to zero. For |DATA_NSQ|-type frames, both exchanges are the same:
+
+::
+
+    tx: -- SYN FRAME(DATA_NSQ) CRC(F) PAYLOAD CRC(P) ----------------------
+    rx: -------------------------------------------------------------------
+
+Here, an error can be detected, but not corrected or indicated to the
+sending party. These exchanges are symmetric, i.e. switching ``rx`` and
+``tx`` results again in a valid exchange. Currently, no longer exchanges are
+known.
+
+
+Commands: Requests, Responses, and Events
+=========================================
+
+Commands are sent as payload inside a data frame. Currently, this is the
+only known payload type of |DATA| frames, with a payload-type value of
+``0x80`` (:c:type:`SSH_PLD_TYPE_CMD <ssh_payload_type>`).
+
+The command-type payload (:c:type:`struct ssh_command <ssh_command>`)
+consists of an eight-byte command structure, followed by optional and
+variable length command data. The length of this optional data is derived
+from the frame payload length given in the corresponding frame, i.e. it is
+``frame.len - sizeof(struct ssh_command)``. The command struct contains the
+following fields, packed together and in order:
+
+.. flat-table:: SSH Command
+   :widths: 1 1 4
+   :header-rows: 1
+
+   * - Field
+     - Type
+     - Description
+
+   * - |TYPE|
+     - |u8|
+     - Type of the payload. For commands always ``0x80``.
+
+   * - |TC|
+     - |u8|
+     - Target category.
+
+   * - |TID|
+     - |u8|
+     - Target ID for commands/messages.
+
+   * - |SID|
+     - |u8|
+     - Source ID for commands/messages.
+
+   * - |IID|
+     - |u8|
+     - Instance ID.
+
+   * - |RQID|
+     - |u16|
+     - Request ID.
+
+   * - |CID|
+     - |u8|
+     - Command ID.
+
+The command struct and data, in general, does not contain any failure
+detection mechanism (e.g. CRCs), this is solely done on the frame level.
+
+Command-type payloads are used by the host to send commands and requests to
+the EC as well as by the EC to send responses and events back to the host.
+We differentiate between requests (sent by the host), responses (sent by the
+EC in response to a request), and events (sent by the EC without a preceding
+request).
+
+Commands and events are uniquely identified by their target category
+(``TC``) and command ID (``CID``). The target category specifies a general
+category for the command (e.g. system in general, vs. battery and AC, vs.
+temperature, and so on), while the command ID specifies the command inside
+that category. Only the combination of |TC| + |CID| is unique. Additionally,
+commands have an instance ID (``IID``), which is used to differentiate
+between different sub-devices. For example ``TC=3`` ``CID=1`` is a
+request to get the temperature on a thermal sensor, where |IID| specifies
+the respective sensor. If the instance ID is not used, it should be set to
+zero. If instance IDs are used, they, in general, start with a value of one,
+whereas zero may be used for instance independent queries, if applicable. A
+response to a request should have the same target category, command ID, and
+instance ID as the corresponding request.
+
+Responses are matched to their corresponding request via the request ID
+(``RQID``) field. This is a 16 bit wrapping counter similar to the sequence
+ID on the frames. Note that the sequence ID of the frames for a
+request-response pair does not match. Only the request ID has to match.
+Frame-protocol wise these are two separate exchanges, and may even be
+separated, e.g. by an event being sent after the request but before the
+response. Not all commands produce a response, and this is not detectable by
+|TC| + |CID|. It is the responsibility of the issuing party to wait for a
+response (or signal this to the communication framework, as is done in
+SAN/ACPI via the ``SNC`` flag).
+
+Events are identified by unique and reserved request IDs. These IDs should
+not be used by the host when sending a new request. They are used on the
+host to, first, detect events and, second, match them with a registered
+event handler. Request IDs for events are chosen by the host and directed to
+the EC when setting up and enabling an event source (via the
+enable-event-source request). The EC then uses the specified request ID for
+events sent from the respective source. Note that an event should still be
+identified by its target category, command ID, and, if applicable, instance
+ID, as a single event source can send multiple different event types. In
+general, however, a single target category should map to a single reserved
+event request ID.
+
+Furthermore, requests, responses, and events have an associated target ID
+(``TID``) and source ID (``SID``). These two fields indicate where a message
+originates from (``SID``) and what the intended target of the message is
+(``TID``). Note that a response to a specific request therefore has the source
+and target IDs swapped when compared to the original request (i.e. the request
+target is the response source and the request source is the response target).
+See (:c:type:`enum ssh_request_id <ssh_request_id>`) for possible values of
+both.
+
+Note that, even though requests and events should be uniquely identifiable by
+target category and command ID alone, the EC may require specific target ID and
+instance ID values to accept a command. A command that is accepted for
+``TID=1``, for example, may not be accepted for ``TID=2`` and vice versa. While
+this may not always hold in reality, you can think of different target/source
+IDs indicating different physical ECs with potentially different feature sets.
+
+
+Limitations and Observations
+============================
+
+The protocol can, in theory, handle up to ``U8_MAX`` frames in parallel,
+with up to ``U16_MAX`` pending requests (neglecting request IDs reserved for
+events). In practice, however, this is more limited. From our testing
+(although via a python and thus a user-space program), it seems that the EC
+can handle up to four requests (mostly) reliably in parallel at a certain
+time. With five or more requests in parallel, consistent discarding of
+commands (ACKed frame but no command response) has been observed. For five
+simultaneous commands, this reproducibly resulted in one command being
+dropped and four commands being handled.
+
+However, it has also been noted that, even with three requests in parallel,
+occasional frame drops happen. Apart from this, with a limit of three
+pending requests, no dropped commands (i.e. command being dropped but frame
+carrying command being ACKed) have been observed. In any case, frames (and
+possibly also commands) should be re-sent by the host if a certain timeout
+is exceeded. This is done by the EC for frames with a timeout of one second,
+up to two re-tries (i.e. three transmissions in total). The limit of
+re-tries also applies to received NAKs, and, in a worst case scenario, can
+lead to entire messages being dropped.
+
+While this also seems to work fine for pending data frames as long as no
+transmission failures occur, implementation and handling of these seems to
+depend on the assumption that there is only one non-acknowledged data frame.
+In particular, the detection of repeated frames relies on the last sequence
+number. This means that, if a frame that has been successfully received by
+the EC is sent again, e.g. due to the host not receiving an |ACK|, the EC
+will only detect this if it has the sequence ID of the last frame received
+by the EC. As an example: Sending two frames with ``SEQ=0`` and ``SEQ=1``
+followed by a repetition of ``SEQ=0`` will not detect the second ``SEQ=0``
+frame as such, and thus execute the command in this frame each time it has
+been received, i.e. twice in this example. Sending ``SEQ=0``, ``SEQ=1`` and
+then repeating ``SEQ=1`` will detect the second ``SEQ=1`` as repetition of
+the first one and ignore it, thus executing the contained command only once.
+
+In conclusion, this suggests a limit of at most one pending un-ACKed frame
+(per party, effectively leading to synchronous communication regarding
+frames) and at most three pending commands. The limit to synchronous frame
+transfers seems to be consistent with behavior observed on Windows.