tools: Pass device part number and rev as args.

[iabdalkader]

Just a proof of concept.

The proposed fix here allows the following:

• Default behavior, load global-cfg.db
• Pass part/rev, which uses the same mechanisms to load the part info.

The part/rev can be set in mpconfigboard.mk, which is very consistent with other ports.

[dpgeorge]

It would be good if all the tools took lots of command line arguments, to specify all the things (part number, revision, file for ToC configuration, etc).

But, the way I was trying to do things (eg with #5) is to keep the changes minimal, and reuse the existing "Alif way" of having config files that specify everything. I know it's not very clean or that easy to use, but if we need to keep rebasing these patches on new Alif releases, then we don't want to make changes that are too big. Otherwise it's a lot of effort to update the toolkit from Alif.

OTOH, if Alif are not interested in improving this tool, and their updates are minor and few, we could do a lot more here to improve the ergonomics of the tools.

[iabdalkader]

to keep the changes minimal, and reuse the existing "Alif way" of having config files that specify everything.

Actually #5 makes more changes, and also adds args, but I feel that many changes have already been made, so maybe we're beyond that and may as well fix this issue in a way that works for us? Note that I kept a fallback to reading the config from file, unchanged, so both options are available. And if part/rev are specified, I use the exiting functions to load them.

The main issue I have with #5 is that there's no way to specify the db file, which I would like to do to avoid adding db files to board files, and then possibly have to also duplicate json config files. So even if we go with #5, we may need yet another option to specify db file name, and add part-specific db files here to this repo.

[dpgeorge]

Actually #5 makes more changes, and also adds args

Well, most of the changes there are in the new paths.py file. And also this PR here is not yet complete so must add more args and code to all the other tools. So in the end I think this one will be bigger.

The main issue I have with #5 is that there's no way to specify the db file, which I would like to do to avoid adding db files to board files

Yes, that would be nice, to specify db files. But then you won't need the changes in this PR, instead you'd add global-cfg-ae3.db etc files, and specify the file you want, instead off all these options for part/rev/jtag/mram.

In that case we should also go ahead and change how the app-device-config.json file is found, and make an option to specify which file to use?

[iabdalkader]

And also this PR here is not yet complete so must add more args and code to all the other tools. So in the end I think this one will be bigger.

It is now complete. However, that's exactly my point: we've already added code and args to all tools, so may as well make this change too.

Yes, that would be nice, to specify db files. But then you won't need the changes in this PR, instead you'd add global-cfg-ae3.db etc files, and specify the file you want, instead off all these options for part/rev/jtag/mram.

There are 3 different approaches/possible fix/improvements here:

1- Allow specifying the global-cfg.db file with an arg.
2- Allow specifying the config dir, where *.dbs can be found .
3- Allow bypassing global-cfg.db by passing part/rev via args

Option 3:

• I think is the most consistent with MicroPython's style (specifying via mpconfigboard.mk and passing them as args from Makefile).
• Requires no duplicated db files (not in toolkit nor in board files).
• Disadvantage: more args/slightly more changes, but as I've mentioned we're beyond this point I think.

Those are not mutually exclusive though, 1&2 are not really needed, but you can still support them if you want.

In that case we should also go ahead and change how the app-device-config.json file is found, and make an option to specify which file to use?

This is specified in the AToC file, and already works very well. Why change it though?

[dpgeorge]

Option 3 is good, let's go with that (and can improve/extend later if needed).

This is specified in the AToC file, and already works very well. Why change it though?

So that it's possible to use the app-device-config-ae3.json file that you added here, without needing to copy it into the board folder.

[iabdalkader]

So that it's possible to use the app-device-config-ae3.json file that you added here, without needing to copy it into the board folder.

But this is already possible via "binary": "app-device-config-ae3.json" in mcu/M55_x_cfg.json`. This is the file that needs to be duplicated, but I actually have a fix for this that will not require duplicating all 3 files.

The fix: we can just run a single json file through CPP, and enable images conditionally. This way only a single file is needed per board, but we could also define this "binary": "app-device-config.json", through a config option.

[dpgeorge]

The fix: we can just run a single json file through CPP, and enable images conditionally. This way only a single file is needed per board, but we could also define this "binary": "app-device-config.json", through a config option.

Just be careful of adding too much complexity, and complexity in lots of different direction. Because if that file is generated it must go in the build-xxx/ directory, and then that directory must be passed to the Alif tools as the input config directory.

And then if one config file is generated, why not generate them all from options set in mpconfigboard.mk?

[iabdalkader]

Because if that file is generated it must go in the build-xxx/ directory, and then that directory must be passed to the Alif tools as the input config directory.

This is the app-cfg file which is already passed to the tools in -f arg. It's not going to be complex, it's just a single run through CPP just like the linker script. I'll send a PoC in a bit.

[dpgeorge]

Tested with an AE3. Works well.