SPM: Add SPMC documentation #107

jellesels-arm · 2021-03-29T07:43:19Z

Add documentation for the following SPMC features:
-Handling FF-A messages
-SP initialisation and start
-Rxtx buffers

Signed-off-by: Jelle Sels [email protected]

jenswi-linaro · 2021-03-29T18:10:29Z

architecture/spmc.rst

+Secure Partitions (SPs) are the endpoints used in the FF-A protocol. When
+OP-TEE is used as a SPMC SPs run primarily inside S-EL0.
+
+The SPMC also exposes the OP-TEE kernel as an secure endpoint itself. This is


s/the OP-TEE kernel/OP-TEE core, privileged mode,/

jenswi-linaro · 2021-03-29T18:12:21Z

architecture/spmc.rst

+OP-TEE is used as a SPMC SPs run primarily inside S-EL0.
+
+The SPMC also exposes the OP-TEE kernel as an secure endpoint itself. This is
+used to handle all OP-TEE legacy operations. All legacy operations will be


It might be a bit early to call the communication with TAs legacy operations. I'm not sure if SPs will ever be able to replace TAs.

@jenswi-linaro agreed, "TAs" in this context means the whole GlobalPlaform programming model...

jenswi-linaro · 2021-03-29T18:13:15Z

architecture/spmc.rst

+when the SPMD is enabled.
+
+SPs run from the initial boot of the system until power down and don't have any
+build-in session management compared to user TA. There only means of


/There/The/

jenswi-linaro · 2021-03-29T18:15:25Z

architecture/spmc.rst

+``Early TSs``. This is done because SP have to be loaded during boot. Early TSs
+are embedded inside the OP-TEE image. This makes it possible for us to load SPs
+during the initialisation of OP-TEE and we are not restricted to as rich-os in
+the normal world. LDELF is used to load the SPs into memory.


s/LDELF/Ldelf/

jenswi-linaro · 2021-03-29T18:16:20Z

architecture/spmc.rst

+The implementation of the SP handling in the SPMC is build upon OP-TEEs
+``Early TSs``. This is done because SP have to be loaded during boot. Early TSs
+are embedded inside the OP-TEE image. This makes it possible for us to load SPs
+during the initialisation of OP-TEE and we are not restricted to as rich-os in


I think this sentence needs to be rephrased a little.

jenswi-linaro · 2021-03-29T18:16:40Z

architecture/spmc.rst

+
+
+Each SP is loaded into the system using ldelf and started. This is done using
+LDELF and is based around the same process as loading the early TAs.


s/LDELF/ldelf/

jenswi-linaro · 2021-03-29T18:17:53Z

architecture/spmc.rst

+
+Each SP is loaded into the system using ldelf and started. This is done using
+LDELF and is based around the same process as loading the early TAs.
+All SPs are run after they are loaded and run until a FFA_MSG_WAIT is send by


Might want to put all FF-A defines inside ``

jenswi-linaro · 2021-03-29T18:18:58Z

architecture/spmc.rst

+        skinparam backgroundcolor transparent
+        participant None_secure_world
+
+        box "S-El3"


jenswi-linaro · 2021-03-29T18:19:45Z

architecture/spmc.rst

+Running and exiting SPs
+-----------------------
+
+THE SPMC resumes/starts q SP by calling the sp_enter(). This will set-up the SP


jforissier

Not too sure about the typographic conventions for structs, function names and file names but I think they should all be in double back ticks.

jforissier · 2021-03-29T19:38:21Z

architecture/spmc.rst

+SPMC
+====
+
+This document describes the SPMC(S-EL1) implementation for OP-TEE.


missing space SPMC (S-EL1)

jforissier · 2021-03-29T19:39:13Z

architecture/spmc.rst

+OP-TEE is used as a SPMC SPs run primarily inside S-EL0.
+
+The SPMC also exposes the OP-TEE kernel as an secure endpoint itself. This is
+used to handle all OP-TEE legacy operations. All legacy operations will be


@jenswi-linaro agreed, "TAs" in this context means the whole GlobalPlaform programming model...

jforissier · 2021-03-29T19:43:28Z

architecture/spmc.rst

+encapsulated inside FF-A messages. The OP-TEE endpoint will unpack the messages
+and afterwards handle them as standard OP-TEE calls. This is needed as
+TF-A (S-EL3) does only allow FF-A messages to be passed to the secure world
+when the SPMD is enabled.


Doesn't this introduces a noticeable overhead in the handling of TA messages? (if so it could be an obstacle to the adoption of the new FF-A protocol).

No, the only overhead is in how shared memory is registered. In the OP-TEE message model it can be a one-step process for temporary shared memory, while with FF-A it's always a two-step process (first register with SPMC and then invoke OP-TEE). It's covered at page 15 in https://static.linaro.org/connect/lvc21/presentations/lvc21-305.pdf

OK, good. Thanks.

jforissier · 2021-03-29T19:45:27Z

architecture/spmc.rst

+
+.. code-block:: Make
+
+	OPTEE_OS_COMMON_FLAGS += CCFG_WITH_SPMC=y CFG_CORE_FFA=y


CFG_WITH_SPMC

jforissier · 2021-03-29T19:47:07Z

conf.py

@@ -41,7 +41,10 @@
 extensions = [
    'sphinx.ext.todo',
    'sphinx.ext.viewcode',
-    'sphinx.ext.graphviz'
+    'sphinx.ext.graphviz',
+     #For plantuml to work you need a plantuml command be on your path. This


Missing space # For...

jforissier · 2021-03-29T20:03:41Z

architecture/spmc.rst

+s to the Normal World buffers.
+
+
+Structs and global objects


I don't think this section is useful at all. It is implementation details that will likely become outdated and which would be much easier documented and understood in the code itself.

jforissier · 2021-03-29T20:05:42Z

architecture/spmc.rst

+        Secure Partitions (Loading the Image, Setting up the stack, HEAP, ...).
+- Routing messages between endpoints:
+        The SPMC is responsible for passing FF-A messages from The Normal World
+        to SPs and back. It also needs to to pass FF-A messages between SPs


jforissier · 2021-03-29T20:08:00Z

architecture/spmc.rst

+The FF-A message is passed to the kernel thread and it will call the
+sp_msg_handler function.
+
+Whenever the SPMC (sp_msg_handler) receives a message not intended for one of


Shouldn't all function names be in double back ticks and have parentheses?

``sp_msg_handler()``

jforissier · 2021-03-29T20:08:23Z

architecture/spmc.rst

+the SPs, it will exit the kernel thread and return to the Normal World passing
+the FF-A message.
+
+Currently only a FFA_MSG_SEND_DIRECT_REQ can be passed from the Normal World to


``FFA_MSG_SEND_DIRECT_REQ``

jforissier · 2021-03-29T20:10:17Z

architecture/spmc.rst

+Transmit and receive buffers are used by the FF-A mechanism to communicate
+larger data structure between the SPMC and an endpoint. To accomplish this, each
+SP has his own rxtx_buf for access to it's transmit and receive buffers. There
+is also 1 defined for the Normal World (nw_buf in thread_spmc.c).


one

``nw_buf`` ``thread_spmc.c``

jbech-linaro · 2021-03-30T15:28:23Z

conf.py

+    'sphinx.ext.graphviz',
+     #For plantuml to work you need a plantuml command be on your path. This
+     # shall start plantuml.jar as: java -jar /path/to/plantuml.jar $@
+     'sphinxcontrib.plantuml'


Unsure how this will work, I've tried Google it, but didn't find any obvious solution for
a) Enable it in Travis
b) Enable it at readthedocs.io

There were some things that seems to be worth trying.

Hi,

there are the following dependencies:

graphviz

java runtime environment

plamtUML

the sphinxcontrib-plantuml PyPI package

On Ubuntu 1-3 can be installed with sudo apt install plantuml . 4 can be installed with pip3 install sphinxcontrib-plantuml.

RTD

The RTD build container has planuml and its dependencies installed.
Additional python packages PyPI packages can be set-up with a configuration file, but TS documentation is using plantuml, build fine and we don't have the mentioned extra config. Although we have a requirements.txt file, which might be taken up by RTD automatically. This is not mentioned in the documentation though.

Travis

I don't know much about the environment used here. sudo apt install plantuml and pip install sphinxcontrib-plantuml should get everything needed, if the environment is ubuntu based.

@gyuri-szing: thanks, that gives a couple of things to try. Having said that, I'm unable to do it right away, so the PR will have to be on hold for a little longer until I've found some time for try it out. Shouldn't be that long though ...

@jbech-linaro did you had any time to look at this?

Haven't had time to look into it yet. Probably have to push it out a bit longer (if no-one else in @OP-TEE/approvers have some bandwidth to spare).

I'll give it a shot.

I'll give it a shot.

Thanks Jens!

jenswi-linaro · 2021-03-31T12:54:49Z

architecture/spmc.rst

+        end box
+
+        autoactivate on
+        None_secure_world -> SPMD: FFA_MSG_SEND_DIRECT_REQ \n <font color=red>SMC


"Non-secure world" instead?

I can't have space in the participant name but I used the 'as' to change the tittle of the participant to "Non-secure world"

jenswi-linaro · 2021-03-31T12:55:22Z

architecture/spmc.rst

+
+        autoactivate on
+        None_secure_world -> SPMD: FFA_MSG_SEND_DIRECT_REQ \n <font color=red>SMC
+        SPMD -> thread_spmc: thread_spmc_msg_recv \n <font color=red>ERET


We should probably have () after each function name to help identifying functions.

jenswi-linaro · 2021-04-06T09:53:01Z

architecture/spmc.rst

+
+The SPMC also exposes the OP-TEE core, privileged mode, as an secure endpoint
+itself. This is used to handle all GlobalPlaform programming mode operations
+All s will GlobalPlatform messages are encapsulated inside FF-A messages.


jenswi-linaro · 2021-04-06T10:10:31Z

architecture/spmc.rst

+
+Starting SPs
+------------
+SP are loaded and started as the last step in OP-TEE's initialisation process.


jenswi-linaro · 2021-04-06T10:11:03Z

architecture/spmc.rst

+:``sp_init_set_registers()``:	Initialise the registers of the SP
+:``sp_msg_handler()``:		Handle the SPs FF-A message
+
+Once all SP are loaded and started we return to the SPMD and the Normal World is


jenswi-linaro · 2021-04-06T10:13:09Z

architecture/spmc.rst

+						is a SP.
+
+When a ``FFA_MSG_SEND_DIRECT_REQ`` message is received by the spmc from the
+Normal World, a new kernel thread is started.


I think you can drop "kernel". Generally we avoid using this word for OP-TEE core since it's easily confused with the Linux kernel.

jenswi-linaro · 2021-04-06T10:14:14Z

architecture/spmc.rst

+Running and exiting SPs
+-----------------------
+
+THE SPMC resumes/starts the SP by calling the sp_enter(). This will set-up the


etienne-lms · 2021-04-13T14:13:06Z

architecture/spmc.rst

+====
+
+This document describes the SPMC (S-EL1) implementation for OP-TEE.
+More information on the SPMC can be found in the FF-A specification can be


remove 2nd occurrence of "can be found in the":

-More information on the SPMC can be found in the FF-A specification can be -found int the +More information on the SPMC can be found in the FF-A specification: `FF-A spec <https://developer.arm.com/documentation/den0077/latest>`_.

etienne-lms · 2021-04-13T14:15:19Z

architecture/spmc.rst

+	The SPMC component is responsible for initialisation of the
+	Secure Partitions (Loading the Image, Setting up the stack, HEAP, ...).
+- Routing messages between endpoints:
+	The SPMC is responsible for passing FF-A messages from The Normal World


s/The Normal World/normal world/

etienne-lms · 2021-04-13T14:15:59Z

architecture/spmc.rst

+	Secure Partitions (Loading the Image, Setting up the stack, HEAP, ...).
+- Routing messages between endpoints:
+	The SPMC is responsible for passing FF-A messages from The Normal World
+	to SPs and back. It also needs to to pass FF-A messages between SPs.


s/It also needs to to pass/It is also responsible for passing/

etienne-lms · 2021-04-13T14:16:55Z

architecture/spmc.rst

+	to SPs and back. It also needs to to pass FF-A messages between SPs.
+- Memory management:
+	The SPMC is responsible for the memory management of the SPs. Memory
+	can be shared between SPs and from a SP to the Normal World.


s/from a SP to the Normal World./between a SP and the normal world./

etienne-lms · 2021-04-13T14:19:16Z

architecture/spmc.rst

+OP-TEE is used as a SPMC SPs run primarily inside S-EL0.
+
+The SPMC also exposes the OP-TEE core, privileged mode, as an secure endpoint
+itself. This is used to handle all GlobalPlaform programming mode operations


add a dot . and end of the above line.

Is the sentence below true? I understand it is a goal but I think this is not yet implemented. If so, maybe this doc should not be stated that as something that is available.

This was implemented by @jenswi-linaro

etienne-lms · 2021-04-13T14:35:24Z

architecture/spmc.rst

+						the source or the destination
+						is a SP.
+
+When a ``FFA_MSG_SEND_DIRECT_REQ`` message is received by the spmc from the


s/spmc/SPMC/g

etienne-lms · 2021-04-13T14:38:05Z

architecture/spmc.rst

+Every message received by the SPMC from the Normal World is handled in the
+``thread_spmc_msg_recv()`` function.
+
+When entering a SPs we need to be running on a thread. This is needed to be


s/SPs/SP/
s/on a thread/in an OP-TEE thread/

etienne-lms · 2021-04-13T14:40:00Z

architecture/spmc.rst

+``sp_enter()`` will copy the FF-A arguments and resume the SP.
+
+When the SPMC needs to have access to the SPs memory, it will call
+``ts_push_current_session()`` to gain access and ts_pop_current_session to


ts_pop_current_session formatting (s/ts_pop_current_session/``ts_pop_current_session``/)

etienne-lms · 2021-04-13T14:40:44Z

architecture/spmc.rst

+The SPMC resumes/starts the SP by calling the sp_enter(). This will set-up the
+SP context and jump into S-EL0.
+Whenever the SP performs a system call it will end-up in ``sp_handle_svc()``.
+``sp_handle_svc()`` store the current context of the SP and makes sure that we


s/store/stores/

etienne-lms · 2021-04-13T14:45:06Z

architecture/spmc.rst

+	OPTEE_OS_COMMON_FLAGS += CFG_WITH_SPMC=y CFG_CORE_FFA=y
+	OPTEE_OS_COMMON_FLAGS += CFG_WITH_SP=y
+	OPTEE_OS_COMMON_FLAGS += SP_PATHS="SP.elf"
+	TF_A_FLAGS += SPD=spmd SPMD_SPM_AT_SEL2=0


OPTEE_OS_COMMON_FLAGS relates to a config from OP-TEE/build.git. Maybe worth mentionning that here: or simply state that OP-TEE configuration should have:
CFG_WITH_SPMC=y
CFG_CORE_FFA=y
CFG_WITH_SP=y
SP_PATHS="path/to/sp-xxx.elf path/to/sp-yyy.elf

(by the way, it would be more consistent to rename SP_PATHS to CFG_SP_PATHS in optee_os.git)

I suppose SP_PATHS follows EARLY_TA_PATHS

true :) both should be CFG_s IMO. But this is probably not a real issue.

true :) both should be CFG_s IMO. But this is probably not a real issue.

No, it is on purpose that they are not CFG_* in order to make it possible to not build the "TA dev kit" twice, see https://github.com/OP-TEE/optee_os/blob/3.12.0/mk/config.mk#L297-L303 (indeed, defining a new CFG_ would cause recompilation). Probably not very important though.

ok, fine. i had forgotten this. i found back your comment.

OPTEE_OS_COMMON_FLAGS are still there

jenswi-linaro · 2021-05-11T13:54:14Z

Please rebase this now that we have enabled plantuml support in travis and readthedocs.

jenswi-linaro · 2021-05-12T06:56:44Z

architecture/spmc.rst

+SPMC Responsibilities
+=====================
+
+The SPMC is a critical component in the FF-A flow. Some of it mayor


s/it mayor/its major/

jenswi-linaro · 2021-05-12T07:00:36Z

architecture/spmc.rst

+FF-A messages to be passed to the secure world when the SPMD is enabled.
+
+SPs run from the initial boot of the system until power down and don't have any
+build-in session management compared to GPD TEE TAs. The only means of


s/build-in/built-in/

jenswi-linaro · 2021-05-12T07:01:21Z

architecture/spmc.rst

+
+SPs run from the initial boot of the system until power down and don't have any
+build-in session management compared to GPD TEE TAs. The only means of
+communicating with the outside world is through the FF-A spec defined messages.


...through messages defined in the FF-A specification.

jenswi-linaro · 2021-05-12T07:02:28Z

architecture/spmc.rst

+SPs run from the initial boot of the system until power down and don't have any
+build-in session management compared to GPD TEE TAs. The only means of
+communicating with the outside world is through the FF-A spec defined messages.
+SPs context is saved between executions.


The context of a SP is saved between executions.

jenswi-linaro · 2021-05-12T07:03:16Z

architecture/spmc.rst

+
+The
+`Trusted Service <https://www.trustedfirmware.org/projects/trusted-services/>`_
+repository includes the libSP libary which export all needed functions to build


Really capital letters in libSP? libraries usually have lower case letters only.

jenswi-linaro · 2021-05-12T07:09:20Z

architecture/spmc.rst

+
+
+Each SP is loaded into the system using ldelf and started. This is done using
+Ldelf and is based around the same process as loading the early TAs.


Mentioning ldelf here again is a bit repetitive. You may want to rephrase these two sentences.

jenswi-linaro · 2021-05-12T07:11:08Z

architecture/spmc.rst

+
+The SPMC is split into 2 main message handlers:
+
+:``thread_spmc_msg_recv()`` thread_spm.c:	Used to handle message coming


thread_spmc.c

jenswi-linaro · 2021-05-12T07:13:20Z

architecture/spmc.rst

+Every message received by the SPMC from the Normal World is handled in the
+``thread_spmc_msg_recv()`` function.
+
+When entering a SP we need to be running in a OP_TEE thread. This is needed to


s/OP_TEE/OP-TEE/

jenswi-linaro · 2021-05-12T07:14:35Z

architecture/spmc.rst

+Running and exiting SPs
+-----------------------
+
+The SPMC resumes/starts the SP by calling the sp_enter(). This will set-up the


sp_enter()

s/set-up/set up/

jenswi-linaro · 2021-05-12T07:15:36Z

architecture/spmc.rst

+
+The SPMC resumes/starts the SP by calling the sp_enter(). This will set-up the
+SP context and jump into S-EL0.
+Whenever the SP performs a system call it will end-up in ``sp_handle_svc()``.


jenswi-linaro · 2021-05-12T08:26:20Z

Reviewed-by: Jens Wiklander <[email protected]>

jellesels-arm · 2021-05-18T12:20:44Z

@jforissier, @etienne-lms any more comments?

jforissier

@jellesels-arm please see below, thanks.

jforissier · 2021-05-18T12:25:53Z

architecture/spmc.rst

+SPMC Responsibilities
+=====================
+
+The SPMC is a critical component in the FF-A flow. Some of its mayor


jforissier · 2021-05-18T12:26:31Z

architecture/spmc.rst

+
+- Initialisation and run-time management of the SPs:
+	The SPMC component is responsible for initialisation of the
+	Secure Partitions (Loading the Image, Setting up the stack, HEAP, ...).


secure partitions (no initial capitalization)? This is arguable, like normal world / Normal World. What is the convention used in Arm documentation?

loading, image, setting, heap

The spec uses Secure Partitions

jforissier · 2021-05-18T12:38:05Z

architecture/spmc.rst

+
+The SPMC also exposes the OP-TEE core, privileged mode, as an secure endpoint
+itself. This is used to handle all GlobalPlaform programming mode operations.
+All GlobalPlatform messages are encapsulated inside FF-A messages.


That is assuming that some config option is set to y. Which one is it? CFG_WITH_SPMC? CFG_CORE_SEL1_SPMC? CFG_CORE_FFA? This needs to be clearly documented.

jforissier · 2021-05-18T12:39:30Z

architecture/spmc.rst

+
+SPMC Program Flow
+=================
+SP images are stored OP-TEE image as early TAs are: the binary images are


in the OP-TEE image

jforissier · 2021-05-18T12:42:25Z

architecture/spmc.rst

+				World.
+
+
+Each SP is loaded into the system using ldelf and started. This is based around


jforissier · 2021-05-18T12:42:50Z

architecture/spmc.rst

+
+Each SP is loaded into the system using ldelf and started. This is done using
+Ldelf and is based around the same process as loading the early TAs.
+All SPs are run after they are loaded and run until a ``FFA_MSG_WAIT`` is send


I still see "send"

jforissier · 2021-05-18T12:44:56Z

architecture/spmc.rst

+embedded in OP-TEE Core in a read-only data section.
+This makes it possible to load SPs during boot. Early TSs are embedded inside
+the OP-TEE image. This makes it possible to load SPs during the initialisation
+of OP-TEE and no rich-os is needed in the normal world. Ldelf is used to load


You are saying the same thing twice, aren't you?

``ldelf``

jforissier · 2021-05-18T12:47:49Z

architecture/spmc.rst

+	OPTEE_OS_COMMON_FLAGS += CFG_WITH_SPMC=y CFG_CORE_FFA=y
+	OPTEE_OS_COMMON_FLAGS += CFG_WITH_SP=y
+	OPTEE_OS_COMMON_FLAGS += SP_PATHS="SP.elf"
+	TF_A_FLAGS += SPD=spmd SPMD_SPM_AT_SEL2=0


OPTEE_OS_COMMON_FLAGS are still there

jforissier · 2021-05-18T13:40:04Z

architecture/spmc.rst

-of OP-TEE and no rich-os is needed in the normal world. Ldelf is used to load
-and initialise the SPs.
+This makes it possible to load SPs during boot and no rich-os is needed in the
+normal world. ``Ldelf`` is used to load and initialise the SPs.


lowercase el

jforissier · 2021-05-18T13:41:48Z

architecture/spmc.rst

-	TF_A_FLAGS += SPD=spmd SPMD_SPM_AT_SEL2=0
+	OPTEE_OS_COMMON_FLAGS += CFG_CORE_FFA=y         #Enable the FF-A transport layer
+	OPTEE_OS_COMMON_FLAGS += SP_PATHS="path/to/sp-xxx.elf path/to/sp-yyy.elf" #Add the SPs to the OP-TEE image
+	TF_A_FLAGS += SPD=spmd SPMD_SPM_AT_SEL2=0       #Build TF-A with the SPMD enabled and without S-EL2


One space after the hash sign please

jforissier · 2021-05-18T13:42:42Z

With my last two comments addressed:
Reviewed-by: Jerome Forissier <[email protected]>

Add documentation for the following SPMC features: -Handling FF-A messages -SP initialisation and start -Rxtx buffers Change-Id: If4a50787036119e30a8cdd92f35a7032aacec73f Signed-off-by: Jelle Sels <[email protected]> Reviewed-by: Jens Wiklander <[email protected]> Reviewed-by: Jerome Forissier <[email protected]>

jbech-linaro · 2021-05-18T15:08:11Z

Thanks @jellesels-arm for the contribution and taking the time to do this. Thanks Jens & Jerome for doing the review. Merging now!

jenswi-linaro reviewed Mar 29, 2021

View reviewed changes

jforissier reviewed Mar 29, 2021

View reviewed changes

jbech-linaro reviewed Mar 30, 2021

View reviewed changes

jellesels-arm force-pushed the spmc_docs branch from a6ee5c0 to fae7012 Compare March 31, 2021 09:46

jenswi-linaro reviewed Mar 31, 2021

View reviewed changes

jellesels-arm force-pushed the spmc_docs branch 2 times, most recently from dae6508 to fd82c6d Compare April 6, 2021 08:02

jenswi-linaro reviewed Apr 6, 2021

View reviewed changes

jellesels-arm force-pushed the spmc_docs branch from fd82c6d to 291ab62 Compare April 6, 2021 10:57

etienne-lms reviewed Apr 13, 2021

View reviewed changes

jellesels-arm force-pushed the spmc_docs branch from 291ab62 to ae34869 Compare April 15, 2021 09:46

jenswi-linaro mentioned this pull request May 10, 2021

Testing with Plantuml #118

Merged

jellesels-arm force-pushed the spmc_docs branch from ae34869 to 031f018 Compare May 12, 2021 06:50

jenswi-linaro reviewed May 12, 2021

View reviewed changes

jellesels-arm force-pushed the spmc_docs branch 2 times, most recently from ba144e2 to 4306977 Compare May 12, 2021 08:10

jellesels-arm force-pushed the spmc_docs branch from 4306977 to 9524461 Compare May 13, 2021 07:39

jforissier reviewed May 18, 2021

View reviewed changes

jellesels-arm force-pushed the spmc_docs branch from 1591ce6 to c20afd8 Compare May 18, 2021 13:50

jbech-linaro merged commit 3ef999a into OP-TEE:master May 18, 2021



		Each SP is loaded into the system using ldelf and started. This is done using
		LDELF and is based around the same process as loading the early TAs.


		.. code-block:: Make

		OPTEE_OS_COMMON_FLAGS += CCFG_WITH_SPMC=y CFG_CORE_FFA=y


		The SPMC is split into 2 main message handlers:

		:``thread_spmc_msg_recv()`` thread_spm.c: Used to handle message coming

		World.


		Each SP is loaded into the system using ldelf and started. This is based around

SPM: Add SPMC documentation #107

SPM: Add SPMC documentation #107

Conversation

jellesels-arm commented Mar 29, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jforissier left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jellesels-arm Mar 31, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

RTD

Travis

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jellesels-arm Mar 31, 2021 •

edited

Loading