Trusted Firmware-A Tests - User Guide

This document describes how to build the Trusted Firmware-A Tests (TF-A Tests) and run them on a set of platforms. It assumes that the reader has previous experience building and running the Trusted Firmware-A (TF-A).

1.   Host machine requirements

The minimum recommended machine specification for building the software and running the FVP models is a dual-core processor running at 2GHz with 12GB of RAM. For best performance, use a machine with a quad-core processor running at 2.6GHz with 16GB of RAM.

The software has been tested on Ubuntu 16.04 LTS (64-bit). Packages used for building the software were installed from that distribution unless otherwise specified.

2.   Tools

Install the required packages to build TF-A Tests with the following command:

sudo apt-get install device-tree-compiler build-essential make git perl libxml-libxml-perl

Download and install the GNU cross-toolchain from Linaro. The TF-A Tests have been tested with version 6.2-2016.11 (gcc 6.2):

In addition, the following optional packages and tools may be needed:

3.   Getting the TF-A Tests source code

Download the TF-A Tests source code using the following command:

git clone https://git.trustedfirmware.org/TF-A/tf-a-tests.git

4.   Building TF-A Tests

  • Before building TF-A Tests, the environment variable CROSS_COMPILE must point to the Linaro cross compiler.

    For AArch64:

    export CROSS_COMPILE=<path-to-aarch64-gcc>/bin/aarch64-linux-gnu-

    For AArch32:

    export CROSS_COMPILE=<path-to-aarch32-gcc>/bin/arm-linux-gnueabihf-
  • Change to the root directory of the TF-A Tests source tree and build.

    For AArch64:

    make PLAT=<platform>

    For AArch32:

    make PLAT=<platform> ARCH=aarch32


    • If PLAT is not specified, fvp is assumed by default. See the Summary of build options for more information on available build options.

    • By default this produces a release version of the build. To produce a debug version instead, build the code with DEBUG=1.

    • The build process creates products in a build/ directory tree, building the objects and binaries for each test image in separate sub-directories. The following binary files are created from the corresponding ELF files:

      • build/<platform>/<build-type>/tftf.bin
      • build/<platform>/<build-type>/ns_bl1u.bin
      • build/<platform>/<build-type>/ns_bl2u.bin
      • build/<platform>/<build-type>/el3_payload.bin
      • build/<platform>/<build-type>/cactus_mm.bin
      • build/<platform>/<build-type>/cactus.bin
      • build/<platform>/<build-type>/ivy.bin
      • build/<platform>/<build-type>/quark.bin

      where <platform> is the name of the chosen platform and <build-type> is either debug or release. The actual number of images might differ depending on the platform.

      Refer to the sections below for more information about each image.

  • Build products for a specific build variant can be removed using:

    make DEBUG=<D> PLAT=<platform> clean

    ... where <D> is 0 or 1, as specified when building.

    The build tree can be removed completely using:

    make realclean
  • Use the following command to list all supported build commands:

    make help

4.1.   TFTF test image

tftf.bin is the main test image to exercise the TF-A features. The other test images provided in this repository are optional dependencies that TFTF needs to test some specific features.

tftf.bin may be built independently of the other test images using the following command:

make PLAT=<platform> tftf

In TF-A boot flow, tftf.bin replaces the BL33 image and should be injected in the FIP image. This might be achieved by running the following command from the TF-A root directory:

BL33=tftf.bin make PLAT=<platform> fip

Please refer to the TF-A User guide for further details.

4.2.   NS_BL1U and NS_BL2U test images

ns_bl1u.bin and ns_bl2u.bin are test images that exercise the Firmware Update (FWU) feature of TF-A [1]. Throughout this document, they will be referred as the FWU test images.

In addition to updating the firmware, the FWU test images also embed some tests that exercise the FWU state machine implemented in the TF-A. They send valid and invalid SMC requests to the TF-A BL1 image in order to test its robustness.

4.2.1.   NS_BL1U test image

The NS_BL1U image acts as the Application Processor (AP) Firmware Update Boot ROM. This typically is the first software agent executing on the AP in the Normal World during a firmware update operation. Its primary purpose is to load subsequent firmware update images from an external interface, such as NOR Flash, and communicate with BL1 to authenticate those images.

The NS_BL1U test image provided in this repository performs the following tasks:

  • Load FWU images from external non-volatile storage (typically flash memory) to Non-Secure RAM.
  • Request TF-A BL1 to copy these images in Secure RAM and authenticate them.
  • Jump to NS_BL2U which carries out the next steps in the firmware update process.

This image may be built independently of the other test images using the following command:

make PLAT=<platform> ns_bl1u

4.2.2.   NS_BL2U test image

The NS_BL2U image acts as the AP Firmware Updater. Its primary responsibility is to load a new set of firmware images from an external interface and write them into non-volatile storage.

The NS_BL2U test image provided in this repository overrides the original FIP image stored in flash with the backup FIP image (see below).

This image may be built independently of the other test images using the following command:

make PLAT=<platform> ns_bl2u

4.2.3.   Putting it all together

The FWU test images should be used in conjunction with the TFTF image, as the latter initiates the FWU process by corrupting the FIP image and resetting the target. Once the FWU process is complete, TFTF takes over again and checks that the firmware was successfully updated.

To sum up, 3 images must be built out of the TF-A Tests repository in order to test the TF-A Firmware Update feature:

  • ns_bl1u.bin
  • ns_bl2u.bin
  • tftf.bin

Once that's done, they must be combined in the right way.

  • ns_bl1u.bin is a standalone image and does not require any further processing.
  • ns_bl2u.bin must be injected into the FWU_FIP image. This might be achieved by setting NS_BL2U=ns_bl2u.bin when building the FWU_FIP image out of the TF-A repository. Please refer to the section Building FIP images with support for Trusted Board Boot in the TF-A User Guide.
  • tftf.bin must be injected in the standard FIP image, as explained in section TFTF test image.

Additionally, on Juno platform, the FWU FIP must contain a SCP_BL2U image. This image can simply be a copy of the standard SCP_BL2 image if no specific firmware update operations need to be carried on the SCP side.

Finally, the backup FIP image must be created. This can simply be a copy of the standard FIP image, which means that the Firmware Update process will restore the original, uncorrupted FIP image.

4.3.   EL3 test payload

el3_payload.bin is a test image exercising the alternative EL3 payload boot flow in TF-A. Refer to the EL3 test payload README file for more details about its behaviour and how to build and run it.

4.4.   SPM test images

This repository contains 3 Secure Partitions that exercise the Secure Partition Manager (SPM) in TF-A [2]. Cactus-MM is designed to test the SPM implementation based on the ARM Management Mode Interface (MM), while Cactus and Ivy can test the SPM implementation based on the SPCI and SPRT draft specifications. Note that it isn't possible to use both communication mechanisms at once: If Cactus-MM is used Cactus and Ivy can't be used.

They run in Secure-EL0 and perform the following tasks:

  • Test that TF-A has correctly setup the secure partition environment: They should be allowed to perform cache maintenance operations, access floating point registers, etc.
  • Test that TF-A accepts to change data access permissions and instruction permissions on behalf of the Secure Partitions for memory regions the latter owns.
  • Test communication with SPM through either MM, or both SPCI and SPRT.

They are only supported on AArch64 FVP. They can be built independently of the other test images using the following command:

make PLAT=fvp cactus ivy cactus_mm

In the TF-A boot flow, the partitions replace the BL32 image and should be injected in the FIP image. To test SPM-MM with Cactus-MM, it is enough to use cactus_mm.bin as BL32 image. To test the SPM based on SPCI and SPRT, it is needed to use sp_tool to build a Secure Partition package that can be used as BL32 image.

To run the full set of tests in the Secure Partitions, they should be used in conjunction with the TFTF image.

For SPM-MM, the following commands can be used to build the tests:


# TF-A-Tests repository:

make PLAT=fvp TESTS=spm-mm tftf cactus_mm

# TF-A repository:

make BL33=path/to/tftf.bin BL32=path/to/cactus_mm.bin PLAT=fvp EL3_EXCEPTION_HANDLING=1 ENABLE_SPM=1 all fip

For SPM based on SPCI and SPRT:


# TF-A-Tests repository:

make PLAT=fvp TESTS=spm tftf cactus ivy

# TF-A repository:

make sptool

tools/sptool/sptool -o sp_package.bin
-i path/to/cactus.bin:path/to/cactus.dtb -i path/to/ivy.bin:path/to/ivy.dtb

make BL33=path/to/tftf.bin BL32=path/to/sp_package.bin PLAT=fvp ENABLE_SPM=1 SPM_MM=0 ARM_BL31_IN_DRAM=1 all fip

Please refer to the TF-A User guide for further details.

4.5.   Summary of build options

As much as possible, TF-A Tests dynamically detect the platform hardware components and available features. There are a few build options to select specific features where the dynamic detection falls short. This section lists them.

Unless mentioned otherwise, these options are expected to be specified at the build command line and are not to be modified in any component makefiles.

Note that the build system doesn't track dependencies for build options. Therefore, if any of the build options are changed from a previous build, a clean build must be performed.

4.5.1.   Build options shared across test images

Most of the build options listed in this section apply to TFTF, the FWU test images and Cactus, unless otherwise specified. These do not influence the EL3 payload, whose simplistic build system is mostly independent.

  • ARCH: Choose the target build architecture for TF-A Tests. It can take either aarch64 or aarch32 as values. By default, it is defined to aarch64. Not all test images support this build option.

  • ARM_ARCH_MAJOR: The major version of Arm Architecture to target when compiling TF-A Tests. Its value must be numeric, and defaults to 8.

  • ARM_ARCH_MINOR: The minor version of Arm Architecture to target when compiling TF-A Tests. Its value must be a numeric, and defaults to 0.

  • DEBUG: Chooses between a debug and a release build. A debug build typically embeds assertions checking the validity of some assumptions and its output is more verbose. The option can take either 0 (release) or 1 (debug) as values. 0 is the default.

  • ENABLE_ASSERTIONS: This option controls whether calls to assert() are compiled out.

    • For debug builds, this option defaults to 1, and calls to assert() are compiled in.
    • For release builds, this option defaults to 0 and calls to assert() are compiled out.

    This option can be set independently of DEBUG. It can also be used to hide any auxiliary code that is only required for the assertion and does not fit in the assertion itself.

  • LOG_LEVEL: Chooses the log level, which controls the amount of console log output compiled into the build. This should be one of the following:


    All log output up to and including the selected log level is compiled into the build. The default value is 40 in debug builds and 20 in release builds.

  • PLAT: Choose a platform to build TF-A Tests for. The chosen platform name must be a subdirectory of any depth under plat/, and must contain a platform makefile named platform.mk. For example, to build TF-A Tests for the Arm Juno board, select PLAT=juno.

  • V: Verbose build. If assigned anything other than 0, the build commands are printed. Default is 0.

4.5.2.   TFTF build options

  • ENABLE_PAUTH: Boolean option to enable ARMv8.3 Pointer Authentication (ARMv8.3-PAuth) support in the Trusted Firmware-A Test Framework itself. If enabled, it is needed to use a compiler that supports the option -msign-return-address. It defaults to 0.

  • NEW_TEST_SESSION: Choose whether a new test session should be started every time or whether the framework should determine whether a previous session was interrupted and resume it. It can take either 1 (always start new session) or 0 (resume session as appropriate). 1 is the default.

  • TESTS: Set of tests to run. Use the following command to list all possible sets of tests:

    make help_tests

    If no set of tests is specified, the standard tests will be selected (see tftf/tests/tests-standard.xml).

  • USE_NVM: Used to select the location of test results. It can take either 0 (RAM) or 1 (non-volatile memory like flash) as test results storage. Default value is 0, as writing to the flash significantly slows tests down.

4.5.3.   FWU test images build options

  • FIRMWARE_UPDATE: Whether the Firmware Update test images (i.e. NS_BL1U and NS_BL2U) should be built. The default value is 0. The platform makefile is free to override this value if Firmware Update is supported on this platform.

5.   Checking source code style

When making changes to the source for submission to the project, the source must be in compliance with the Linux style guide. To assist with this, the project Makefile provides two targets, which both utilise the checkpatch.pl script that ships with the Linux source tree.

To check the entire source tree, you must first download copies of checkpatch.pl, spelling.txt and const_structs.checkpatch available in the Linux master tree scripts directory, then set the CHECKPATCH environment variable to point to checkpatch.pl (with the other 2 files in the same directory).

Then use the following command:

make CHECKPATCH=<path-to-linux>/linux/scripts/checkpatch.pl checkcodebase

To limit the coding style checks to your local changes, use:

make CHECKPATCH=<path-to-linux>/linux/scripts/checkpatch.pl checkpatch

By default, this will check all patches between origin/master and your local branch. If you wish to use a different reference commit, this can be specified using the BASE_COMMIT variable.

6.   Running the TF-A Tests

Refer to the sections Running the software on FVP and Running the software on Juno in TF-A User Guide. The same instructions mostly apply to run the TF-A Tests on those 2 platforms. The difference is that the following images are not needed here:

  • Normal World bootloader. The TFTF replaces it in the boot flow;
  • Linux Kernel;
  • Device tree;
  • Filesystem.

In other words, only the following software images are needed:

  • BL1 firmware image;
  • FIP image containing the following images:
    • BL2;
    • SCP_BL2 if required by the platform (e.g. Juno);
    • BL31;
    • BL32 (optional);
    • tftf.bin (standing as the BL33 image).

6.1.   Running the manual tests on FVP

The manual tests rely on storing state in non-volatile memory (NVM) across reboot. On FVP the NVM is not persistent across reboots, so the following flag must be used to write the NVM to a file when the model exits.

-C bp.flashloader0.fnameWrite=[filename]

To ensure the model exits on shutdown the following flag must be used:

-C bp.ve_sysregs.exit_on_shutdown=1

After the model has been shutdown, this file must be fed back in to continue the test. Note this flash file includes the FIP image, so the original fip.bin does not need to be passed in. The following flag is used:

-C bp.flashloader0.fname=[filename]

6.2.   Running the FWU tests

As previously mentioned in section Putting it all together, there are a couple of extra images involved when running the FWU tests. They need to be loaded at the right addresses, which depend on the platform.

6.2.1.   FVP

In addition to the usual BL1 and FIP images, the following extra images must be loaded:

  • NS_BL1U image at address 0x0BEB8000 (i.e. NS_BL1U_BASE macro in TF-A)
  • FWU_FIP image at address 0x08400000 (i.e. NS_BL2U_BASE macro in TF-A)
  • Backup FIP image at address 0x09000000 (i.e. FIP_BKP_ADDRESS macro in TF-A tests).

An example script is provided in scripts/run_fwu_fvp.sh.

6.2.2.   Juno

The same set of extra images and load addresses apply for Juno as for FVP.

The new images must be programmed in flash memory by adding some entries in the SITE1/HBI0262x/images.txt configuration file on the Juno SD card (where x depends on the revision of the Juno board). Refer to the Juno Getting Started Guide, section 2.3 "Flash memory programming" for more information. Users should ensure these do not overlap with any other entries in the file.

Addresses in this file are expressed as an offset from the base address of the flash (that is, 0x08000000).

NOR10UPDATE: AUTO                       ; Image Update:NONE/AUTO/FORCE
NOR10ADDRESS: 0x00400000                ; Image Flash Address
NOR10FILE: \SOFTWARE\fwu_fip.bin        ; Image File Name
NOR10LOAD: 00000000                     ; Image Load Address
NOR10ENTRY: 00000000                    ; Image Entry Point

NOR11UPDATE: AUTO                       ; Image Update:NONE/AUTO/FORCE
NOR11ADDRESS: 0x03EB8000                ; Image Flash Address
NOR11FILE: \SOFTWARE\ns_bl1u.bin        ; Image File Name
NOR11LOAD: 00000000                     ; Image Load Address
NOR11ENTRY: 00000000                    ; Image Load Address

NOR12UPDATE: AUTO                       ; Image Update:NONE/AUTO/FORCE
NOR12ADDRESS: 0x01000000                ; Image Flash Address
NOR12FILE: \SOFTWARE\backup_fip.bin     ; Image File Name
NOR12LOAD: 00000000                     ; Image Load Address
NOR12ENTRY: 00000000                    ; Image Entry Point

[1]Therefore, the Trusted Board Boot feature must be enabled in TF-A for the FWU test images to work. Please refer the TF-A User guide for further details.
[2]Therefore, the Secure Partition Manager must be enabled in TF-A for any of the test Secure Partitions to work. Please refer to the TF-A User guide for further details.

Copyright (c) 2018-2019, Arm Limited. All rights reserved.