docs/components/secure-partition-manager.rst

Secure Partition Manager
************************

.. contents::

Acronyms
========

+--------+-----------------------------------+
| DMA    | Direct Memory Access              |
+--------+-----------------------------------+
| DTB    | Device Tree Blob                  |
+--------+-----------------------------------+
| DTS    | Device Tree Source                |
+--------+-----------------------------------+
| EC     | Execution Context                 |
+--------+-----------------------------------+
| FIP    | Firmware Image Package            |
+--------+-----------------------------------+
| FF-A   | Firmware Framework for A-class    |
+--------+-----------------------------------+
| IPA    | Intermediate Physical Address     |
+--------+-----------------------------------+
| NWd    | Normal World                      |
+--------+-----------------------------------+
| ODM    | Original Design Manufacturer      |
+--------+-----------------------------------+
| OEM    | Original Equipment Manufacturer   |
+--------+-----------------------------------+
| PA     | Physical Address                  |
+--------+-----------------------------------+
| PE     | Processing Element                |
+--------+-----------------------------------+
| PVM    | Primary VM                        |
+--------+-----------------------------------+
| SMMU   | System Memory Management Unit     |
+--------+-----------------------------------+
| SP     | Secure Partition                  |
+--------+-----------------------------------+
| SPM    | Secure Partition Manager          |
+--------+-----------------------------------+
| SPMC   | SPM Core                          |
+--------+-----------------------------------+
| SPMD   | SPM Dispatcher                    |
+--------+-----------------------------------+
| SiP    | Silicon Provider                  |
+--------+-----------------------------------+
| SWd    | Secure World                      |
+--------+-----------------------------------+
| TLV    | Tag-Length-Value                  |
+--------+-----------------------------------+
| TOS    | Trusted Operating System          |
+--------+-----------------------------------+
| VM     | Virtual Machine                   |
+--------+-----------------------------------+

Foreword
========

Two implementations of a Secure Partition Manager co-exist in the TF-A codebase:

-  SPM based on the FF-A specification `[1]`_.
-  SPM based on the MM interface to communicate with an S-EL0 partition `[2]`_.

Both implementations differ in their architectures and only one can be selected
at build time.

This document:

-  describes the FF-A implementation where the Secure Partition Manager
   resides at EL3 and S-EL2 (or EL3 and S-EL1).
-  is not an architecture specification and it might provide assumptions
   on sections mandated as implementation-defined in the specification.
-  covers the implications to TF-A used as a bootloader, and Hafnium
   used as a reference code base for an S-EL2 secure firmware on
   platforms implementing Armv8.4-SecEL2.

Terminology
-----------

-  Hypervisor refers to the NS-EL2 component managing Virtual Machines (or
   partitions) in the Normal World.
-  SPMC refers to the S-EL2 component managing Virtual Machines (or Secure
   Partitions) in the Secure World when Armv8.4-SecEL2 extension is implemented.
-  Alternatively, SPMC can refer to an S-EL1 component, itself being a Secure
   Partition and implementing the FF-A ABI on pre-Armv8.4 platforms.
-  VM refers to a Normal World Virtual Machine managed by an Hypervisor.
-  SP refers to a Secure World "Virtual Machine" managed by the SPMC component.

Support for legacy platforms
----------------------------

In the implementation, the SPM is split into SPMD and SPMC components
(although not strictly mandated by the specification). SPMD is located
at EL3 and principally relays FF-A messages from NWd (Hypervisor or OS
kernel) to SPMC located either at S-EL1 or S-EL2.

Hence TF-A must support both cases where SPMC is either located at:

-  S-EL1 supporting pre-Armv8.4 platforms. SPMD conveys FF-A protocol
   from EL3 to S-EL1.
-  S-EL2 supporting platforms implementing Armv8.4-SecEL2 extension.
   SPMD conveys FF-A protocol from EL3 to S-EL2.

The same SPMD component is used to support both configurations. The SPMC
execution level is a build time choice.

Sample reference stack
======================

The following diagram illustrates a possible configuration with SPMD and SPMC,
one or multiple Secure Partitions, with or without an optional Hypervisor:

.. image:: ../resources/diagrams/ff-a-spm-sel2.png

TF-A build options
==================

The following TF-A build options are provisioned:

-  **SPD=spmd**: this option selects the SPMD component to relay FF-A
   protocol from NWd to SWd back and forth. It is not possible to
   enable another Secure Payload Dispatcher when this option is chosen.
-  **SPMD_SPM_AT_SEL2**: this option adjusts the SPMC execution
   level to being S-EL1 or S-EL2. It defaults to enabled (value 1) when
   SPD=spmd is chosen.
-  **CTX_INCLUDE_EL2_REGS**: this option permits saving (resp.
   restoring) the EL2 system register context before entering (resp.
   after leaving) the SPMC. It is mandatory when ``SPMD_SPM_AT_SEL2`` is
   enabled. The context save/restore routine and exhaustive list of
   registers is visible at `[4]`_.
-  **SP_LAYOUT_FILE**: this option provides a text description file
   providing paths to SP binary images and DTS format manifests
   (see `Specifying partition binary image and DT`_). It
   is required when ``SPMD_SPM_AT_SEL2`` is enabled hence when multiple
   secure partitions are to be loaded on behalf of SPMC.

+------------------------------+----------------------+------------------+
|                              | CTX_INCLUDE_EL2_REGS | SPMD_SPM_AT_SEL2 |
+------------------------------+----------------------+------------------+
| SPMC at S-EL1 (e.g. OP-TEE)  |           0          |        0         |
+------------------------------+----------------------+------------------+
| SPMC at S-EL2 (e.g. Hafnium) |           1          | 1 (default when  |
|                              |                      |    SPD=spmd)     |
+------------------------------+----------------------+------------------+

Other combinations of such build options either break the build or are not
supported.

Note, the ``CTX_INCLUDE_EL2_REGS`` option provides the generic support for
barely saving/restoring EL2 registers from an Arm arch perspective. As such
it is decoupled from the ``SPD=spmd`` option.

BL32 option is re-purposed to specify the SPMC image. It can specify either the
Hafnium binary path (built for the secure world) or the path to a TEE binary
implementing the FF-A protocol.

BL33 option can specify either:

-  the TFTF binary or
-  the Hafnium binary path (built for the normal world) if VMs were loaded by
   TF-A beforehand or
-  a minimal loader performing the loading of VMs and Hafnium.

Sample TF-A build command line when SPMC is located at S-EL1
(typically pre-Armv8.4):

.. code:: shell

    make \
    CROSS_COMPILE=aarch64-none-elf- \
    SPD=spmd \
    SPMD_SPM_AT_SEL2=0 \
    BL32=<path-to-tee-binary> \
    BL33=<path-to-nwd-binary> \
    PLAT=fvp \
    all fip

Sample TF-A build command line for an Armv8.4-SecEL2 enabled system
where SPMC is located at S-EL2:

.. code:: shell

    make \
    CROSS_COMPILE=aarch64-none-elf- \
    SPD=spmd \
    CTX_INCLUDE_EL2_REGS=1 \
    ARM_ARCH_MINOR=4 \
    BL32=<path-to-swd-hafnium-binary>
    BL33=<path-to-nwd-binary> \
    SP_LAYOUT_FILE=sp_layout.json \
    PLAT=fvp \
    all fip

Build options to enable secure boot:

.. code:: shell

    make \
    CROSS_COMPILE=aarch64-none-elf- \
    SPD=spmd \
    CTX_INCLUDE_EL2_REGS=1 \
    ARM_ARCH_MINOR=4 \
    BL32=<path-to-swd-hafnium-binary>
    BL33=<path-to-nwd-binary> \
    SP_LAYOUT_FILE=../tf-a-tests/build/fvp/debug/sp_layout.json \
    MBEDTLS_DIR=<path-to-mbedtls-lib> \
    TRUSTED_BOARD_BOOT=1 \
    COT=dualroot \
    ARM_ROTPK_LOCATION=devel_rsa \
    ROT_KEY=plat/arm/board/common/rotpk/arm_rotprivk_rsa.pem \
    GENERATE_COT=1 \
    PLAT=fvp \
    all fip

Boot process
============

Loading Hafnium and Secure Partitions in the secure world
---------------------------------------------------------

The Hafnium implementation in normal world requires VMs to be loaded in
memory prior to booting. The mechanism upon which VMs are loaded and
exposed to Hafnium are either:

-  by supplying a ramdisk image where VM images are concatenated (1)
-  or by providing VM load addresses within Hafnium manifest (2)

TF-A is the bootlader for the Hafnium and SPs in the secure world. TF-A
does not provide tooling or libraries manipulating ramdisks as required
by (1). Thus BL2 loads SPs payloads independently.
SPs may be signed by different parties (SiP, OEM/ODM, TOS vendor, etc.).
Thus they are supplied as distinct “self-contained” signed entities within
the FIP flash image. The FIP image itself is not signed hence providing
ability to upgrade SPs in the field.

Booting through TF-A
--------------------

SP manifests
~~~~~~~~~~~~

An SP manifest describes SP attributes as defined in `[1]`_
section 3.1 (partition manifest at virtual FF-A instance) in DTS text format. It
is represented as a single file associated with the SP. A sample is
provided by `[5]`_. A binding document is provided by `[6]`_.

Secure Partition packages
~~~~~~~~~~~~~~~~~~~~~~~~~

Secure Partitions are bundled as independent package files consisting
of:

-  a header
-  a DTB
-  an image payload

The header starts with a magic value and offset values to SP DTB and
image payload. Each SP package is loaded independently by BL2 loader
and verified for authenticity and integrity.

The SP package identified by its UUID (matching FF-A uuid) is inserted
as a single entry into the FIP at end of the TF-A build flow as shown:

.. code:: shell

    Trusted Boot Firmware BL2: offset=0x1F0, size=0x8AE1, cmdline="--tb-fw"
    EL3 Runtime Firmware BL31: offset=0x8CD1, size=0x13000, cmdline="--soc-fw"
    Secure Payload BL32 (Trusted OS): offset=0x1BCD1, size=0x15270, cmdline="--tos-fw"
    Non-Trusted Firmware BL33: offset=0x30F41, size=0x92E0, cmdline="--nt-fw"
    HW_CONFIG: offset=0x3A221, size=0x2348, cmdline="--hw-config"
    TB_FW_CONFIG: offset=0x3C569, size=0x37A, cmdline="--tb-fw-config"
    SOC_FW_CONFIG: offset=0x3C8E3, size=0x48, cmdline="--soc-fw-config"
    TOS_FW_CONFIG: offset=0x3C92B, size=0x427, cmdline="--tos-fw-config"
    NT_FW_CONFIG: offset=0x3CD52, size=0x48, cmdline="--nt-fw-config"
    B4B5671E-4A90-4FE1-B81F-FB13DAE1DACB: offset=0x3CD9A, size=0xC168, cmdline="--blob"
    D1582309-F023-47B9-827C-4464F5578FC8: offset=0x48F02, size=0xC168, cmdline="--blob"

.. uml:: ../resources/diagrams/plantuml/fip-secure-partitions.puml

Specifying partition binary image and DT
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

A description file (json format) is passed to the build flow specifying
paths to the SP binary image and associated DTS partition manifest file.
The latter is going through the dtc compiler to generate the dtb fed into
the SP package.
This file also specifies the owner of the SP, which is an optional field and
identifies the signing domain in case of dualroot CoT.
The possible owner of an SP could either be Silicon Provider or Platform, and
the corresponding "owner" field value could either be "SiP" or "Plat".
In absence of "owner" field, it defaults to "SiP".

.. code:: shell

    {
        "tee1" : {
            "image": "tee1.bin",
             "pm": "tee1.dts",
             "owner": "SiP"
        },

        "tee2" : {
            "image": "tee2.bin",
            "pm": "tee2.dts",
            "owner": "Plat"
        }
    }

SPMC manifest
~~~~~~~~~~~~~

This manifest contains an SPMC attributes node consumed by SPMD at boot time. It
is implementing the description from `[1]`_ section 3.2 (SP manifest at physical
FF-A instance). The SP manifest at physical FF-A instance is used by the SPMD to
setup a SP that co-resides with the SPMC and executes at S-EL1 or Secure
Supervisor mode.

In this implementation its usage is extended to the secure physical FF-A
instance where SPMC executes at S-EL2.

.. code:: shell

    attribute {
        spmc_id = <0x8000>;
        maj_ver = <0x1>;
        min_ver = <0x0>;
        exec_state = <0x0>;
        load_address = <0x0 0x6000000>;
        entrypoint = <0x0 0x6000000>;
        binary_size = <0x60000>;
    };

-  *spmc_id* defines the endpoint ID value that SPMC can query through
   ``FFA_ID_GET``.
-  *maj_ver/min_ver*. SPMD checks provided version versus its internal
   version and aborts if not matching.
-  *exec_state* defines SPMC execution state (can be AArch64 for
   Hafnium, or AArch64/AArch32 for OP-TEE at S-EL1).
-  *load_address* and *binary_size* are mostly used to verify secondary
   entry points fit into the loaded binary image.
-  *entrypoint* defines the cold boot primary core entry point used by
   SPMD (currently matches ``BL32_BASE``)

Other nodes in the manifest are consumed by Hafnium in the secure world.
A sample can be found at [7]:

-  The *chosen* node is currently unused in SWd. It is meant for NWd to
   specify the init ramdisk image.
-  The *hypervisor* node describes SPs. *is_ffa_partition* boolean
   attribute indicates an SP. Load-addr field specifies the load address
   at which TF-A loaded the SP package.
-  *cpus* node provide the platform topology and allows MPIDR to VMPIDR
   mapping. Notice with current implementation primary cpu is declared
   first, then secondary cpus must be declared in reverse order.

SPMC boot
~~~~~~~~~

The SPMC is loaded by BL2 as the BL32 image.

The SPMC manifest is loaded by BL2 as the ``TOS_FW_CONFIG`` image.

BL2 passes the SPMC manifest address to BL31 through a register.

BL31(SPMD) runs from primary core, initializes the core contexts and
launches BL32 passing the SPMC manifest address through a register.

Loading of SPs
~~~~~~~~~~~~~~

.. uml:: ../resources/diagrams/plantuml/bl2-loading-sp.puml


Notice this boot flow is an implementation sample on Arm's FVP platform. Platforms
not using FW_CONFIG would adjust to a different implementation.

Secure boot
~~~~~~~~~~~

The SP content certificate is inserted as a separate FIP item so that BL2 loads SPMC,
SPMC manifest and Secure Partitions and verifies them for authenticity and integrity.
Refer to TBBR specification `[3]`_.

The multiple-signing domain feature (in current state dual signing domain) allows
the use of two root keys namely S-ROTPK and NS-ROTPK (see `[8]`_):

-  SPMC (BL32) and SPMC manifest are signed by the SiP using the S-ROTPK.
-  BL33 may be signed by the OEM using NS-ROTPK.
-  An SP may be signed either by SiP (using S-ROTPK) or by OEM (using NS-ROTPK).

Longer term multiple signing domain will allow additional signing keys, e.g.
if SPs originate from different parties.

See `TF-A build options`_ for a sample build command line.

Hafnium in the secure world
===========================

**NOTE: this section is work in progress. Descriptions and implementation choices
are subject to evolve.**

General considerations
----------------------

Build platform for the secure world
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The implementation might add specific code parts only relevant to the
secure world. Such code parts might be isolated into different files
and/or conditional code enclosed by a ``SECURE_WORLD`` macro.

Secure Partitions CPU scheduling
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

In the normal world, VMs are scheduled by the FFA_RUN ABI invoked from the
primary scheduler (in the primary VM), or by a direct message request or
response.

With the FF-A EAC specification, Secure Partitions are scheduled by direct
message invocations from a NWd VM or another SP.

Platform topology
~~~~~~~~~~~~~~~~~

As stated in `[1]`_ section 4.4.1 the SPMC implementation assumes the
following SP types:

-  Pinned MP SPs: an Execution Context id matches a physical PE id. MP
   SPs must implement the same number of ECs as the number of PEs in the
   platform. Hence the *execution-ctx-count* as defined by
   `[1]`_ (or NWd-Hafnium *vcpu_count*) can only take the
   value of one or the number of physical PEs.
-  Migratable UP SPs: a single execution context can run and be migrated
   on any physical PE. It declares a single EC in its SP manifest. An UP
   SP can receive a direct message request on any physical core.

Usage of PSCI services in the secure world
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- The normal world Hypervisor (optional) or OS kernel issues PSCI service
  invocations e.g. to request PSCI version, wake-up a secondary core, or request
  core suspend. This happens at the non-secure physical FF-A instance. In the
  example case of Hafnium in the normal world, it boots on the primary core and
  one of the first initialization step is to request the PSCI version. It then
  launches the primary VM. The primary VM upon initializing performs PSCI service
  calls (at non-secure virtual FF-A instance) which are trapped by the
  Hypervisor. Invocation from OS Kernel ends straight at EL3. The PVM issues
  ``PSCI_CPU_ON`` service calls to wake-up secondary cores by passing an
  ``MPIDR``, entry point address and a CPU context address. The EL3 PSCI layer
  then performs an exception return to the secondary core entry point on the
  targeted core. Other PSCI calls can happen at run-time from the PVM e.g. to
  request core suspend.
- In the existing TF-A PSCI standard library, PSCI service calls are filtered at
  EL3 to only originate from the NWd. Thus concerning the SPMC (at secure
  physical FF-A instance) the PSCI service invocations cannot happen as in the
  normal world. For example, a ``PSCI_CPU_ON`` service invocation from the SPMC
  does not reach the PSCI layer.

Parsing SP partition manifests
------------------------------

Hafnium must be able to consume SP manifests as defined in
`[1]`_ section 3.1, at least for the mandatory fields.

The SP manifest may contain memory and device regions nodes.

-  Memory regions shall be mapped in the SP Stage-2 translation regime at
   load time. A memory region node can specify RX/TX buffer regions in which
   case it is not necessary for an SP to explicitly call the ``FFA_RXTX_MAP``
   service.
-  Device regions shall be mapped in SP Stage-2 translation regime as
   peripherals and possibly allocate additional resources (e.g. interrupts)

Base addresses for memory and device region nodes are IPAs provided SPMC
identity maps IPAs to PAs within SP Stage-2 translation regime.

Note: currently both VTTBR_EL2 and VSTTBR_EL2 resolve to the same set of page
tables. It is still open whether two sets of page tables shall be provided per
SP. The memory region node as defined in the spec (section 3.1 Table 10)
provides a memory security attribute hinting to map either to the secure or
non-secure stage-2 table.

Passing boot data to the SP
---------------------------

`[1]`_ Section 3.4.2 “Protocol for passing data” defines a
method to passing boot data to SPs (not currently implemented).

Provided that the whole Secure Partition package image (see `Secure
Partition packages`_) is mapped to the SP's secure Stage-2 translation
regime, an SP can access its own manifest DTB blob and extract its partition
manifest properties.

SP Boot order
-------------

SP manifests provide an optional boot order attribute meant to resolve
dependencies such as an SP providing a service required to properly boot
another SP.

Boot phases
-----------

Primary core boot-up
~~~~~~~~~~~~~~~~~~~~

The SPMC performs its platform initializations then loads and creates
secure partitions based on SP packages and manifests. Then each secure
partition is launched in sequence (see `SP Boot order`_) on their primary
Execution Context.

Notice the primary physical core may not be core 0. Hence if the primary
core linear id is N, the 1:1 mapping requires MP SPs are launched using
EC[N] on PE[N] (see `Platform topology`_).

The SP's primary Execution Context (or the EC used when the partition is booted)
exits through ``FFA_MSG_WAIT`` to indicate successful initialization.

Secondary physical core boot-up
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Upon boot-up, the SPMC running on the primary core performs
implementation-defined SPMD service calls at secure physical FF-A instance
to register the secondary physical cores entry points and context information:

-  This is done through a direct message request invocation to the SPMD
   (``SET_ENTRY_POINT``). This service call does not wake-up the targeted
   core immediately. The secondary core is woken up later by a NWd
   ``PSCI_CPU_ON`` service invocation. A notification is passed from EL3
   PSCI layer to the SPMD, and then to SPMC through an implementation-defined
   interface.
-  The SPMC/SPMD interface can consist of FF-A direct message requests/responses
   transporting PM events.

If there is no Hypervisor in the normal world, the OS Kernel issues
``PSCI_CPU_ON`` calls that are directly trapped to EL3.

When a secondary physical core wakes-up the SPMD notifies the SPMC which updates
its internal states reflecting current physical core is being turned on.
It might then return straight to the SPMD and then to the NWd.

*(under discussion)* There may be possibility that an SP registers "PM events"
(during primary EC boot stage) through an ad-hoc interface. Such events would
be relayed by SPMC to one or more registered SPs on need basis
(see `Power management`_).

Secondary virtual core boot-up
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

In the example case where Hafnium exists in the normal world, secondary VMs
issue a ``PSCI_CPU_ON`` service call which is trapped to the Hypervisor. The
latter then enables the vCPU context for the targeted core, and switches to
the PVM down to the kernel driver with an ``HF_WAKE_UP`` message. The NWd
driver in PVM can then schedule the newly woken up vCPU context.

In the secure world the primary EC of a given SP passes the secondary EC entry
point and context. The SMC service call is trapped into the SPMC. This can be
either *(under discussion)*:

-  a specific interface registering the secondary EC entry point,
   similarly to above ``SET_ENTRY_POINT`` service.
-  Re-purposing the ``PSCI_CPU_ON`` function id. It is
   assumed that even if the input arguments are the same as the ones defined in
   the PSCI standard, the usage deviates by the fact the secondary EC is not
   woken up immediately. At least for the FF-A EAC where only
   direct messaging is allowed, it is only after the first direct
   message invocation that the secondary EC is entered. This option
   might be preferred when the same code base is re-used for a VM or
   an SP. The ABI to wake-up a secondary EC can remain similar.

SPs are always scheduled from the NWd, this paradigm did not change from legacy
TEEs. There must always be some logic (or driver) in the NWd to relinquish CPU
cycles to the SWd. If primary core is 0, an SP EC[x>0] entry point is supplied
by the SP EC[0] when the system boots in SWd. But this EC[x] is not immediately
entered at boot. Later in the boot process when NWd is up, a direct message
request issued from physical core 1 ends up in SP EC[1], and only at this stage
this context is effectively scheduled.

It should be possible for an SP to call into another SP through direct message
provided the latter SP has been booted already. The "boot-order" field in
partition manifests (`SP Boot order`_) fulfills the dependency towards availability
of a service within an SP offered to another SP.

Mandatory interfaces
--------------------

The following interfaces must be exposed to any VM or SP:

-  ``FFA_STATUS``
-  ``FFA_ERROR``
-  ``FFA_INTERRUPT``
-  ``FFA_VERSION``
-  ``FFA_FEATURES``
-  ``FFA_RX_RELEASE``
-  ``FFA_RXTX_MAP``
-  ``FFA_RXTX_UNMAP``
-  ``FFA_PARTITION_INFO_GET``
-  ``FFA_ID_GET``

FFA_VERSION
~~~~~~~~~~~

Per `[1]`_ section 8.1 ``FFA_VERSION`` requires a
*requested_version* parameter from the caller.

In the current implementation when ``FFA_VERSION`` is invoked from:

-  Hypervisor in NS-EL2: the SPMD returns the SPMC version specified
   in the SPMC manifest.
-  OS kernel in NS-EL1 when NS-EL2 is not present: the SPMD returns the
   SPMC version specified in the SPMC manifest.
-  VM in NWd: the Hypervisor returns its implemented version.
-  SP in SWd: the SPMC returns its implemented version.
-  SPMC at S-EL1/S-EL2: the SPMD returns its implemented version.

FFA_FEATURES
~~~~~~~~~~~~

FF-A features may be discovered by Secure Partitions while booting
through the SPMC. However, SPMC cannot get features from Hypervisor
early at boot time as NS world is not setup yet.

The Hypervisor may decide to gather FF-A features from SPMC through SPMD
once at boot time and store the result. Later when a VM requests FF-A
features, the Hypervisor can adjust its own set of features with what
SPMC advertised, if necessary. Another approach is to always forward FF-A
features to the SPMC when a VM requests it to the Hypervisor. Although
the result is not supposed to change over time so there may not be added
value doing the systematic forwarding.

FFA_RXTX_MAP/FFA_RXTX_UNMAP
~~~~~~~~~~~~~~~~~~~~~~~~~~~

VM mailboxes are re-purposed to serve as SP RX/TX buffers. The RX/TX
map API maps the send and receive buffer IPAs to the SP Stage-2 translation regime.

Hafnium in the normal world defines VMs and their attributes as logical structures,
including a mailbox used for FF-A indirect messaging, memory sharing, or the
`FFA_PARTITION_INFO_GET`_  ABI.
This same mailbox structure is re-used in the SPMC. `[1]`_ states only direct
messaging is allowed to SPs. Thus mailbox usage is restricted to implementing
`FFA_PARTITION_INFO_GET`_ and memory sharing ABIs.

FFA_PARTITION_INFO_GET
~~~~~~~~~~~~~~~~~~~~~~

Partition info get service call can originate:

-  from SP to SPM
-  from VM to Hypervisor
-  from Hypervisor to SPM

For the latter case, the service call must be forwarded through the SPMD.

FFA_ID_GET
~~~~~~~~~~

The SPMD returns:

-  a default zero value on invocation from the Hypervisor.
-  The ``spmc_id`` value specified in the SPMC manifest on invocation from
   the SPMC (see `SPMC manifest`_)

The FF-A id space is split into a non-secure space and secure space:

-  FF-A id with bit 15 clear refer to normal world VMs.
-  FF-A id with bit 15 set refer to secure world SPs

Such convention helps the SPMC discriminating the origin and destination worlds
in an FF-A service invocation. In particular the SPMC shall filter unauthorized
transactions in its world switch routine. It must not be permitted for a VM to
use a secure FF-A id as origin world through spoofing:

-  A VM-to-SP messaging passing shall have an origin world being non-secure
   (FF-A id bit 15 clear) and destination world being secure (FF-A id bit 15
   set).
-  Similarly, an SP-to-SP message shall have FF-A id bit 15 set for both origin
   and destination ids.

An incoming direct message request arriving at SPMD from NWd is forwarded to
SPMC without a specific check. The SPMC is resumed through eret and "knows" the
message is coming from normal world in this specific code path. Thus the origin
endpoint id must be checked by SPMC for being a normal world id.

An SP sending a direct message request must have bit 15 set in its origin
endpoint id and this can be checked by the SPMC when the SP invokes the ABI.

The SPMC shall reject the direct message if the claimed world in origin endpoint
id is not consistent:

-  It is either forwarded by SPMD and thus origin endpoint id must be a "normal
   world id",
-  or initiated by an SP and thus origin endpoint id must be a "secure world id".

Direct messaging
----------------

This is a mandatory interface for Secure Partitions consisting in direct
message request and responses.

The ``ffa_handler`` Hafnium function may:

-  trigger a world change e.g. when an SP invokes the direct message
   response ABI to a VM.
-  handle multiple requests from the NWd without resuming an SP.

SP-to-SP
~~~~~~~~

-  An SP can send a direct message request to another SP
-  An SP can receive a direct message response from another SP.

VM-to-SP
~~~~~~~~

-  A VM can send a direct message request to an SP
-  An SP can send a direct message response to a VM

SPMC-SPMD messaging
~~~~~~~~~~~~~~~~~~~

Specific implementation-defined endpoint IDs are allocated to the SPMC and SPMD.
Referring those IDs in source/destination fields of a direct message
request/response permits SPMD to SPMC messaging back and forth.

Per `[1]`_ Table 114 Config No. 1 (physical FF-A instance):

-  SPMC=>SPMD direct message request uses SMC conduit
-  SPMD=>SPMC direct message request uses ERET conduit

Per `[1]`_ Table 118 Config No. 1 (physical FF-A instance):

-  SPMC=>SPMD direct message response uses SMC conduit
-  SPMD=>SPMC direct message response uses ERET conduit

Memory management
-----------------

This section only deals with the PE MMU configuration.

Hafnium in the normal world deals with NS buffers only and provisions
a single root page table directory to VMs. In context of S-EL2 enabled
firmware, two IPA spaces are output from Stage-1 translation (secure
and non-secure). The Stage-2 translation handles:

-  A single secure IPA space when an SP Stage-1 MMU is disabled.
-  Two IPA spaces (secure and non-secure) when Stage-1 MMU is enabled.

``VTCR_EL2`` and ``VSTCR_EL2`` provide additional bits for controlling the
NS/S IPA translations (``VSTCR_EL2.SW``, ``VSTCR_EL2.SA``, ``VTCR_EL2.NSW``,
``VTCR_EL2.NSA``). There may be two approaches:

-  secure and non-secure mappings are rooted as two separate root page
   tables
-  secure and non-secure mappings use the same root page table. Access
   from S-EL1 to an NS region translates to a secure physical address
   space access.

Interrupt management
--------------------

Road to a para-virtualized interface
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Current Hafnium implementation uses an ad-hoc mechanism for a VM to get
a pending interrupt number through an hypercall. The PVM injects
interrupts to VMs by delegation from the Hypervisor. The PVM probes a
pending interrupt directly from the GIC distributor.

The short-term plan is to have Hafnium/SPMC in the secure world owner
of the GIC configuration.

The SPMC fully owns the GIC configuration at S-EL2. The SPMC manages
interrupt resources and allocates interrupt ID based on SP manifests.
The SPMC acknowledges physical interrupts and injects virtual interrupts
by setting the vIRQ bit when resuming an SP. A Secure Partition gathers
the interrupt number through an hypercall.

Notice the SPMC/SPMD has to handle Group0 secure interrupts in addition
to Group1 S/NS interrupts.

Power management
----------------

Assumption on the Nwd:

-  NWd is the best candidate to own the platform Power Management
   policy. It is master to invoking PSCI service calls from physical
   CPUs.
-  EL3 monitor is in charge of the PM control part (its PSCI layer
   actually writing to platform registers).
-  It is fine for the Hypervisor to trap PSCI calls and relay to EL3, or
   OS kernel driver to emit PSCI service calls.

PSCI notification are relayed through the SPMD/SPD PM hooks to the SPMC.
This can either be through re-use of PSCI FIDs or an FF-A direct message
from SPMD to SPMC.

The SPMD performs an exception return to the SPMC which is resumed to
its ``eret_handler`` routine. It is then either consuming a PSCI FID or
an FF-A FID. Depending on the servicing, the SPMC may return directly to
the SPMD (and then NWd) without resuming an SP at this stage. An example
of this is invocation of ``FFA_PARTITION_INFO_GET`` from NWd relayed by
the SPMD to the SPMC. The SPMC returns the needed partition information
to the SPMD (then NWd) without actually resuming a partition in secure world.

*(under discussion)*
About using PSCI FIDs from SPMD to SPMC to notify of PM events, it is still
questioned what to use as the return code from the SPMC.
If the function ID used by the SPMC is not an FF-A ID when doing SMC, then the
EL3 std svc handler won't route the response to the SPMD. That's where comes the
idea to embed the notification into an FF-A message. The SPMC can discriminate
this message as being a PSCI event, process it, and reply with an FF-A return
message that the SPMD receives as an acknowledgement.

SP notification
---------------

Power management notifications are conveyed from PSCI library to the
SPMD / SPD hooks. A range of events can be relayed to SPMC.

SPs may need to be notified about specific PM events.

-  SPs might register PM events to the SPMC
-  On SPMD to SPMC notification, a limited range of SPs may be notified
   through a direct message.
-  This assumes the mentioned SPs supports managed exit.

The SPMC is the first to be notified about PM events from the SPMD. It is up
to the SPMC to arbitrate to which SP it needs to send PM events.
An SP explicitly registers to receive notifications to specific PM events.
The register operation can either be an implementation-defined service call
to the SPMC when the primary SP EC boots, or be supplied through the SP
manifest.

Support for SMMUv3 in Hafnium
=============================

An SMMU is analogous to an MMU in a CPU. It performs address translations for
Direct Memory Access (DMA) requests from system I/O devices.
The responsibilities of an SMMU include:

-  Translation: Incoming DMA requests are translated from bus address space to
   system physical address space using translation tables compliant to
   Armv8/Armv7 VMSA descriptor format.
-  Protection: An I/O device can be prohibited from read, write access to a
   memory region or allowed.
-  Isolation: Traffic from each individial device can be independently managed.
   The devices are differentiated from each other using unique translation
   tables.

The following diagram illustrates a typical SMMU IP integrated in a SoC with
several I/O devices along with Interconnect and Memory system.

.. image:: ../resources/diagrams/MMU-600.png

SMMU has several versions including SMMUv1, SMMUv2 and SMMUv3. Hafnium provides
support for SMMUv3 driver in both Normal and Secure World. A brief introduction
of SMMUv3 functionality and the corresponding software support in Hafnium is
provided here.

SMMUv3 features
---------------

-  SMMUv3 provides Stage1, Stage2 translation as well as nested (Stage1 + Stage2)
   translation support. It can either bypass or abort incoming translations as
   well.
-  Traffic (memory transactions) from each upstream I/O peripheral device,
   referred to as Stream, can be independently managed using a combination of
   several memory based configuration structures. This allows the SMMUv3 to
   support a large number of streams with each stream assigned to a unique
   translation context.
-  Support for Armv8.1 VMSA where the SMMU shares the translation tables with
   a Processing Element. AArch32(LPAE) and AArch64 translation table format
   are supported by SMMUv3.
-  SMMUv3 offers non-secure stream support with secure stream support being
   optional. Logically, SMMUv3 behaves as if there is an indepdendent SMMU
   instance for secure and non-secure stream support.
-  It also supports sub-streams to differentiate traffic from a virtualized
   peripheral associated with a VM/SP.
-  Additionally, SMMUv3.2 provides support for PEs implementing Armv8.4-A
   extensions. Consequently, SPM depends on Secure EL2 support in SMMUv3.2
   for providing Secure Stage2 translation support to upstream peripheral
   devices.

SMMUv3 Programming Interfaces
-----------------------------

SMMUv3 has three software interfaces that are used by the Hafnium driver to
configure the behaviour of SMMUv3 and manage the streams.

-  Memory based data strutures that provide unique translation context for
   each stream.
-  Memory based circular buffers for command queue and event queue.
-  A large number of SMMU configuration registers that are memory mapped during
   boot time by Hafnium driver. Except a few registers, all configuration
   registers have independent secure and non-secure versions to configure the
   behaviour of SMMUv3 for translation of secure and non-secure streams
   respectively.

Peripheral device manifest
--------------------------

Currently, SMMUv3 driver in Hafnium only supports dependent peripheral devices.
These devices are dependent on PE endpoint to initiate and receive memory
management transactions on their behalf. The acccess to the MMIO regions of
any such device is assigned to the endpoint during boot. Moreover, SMMUv3 driver
uses the same stage 2 translations for the device as those used by partition
manager on behalf of the PE endpoint. This ensures that the peripheral device
has the same visibility of the physical address space as the endpoint. The
device node of the corresponding partition manifest (refer to `[1]`_ section 3.2
) must specify these additional properties for each peripheral device in the
system :

-  smmu-id: This field helps to identify the SMMU instance that this device is
   upstream of.
-  stream-ids: List of stream IDs assigned to this device.

.. code:: shell

    smmuv3-testengine {
        base-address = <0x00000000 0x2bfe0000>;
        pages-count = <32>;
        attributes = <0x3>;
        smmu-id = <0>;
        stream-ids = <0x0 0x1>;
        interrupts = <0x2 0x3>, <0x4 0x5>;
        exclusive-access;
    };

SMMUv3 driver limitations
-------------------------

The primary design goal for the Hafnium SMMU driver is to support secure
streams.

-  Currently, the driver only supports Stage2 translations. No support for
   Stage1 or nested translations.
-  Supports only AArch64 translation format.
-  No support for features such as PCI Express (PASIDs, ATS, PRI), MSI, RAS,
   Fault handling, Performance Monitor Extensions, Event Handling, MPAM.
-  No support for independent peripheral devices.

References
==========

.. _[1]:

[1] `Arm Firmware Framework for Armv8-A <https://developer.arm.com/docs/den0077/latest>`__

.. _[2]:

[2] :ref:`Secure Partition Manager using MM interface<Secure Partition Manager (MM)>`

.. _[3]:

[3] `Trusted Boot Board Requirements
Client <https://developer.arm.com/docs/den0006/latest/trusted-board-boot-requirements-client-tbbr-client-armv8-a>`__

.. _[4]:

[4] https://git.trustedfirmware.org/TF-A/trusted-firmware-a.git/tree/lib/el3_runtime/aarch64/context.S#n45

.. _[5]:

[5] https://git.trustedfirmware.org/TF-A/tf-a-tests.git/tree/spm/cactus/cactus.dts

.. _[6]:

[6] https://trustedfirmware-a.readthedocs.io/en/latest/components/ffa-manifest-binding.html

.. _[7]:

[7] https://git.trustedfirmware.org/TF-A/trusted-firmware-a.git/tree/plat/arm/board/fvp/fdts/fvp_spmc_manifest.dts

.. _[8]:

[8] https://developer.trustedfirmware.org/w/tf_a/poc-multiple-signing-domains/

--------------

*Copyright (c) 2020-2021, Arm Limited and Contributors. All rights reserved.*