Ghostfat test boards

[henrygab]

The goal of this branch is to enable, as part of the normal build process, validation of GhostFat image generation.

• Create framework to execute own custom test_main() function, using existing code
• Ensure ghostfat initialization completes prior to calling test_main()
• Generate disk image file from test_main()
• Validate bit-for-bit accuracy of disk image file from test_main()
• Use common framework code for variations of desired file system image
• Check in known-good file system image(s) to validate against
• Double-compress known-good disk images
• Ensure CI catches failures

NOTE: When the ghostfat image that is generated would need to change, such as for a bugfix, then updated versions of the known-good file system images would also need to also be committed.

[henrygab]

Collapsing to reduce thread length

@hathach -- Can you help with two things?

1. Getting this to compile in the CI builds
2. Getting the compiled , and ensuring CI will execute the resulting binary.

Compilation:

On my Ubuntu system, to make the test ports compile requires removing two options from make.mk and rules.mk:

tinyuf2/ports/make.mk

Lines 136 to 137 in 54cff8b

	-specs=nosys.specs \
	-specs=nano.specs \

tinyuf2/ports/rules.mk

Line 12 in 54cff8b

LIBS += -lgcc -lm -lnosys -lc

[hathach]

Collapsing to reduce thread length
@hathach -- Can you help with two things?

1. Getting this to compile in the CI builds
2. Getting the compiled , and ensuring CI will execute the resulting binary.

Compilation:

On my Ubuntu system, to make the test ports compile requires removing two options from make.mk and rules.mk:

tinyuf2/ports/make.mk

Lines 136 to 137 in 54cff8b

	-specs=nosys.specs \
	-specs=nano.specs \

tinyuf2/ports/rules.mk

Line 12 in 54cff8b

LIBS += -lgcc -lm -lnosys -lc

no problems at all, leave that to me. I just resolved the conflict with master and push the update. Give me a bit of time to go through the code. I think I could setup unit tests with CMock/Unity using your tests. That would be easier to run on ci and a bit cleaner look for the ports. Will post more update.

Update: I just realized this is 6 days ago, hmm, the collapsible probably have me miss your message earlier 🤕

[henrygab]

@hathach -- Yes, it was a few days back. Since then, I got it to work with CI.

So, you can merge as-is, no work, and get GhostFAT validation.

[henrygab]

OK, this is ready for review. See success in checks from commit deae8ca ... listing the four GhostFAT code compiling, comparing against known-good images, and even ignoring the two very narrow exceptions required.

Two exceptions detailed

Two variations allowed from the known good file system:

1. Contents of UF2_INFO.TXT file will have a variable-length string in the middle that differs for each commit.
2. The directory entry for UF2_INFO.TXT may indicate a different file size.

The code allows very narrow exceptions with multiple checks, to allow these two differences, without allowing other regressions to fall through.

[hathach]

Brilliant !! Thank you very much for your effort and time put into this PR. Though I still think we should separate the native build from current build system for unit test cmock/unity. I will merge this now and make a follow up PR to decouple the native build. Thank you very much for your awesome work !!!!

[henrygab]

Sorry, please ignore this "review request" ... accidental click.

[hathach]

no problem at all :)