Hello,
The agenda for the forum:
1. Interrupt handling in PSA FF-M v1.1
2. Ongoing open issues, discussed on the Forum
3. AOB
See you,
Anton
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org> On Behalf Of Anton Komlev via TF-M
Sent: 09 October 2020 11:11
To: tf-m(a)lists.trustedfirmware.org
Cc: nd <nd(a)arm.com>
Subject: [TF-M] TF-M Technical Forum call - October 15
Hello,
The next Technical Forum is planned on Thursday, October 15 at 6:00-07:00 UTC (Asia time zone).
Please reply on this email with your proposals for agenda topics.
Recording and slides of previous meetings are here:
https://www.trustedfirmware.org/meetings/tf-m-technical-forum/
Best regards,
Anton
Hi, Raef,
I can get past the CMSIS issue by dup'ing the GNU links. It now fails for lack of the correct Python.
Using the -G option, CMakeCache.txt lists all the pythons in my system and the build completes:
$ grep python cmake_build/CMakeCache.txt
PYTHON_EXECUTABLE:FILEPATH=C:/Python/Python27/python.exe
FIND_PACKAGE_MESSAGE_DETAILS_Python3:INTERNAL=[C:/Users/cXXXXX/AppData/Local/Programs/Python/Python38-32/python.exe][cfound components: Interpreter ][v3.8.5()]
_Python3_EXECUTABLE:INTERNAL=C:/Users/cXXXXX/AppData/Local/Programs/Python/Python38-32/python.exe
Without the -G option, the compiler is MSVC, and the cache has no entry for python at all.
Building Custom Rule C:/Users/cXXXXX/Git/arm/TF-M/trusted-firmware-m/cmake_build/lib/ext/mcuboot-subbuild/CMakeLists.txt
Building Custom Rule C:/Users/cXXXXX/Git/arm/TF-M/trusted-firmware-m/cmake_build/lib/ext/mcuboot-subbuild/CMakeLists.txt
-- Could NOT find Python3 (missing: Python3_EXECUTABLE Interpreter)
Reason given by package:
Interpreter: Wrong major version for the interpreter "C:/Python/Python27/python.exe"
Hi, Raef,
The combination that worked was the most-recent commit and gnu tools (-G option). Using VS it fails at lib/ext/CMSIS_5/CMakeLists.txt:25 for lack of CMSIS RTX static libraries.
...this command worked....
$ cmake -S . -B cmake_build -DTFM_PLATFORM=mps2/an521 -DCMAKE_TOOLCHAIN_FILE=toolchain_GNUARM.cmake -G"Unix Makefiles"
-- The C compiler identification is GNU 9.3.1
-- The ASM compiler identification is GNU
-- Found assembler: C:/Program Files (x86)/GNU Arm Embedded Toolchain/9 2020-q2-update/bin/arm-none-eabi-gcc.exe
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: C:/Program Files (x86)/GNU Arm Embedded Toolchain/9 2020-q2-update/bin/arm-none-eabi-gcc.exe - skipped
-- Detecting C compile features
-- Detecting C compile features - done
Thanks. I'm interested if you've managed to build TFM using the visual studio generator? That check was actually added because we had a problem with at least one of the visual studio generators (VS10) setting the C compiler to `MSVC`. If you've managed to get it to build with VS16, then we can look in to adding that as a known good generator.
Raef
________________________________________
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org> on behalf of Kevin Kilzer via TF-M <tf-m(a)lists.trustedfirmware.org>
Sent: 13 October 2020 15:30
To: tf-m(a)lists.trustedfirmware.org
Subject: Re: [TF-M] Following the TF-M build example
Hi, Raef,
I added to the generator qualifier in CMakeLists.txt line 17. Since cmake is not my native language, I put in the full string, as you see. The string comes from the display of CMAKE_GENERATOR in the error message.
if(NOT ${CMAKE_GENERATOR} STREQUAL "Unix Makefiles" AND
NOT ${CMAKE_GENERATOR} STREQUAL "Visual Studio 16 2019" AND
NOT ${CMAKE_GENERATOR} STREQUAL "Ninja")
Message(FATAL_ERROR "Unsupported generator ${CMAKE_GENERATOR}. Hint: Try -G\"Unix Makefiles\"")
endif()
Hi, Raef,
I added to the generator qualifier in CMakeLists.txt line 17. Since cmake is not my native language, I put in the full string, as you see. The string comes from the display of CMAKE_GENERATOR in the error message.
if(NOT ${CMAKE_GENERATOR} STREQUAL "Unix Makefiles" AND
NOT ${CMAKE_GENERATOR} STREQUAL "Visual Studio 16 2019" AND
NOT ${CMAKE_GENERATOR} STREQUAL "Ninja")
Message(FATAL_ERROR "Unsupported generator ${CMAKE_GENERATOR}. Hint: Try -G\"Unix Makefiles\"")
endif()
Note that the aforementioned patch has now been merged - windows build should now be working again on master
Raef
________________________________________
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org> on behalf of Raef Coles via TF-M <tf-m(a)lists.trustedfirmware.org>
Sent: 13 October 2020 10:24
To: tf-m(a)lists.trustedfirmware.org; Kevin.Kilzer(a)microchip.com
Subject: Re: [TF-M] Following the TF-M build example
Hi,
I'm interested in the changes that you made to the validity checks, would you mind sending a patch / outlining what you had to change. The windows generator checks are still not working exactly as they should and I'd like to know what your experience was.
For the build failure, I believe this might be related to an issue with the windows PSA file generation. We've got a patch in review for this at https://review.trustedfirmware.org/c/TF-M/trusted-firmware-m/+/6386, can you test and see if that fixes the problem.
Raef
________________________________________
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org> on behalf of Kevin Kilzer via TF-M <tf-m(a)lists.trustedfirmware.org>
Sent: 13 October 2020 00:20
To: tf-m(a)lists.trustedfirmware.org
Subject: Re: [TF-M] Following the TF-M build example
Thanks for the notes. Since last week I’ve now:
1. downloaded today’s latest repo (12 October, commit 8bebd05745a8b27dccc6403f0215fa6e39de3bc1, and
2. added the VS compiler to the “valid” checks at CmakeLists.txt line 17.
Using the -G option allows the make to complete (apparently), but the install script fails (in both GitBash and CMD).
Thanks for any help.
==========
-- Build files have been written to: C:/Users/cXXXXX/Git/arm/TF-M/trusted-firmware-m/cmake_build
cXXXXX@LT-cXXXXXA MINGW64 ~/Git/arm/TF-M/trusted-firmware-m (master)
$ cmake --build cmake_build -- install
tools/CMakeFiles/tfm_generated_files.dir/build.make:93: *** target pattern contains no '%'. Stop.
CMakeFiles/Makefile2:944: recipe for target 'tools/CMakeFiles/tfm_generated_files.dir/all' failed
make.exe[1]: *** [tools/CMakeFiles/tfm_generated_files.dir/all] Error 2
Makefile:148: recipe for target 'all' failed
make.exe: *** [all] Error 2==========
--
TF-M mailing list
TF-M(a)lists.trustedfirmware.org
https://lists.trustedfirmware.org/mailman/listinfo/tf-m
Hi,
I'm interested in the changes that you made to the validity checks, would you mind sending a patch / outlining what you had to change. The windows generator checks are still not working exactly as they should and I'd like to know what your experience was.
For the build failure, I believe this might be related to an issue with the windows PSA file generation. We've got a patch in review for this at https://review.trustedfirmware.org/c/TF-M/trusted-firmware-m/+/6386, can you test and see if that fixes the problem.
Raef
________________________________________
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org> on behalf of Kevin Kilzer via TF-M <tf-m(a)lists.trustedfirmware.org>
Sent: 13 October 2020 00:20
To: tf-m(a)lists.trustedfirmware.org
Subject: Re: [TF-M] Following the TF-M build example
Thanks for the notes. Since last week I’ve now:
1. downloaded today’s latest repo (12 October, commit 8bebd05745a8b27dccc6403f0215fa6e39de3bc1, and
2. added the VS compiler to the “valid” checks at CmakeLists.txt line 17.
Using the -G option allows the make to complete (apparently), but the install script fails (in both GitBash and CMD).
Thanks for any help.
==========
-- Build files have been written to: C:/Users/cXXXXX/Git/arm/TF-M/trusted-firmware-m/cmake_build
cXXXXX@LT-cXXXXXA MINGW64 ~/Git/arm/TF-M/trusted-firmware-m (master)
$ cmake --build cmake_build -- install
tools/CMakeFiles/tfm_generated_files.dir/build.make:93: *** target pattern contains no '%'. Stop.
CMakeFiles/Makefile2:944: recipe for target 'tools/CMakeFiles/tfm_generated_files.dir/all' failed
make.exe[1]: *** [tools/CMakeFiles/tfm_generated_files.dir/all] Error 2
Makefile:148: recipe for target 'all' failed
make.exe: *** [all] Error 2==========
Hi,
While digging the clean issue brought up by Soby, I started wondering if external dependency handling would be better in a slightly different way. There is a lot of hack in the build system around including the psa-arch test project, mostly to work around cmake limitations on namespaces and symbol separation. A stronger barrier could eliminate the mess. In TS we use the following pattern (let's call it "Internal Project"):
* External dependencies are fetched with fetch_content()
* Right after the fetch, execute_process() is called to start the build of the component. So external component builds configuration time.
* The project get's installed into a directory and the main project is using the installed content, possibly through find_package().
* Benefits:
* This gives a stronger separation, elimination any name clash between the main project and the external dependency. Also global settings cannot collide like when a dependency sets CMAKE_BUILD_TYPE.
* Faster main project build times, as external projects are only built once.
* Makes it more "natural" to use an externally built binary for an external component. This might be handy from QA perspective if binary releases are going to happen. (If ever of course.)
* Strong separation could allow using different version of the same tools for components. (i.e. main project is built with GCC, component with IAR.)
* Drawbacks:
* It is harder to develop the external component together with tf-m s tracking changes is more difficult. Might be a problem if debugging tf-m vs external component interaction. This should be rare and might be an acceptable issue.
* It is unnatural to run builds configuration time in cmake world.
* Configuration phase will take longer.
* Since the build happens right where the external component is added (point A), cmake execution flow might need to be different to ensure all information needed to configure the external component is present at point A.
* Since external component is built by a separated cmake run, tool detection happens separate. This means the same tools will be searched for multiple times. Initial cache files can be a workaround.
This is very similar to how external projects work in cmake, but makes better integration possible. The main project can use information from the dependency as it's source and output files become available configuration time. In turn external project changes are harder to track.
/George
Hi,
I tried to dig deeper into this, but the cmake command used by Soby fails for me.
"
[ 33%] Performing patch step for 'psa_arch_tests-populate'
error: patch failed: api-tests/platform/targets/tgt_dev_apis_tfm_an521/nspe/pal_driver_intf.c:128
...
"
It would be nice to understand why cmake fails to clean properly, but well, I cannot deep-dive due to the above. What I wanted to check is if https://cmake.org/cmake/help/v3.15/prop_tgt/ADDITIONAL_CLEAN_FILES.html could be used to get "make clean" remove the psa-arch test binary directory. But:
* I am not sure which build directory is used. I have the feeling we use ${CMAKE_CURRENT_BINARY_DIR}/psa_api_tests and not psa_arch_tests_BINARY_DIR, which would be build\lib\ext\psa_arch_tests-build. Strange.
* ADDITIONAL_CLEAN_FILES was introduced in cmake v3.15 and if my memories are correct tf-m allows an older version if not using IAR.
Soby: can you please test if ADDITIONAL_CLEAN_FILES works? This solution would give a more streamlined.
/George
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org> On Behalf Of Anton Komlev via TF-M
Sent: 12 October 2020 23:00
To: tf-m(a)lists.trustedfirmware.org
Cc: nd <nd(a)arm.com>
Subject: Re: [TF-M] New TF-M Build doesn't track config changes
Hi,
Thanks Soby for sorting it out.
Sounds like a right way to go and cleanall shall do that job.
For me it looks like an exceptional case while the main scenario for a daily development shall be the one, described by Karl : downloaded dependencies explicitly specified by paths outside of TF-M tree via command line, or via project config file (suggested).
And true, both cases shall be explicitly documented.
Hope it helps,
Anton
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org<mailto:tf-m-bounces@lists.trustedfirmware.org>> On Behalf Of Soby Mathew via TF-M
Sent: 12 October 2020 17:29
To: Gyorgy Szing <Gyorgy.Szing(a)arm.com<mailto:Gyorgy.Szing@arm.com>>
Cc: nd <nd(a)arm.com<mailto:nd@arm.com>>; tf-m(a)lists.trustedfirmware.org<mailto:tf-m@lists.trustedfirmware.org>
Subject: Re: [TF-M] New TF-M Build doesn't track config changes
Thanks Karl, Gyorgy for your inputs.
I agree with the principle that `BUILD` shall be only folder the cmake modifies. The trouble is, after a `make clean`, there are still artefacts from the previous configuration which affects the new build and gives the wrong output. Hence the suggestion to introduce a `cleanall` custom target which endeavours to clean the all the remnant config information from previous build and leave auto cloned dependant repositories untouched (or maybe print some status info).
Does that sound like a good plan then ?
Best Regards
Soby Mathew
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org<mailto:tf-m-bounces@lists.trustedfirmware.org>> On Behalf Of Gyorgy Szing via TF-M
Sent: 12 October 2020 06:29
To: tf-m(a)lists.trustedfirmware.org<mailto:tf-m@lists.trustedfirmware.org>
Cc: nd <nd(a)arm.com<mailto:nd@arm.com>>
Subject: Re: [TF-M] New TF-M Build doesn't track config changes
Hi,
I think the build directory is owned by cmake and the same rules shall apply to all files there. Also the only directory cmake does modify without the users consent shall be the build directory. As long as cmake is owning the external dependencies it is the responsibility of the build system to keep the dependencies in a healthy state and to ensure the correct version is built. To do this safely the "clean" operation, which is used to get out of a "non-operational" state, shall fix the dependencies too.
So the correct operation (in my opinion) is to make the dependency download work in the following way:
* If the dependency is already present at the target location, cmake shall use it as is. Possibly some status information should be printed (i.e. version number, if the git working copy is dirty etc...)
* If not cmake shall do the fetch.
This way if the user specify an external location (one not in the build directory), cmake will "export" the dependency when the first fetch is done, and do no modifications after. This gives us a well-defined act of handing over the responsibility of keeping the dependency clean.
As far as I can see (was not digging into the details) this more or less matches how the current implementation works, and what is missing is more details in the documentation.
/George
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org<mailto:tf-m-bounces@lists.trustedfirmware.org>> On Behalf Of Karl Zhang via TF-M
Sent: 10 October 2020 08:14
To: tf-m(a)lists.trustedfirmware.org<mailto:tf-m@lists.trustedfirmware.org>; Soby Mathew <Soby.Mathew(a)arm.com<mailto:Soby.Mathew@arm.com>>
Cc: nd <nd(a)arm.com<mailto:nd@arm.com>>
Subject: Re: [TF-M] New TF-M Build doesn't track config changes
Hi Soby,
I met the same problem before, and I think your suggestions are helpful. There might be more check needed if the 'make clean' does not delete the auto cloned repos. Because the dependencies may update by a new TFM commit.
The new build system supports to specify the patch of each dependency, which can avoid clone automatically to the build folder each time. Hope it can mitigate the inconvenient scenario.
-DMBEDCRYPTO_PATH=
-DTFM_TEST_REPO_PATH=
-DMCUBOOT_PATH=
-DPSA_ARCH_TESTS_PATH=
There is an example from CI for build command:
https://ci.trustedfirmware.org/view/TF-M/job/tf-m-build-config/lastStableBu…
cmake -DTFM_PLATFORM=mps2/an519 -DCMAKE_TOOLCHAIN_FILE=/home/buildslave/workspace/tf-m-build-config/trusted-firmware-m/toolchain_GNUARM.cmake -DTFM_PSA_API=True -DTFM_ISOLATION_LEVEL=1 -DTEST_NS=False -DTEST_S=False -DTEST_PSA_API=OFF -DCMAKE_BUILD_TYPE=Debug -DCRYPTO_HW_ACCELERATOR_OTP_STATE=False -DBL2=False -DNS=False -DTFM_TEST_REPO_PATH=/home/buildslave/workspace/tf-m-build-config/trusted-firmware-m/../tf-m-tests -DMBEDCRYPTO_PATH=/home/buildslave/workspace/tf-m-build-config/trusted-firmware-m/../mbedtls -DPSA_ARCH_TESTS_PATH=/home/buildslave/workspace/tf-m-build-config/trusted-firmware-m/../psa-arch-tests -DMCUBOOT_PATH=/home/buildslave/workspace/tf-m-build-config/trusted-firmware-m/../mcuboot -DTFM_PROFILE= /home/buildslave/workspace/tf-m-build-config/trusted-firmware-m
BR
Karl
________________________________
From: TF-M <tf-m-bounces(a)lists.trustedfirmware.org<mailto:tf-m-bounces@lists.trustedfirmware.org>> on behalf of Soby Mathew via TF-M <tf-m(a)lists.trustedfirmware.org<mailto:tf-m@lists.trustedfirmware.org>>
Sent: Friday, October 2, 2020 8:40 PM
To: tf-m(a)lists.trustedfirmware.org<mailto:tf-m@lists.trustedfirmware.org> <tf-m(a)lists.trustedfirmware.org<mailto:tf-m@lists.trustedfirmware.org>>
Cc: nd <nd(a)arm.com<mailto:nd@arm.com>>
Subject: [TF-M] New TF-M Build doesn't track config changes
Hi,
With the new build system, TF-M downloads all dependant repositories to the BUILD folder as part of CMAKE configuration. CMake does provide `make clean` target to clean the build and rebuild but this doesn't track any config changes between the builds. Previously, the CMake workflow was to delete all contents in the BUILD folder before re-configuring for a new build but now this means the developer has to re-download all the dependant git repositories before the project can be build again. This can be a slow and cumbersome process for developers. Deleting the CMakecache.txt doesn't seem to solve the problem either.
The most obvious solution is to move the cloned repositories outside the BUILD/ folder , so we can follow the previous workflow of deleting the BUILD folder before re-build. The sample test sequence to reproduce the problem is given below:
/* Build secure regressions test suite */
$ make clean; cmake .. -DTFM_PLATFORM=mps2/an521 -DCMAKE_TOOLCHAIN_FILE=../toolchain_GNUARM.cmake -DTEST_S=ON -DCMAKE_BUILD_TYPE=Debug; make install
/* reconfigure and build PSA Crypto API test suite */
$ make clean; cmake .. -DTFM_PLATFORM=mps2/an521 -DCMAKE_TOOLCHAIN_FILE=../toolchain_GNUARM.cmake -DCMAKE_BUILD_TYPE=Debug -DTEST_PSA_API=CRYPTO; make install
Built binary is still for regression test suite.
There are multiple suggestions to solve the problem, One is to move the clone repositories outside the BUILD/ folder, or introduce a build target like `make cleanall` which does the equivalent of rm -rf !(lib) within the BUILD folder (i.e the build target will remove all folders except the lib folder which has the cloned repositories). Please let us know of your suggestions on this.
Best Regards
Soby Mathew
