Supporting cores with private flash partitions

[maxd-nordic]

Hi! I'm looking into supporting the nRF5340 in pyocd.
Although it seems easy enough to have multiple flasher algos for different memory partitions, I couldn't find a way to specify on which core the algo needs to run. We would usually have a merged HEX file containing all the partitions that is supplied to the flasher tool. It would be great if someone could help me out here. :)

[flit]

Unfortunately… pyocd currently can only run the flash algo on the main core. ☹️ It's definitely a serious deficiency for certain devices!

[flit]

It looks like it wouldn't be too hard to implement. The pyocd.flash.flash.Flash class that's responsible for running the flash algo already takes a Target class used to access the core. If the target instance for a core (rather than SoC) is passed in, the algo would be run on that core.

The main changes would be

• Tracking the core the algo needs to on. Using a FlashRegion attribute with the core name (or number) would be fine. (Default of None needs to be added to FlashRegion.DEFAULT_ATTRS dict.)
• Update CoreSightTarget.create_flash() to look up the core and pass it to the Flash class constructor, or use the primary core if not specified.
• Update pyocd.target.pack.cmsis_pack.CmsisPackDevice to set the flash algo core name attribute.

[maxd-nordic]

Thanks for the insight! I'm currently tinkering with this on my develop branch.
Do you have an idea why

• the second core is not automatically generated
• only the first flash region is actually saved as a flash regions, while the others are ROM regions?

[maxd-nordic]

Got some more progress, here's a PR: #1656