Feature: Arterytek option bytes and more part IDs #1803

ALTracer commented 2024年04月19日 22:23:02 +02:00 (Migrated from github.com)

Detailed description

• This is a new feature.
• Currently there's no easy way to extend SRAM of AT32F403A via monitor option commands because EOPB0 is out of reach for the normal 16-byte driver.
• The PR solves it by conditionally bumping the option byte ranges.

Past first commit see my work of adding yet more part ID lists for other families in AT32F4 lineup. This is probably getting unmanageable, so I started incorporating F_SIZE and "Product ID" readouts from UID (in their DESig analog). If this is fine, e.g. does not result in problems for RDP1-enabled targets, then all such lists can be dropped or gated under ENABLE_DEBUG to not hog precious BMF flash -- it only needs to know flash parameters (banks, sector size, total size), not exact package & pinout. I may move this to a separate PR (and a split translation unit) as no users have probably asked for F421, F413 or F425. F423 and F402/F405 are plain unimplemented (F405 with USB HS PHY would be nice to have though). I have none of these chips to test on.

Your checklist for this pull request

• I've read the Code of Conduct
• I've read the guidelines for contributing to this repository
• It builds for hardware native (see Building the firmware) ...and overflows flash
• It builds as BMDA (see Building the BMDA)
• I've tested it to the best of my ability
• My commit messages provide a useful short description of what the commits do

Closing issues

Allows operating Arterytek User System Data, including Flash Access Protection enable/disable ops.

## Detailed description * This is a new feature. * Currently there's no easy way to extend SRAM of AT32F403A via `monitor option` commands because EOPB0 is out of reach for the normal 16-byte driver. * The PR solves it by conditionally bumping the option byte ranges. Past first commit see my work of adding yet more part ID lists for other families in AT32F4 lineup. This is probably getting unmanageable, so I started incorporating F_SIZE and "Product ID" readouts from UID (in their DESig analog). If this is fine, e.g. does not result in problems for RDP1-enabled targets, then all such lists can be dropped or gated under ENABLE_DEBUG to not hog precious BMF flash -- it only needs to know flash parameters (banks, sector size, total size), not exact package & pinout. I may move this to a separate PR (and a split translation unit) as no users have probably asked for F421, F413 or F425. F423 and F402/F405 are plain unimplemented (F405 with USB HS PHY would be nice to have though). I have none of these chips to test on. ## Your checklist for this pull request * [x] I've read the [Code of Conduct](https://github.com/blackmagic-debug/blackmagic/blob/main/CODE_OF_CONDUCT.md) * [x] I've read the [guidelines for contributing](https://github.com/blackmagic-debug/blackmagic/blob/main/CONTRIBUTING.md) to this repository * [x] It builds for hardware native (see [Building the firmware](https://github.com/blackmagic-debug/blackmagic?tab=readme-ov-file#building-black-magic-debug-firmware)) ...and overflows flash * [x] It builds as BMDA (see [Building the BMDA](https://github.com/blackmagic-debug/blackmagic?tab=readme-ov-file#building-black-magic-debug-app)) * [x] I've tested it to the best of my ability * [x] My commit messages provide a useful short description of what the commits do ## Closing issues Allows operating Arterytek User System Data, including Flash Access Protection enable/disable ops.

ALTracer commented 2024年09月29日 22:47:46 +02:00 (Migrated from github.com)

Rebased and clang-formatted.
I only wanted 48 option bytes for AT32F403A (especially RDP1 and EOPB0). Even that impl is bad -- it would affect GD32F1 etc. because of TOPT_32BIT criteria. The stm32f1-generic flash driver does not record option byte region size. If you have any better ideas -- I'll try to edit them in for a minor but effective change (no large driver-scope refactoring).
I would rather migrate/move the rest of chips/targets into at32f43x.c sometime in the future, but that should be made more generic as well to handle that. And it'd shrink the stm32f1.o (always compiled).

Rebased and clang-formatted. I only wanted 48 option bytes for AT32F403A (especially RDP1 and EOPB0). Even that impl is bad -- it would affect GD32F1 etc. because of TOPT_32BIT criteria. The stm32f1-generic flash driver does not record option byte region size. If you have any better ideas -- I'll try to edit them in for a minor but effective change (no large driver-scope refactoring). I would rather migrate/move the rest of chips/targets into at32f43x.c sometime in the future, but that should be made more generic as well to handle that. And it'd shrink the stm32f1.o (always compiled).

ALTracer commented 2024年10月02日 20:45:43 +02:00 (Migrated from github.com)

While I don't think it ends good for GD32 in current form, I'd like to ask to if(!strcmp(target->driver, "AT32")) instead of TOPT so both would work as expected.
Rebased away the addition of AT32F405/F402 (HS USB) because their FPEC is at 0x40023C00 like F435/F437, not 0x40022000 as this driver is hardcoded to work on. Same for AT32F423 (three series, FPEC at incompatible offset, unsupported here).

While I don't think it ends good for GD32 in current form, I'd like to ask to `if(!strcmp(target->driver, "AT32"))` instead of TOPT so both would work as expected. Rebased away the addition of AT32F405/F402 (HS USB) because their FPEC is at 0x40023C00 like F435/F437, not 0x40022000 as this driver is hardcoded to work on. Same for AT32F423 (three series, FPEC at incompatible offset, unsupported here).

dragonmux commented 2024年10月12日 02:04:55 +02:00

Once the single outstanding review note has been sorted, we're happy to get this merged. This is looking really good.

Once the single outstanding review note has been sorted, we're happy to get this merged. This is looking really good.

ALTracer commented 2024年10月12日 17:33:54 +02:00 (Migrated from github.com)

Replaced the option byte size dispatch check from TOPT 32-bit writes (shared with Gigadevices) to direct strncmp. This triggers correctly, 48 bytes (24 halfwords) for AT32F403A/407, but still 16 bytes (8 halfwords) for GD32F103 etc.
Did not register cmd_option for older F403 or newer F413, as I cannot test on them. Did not register it for F415 either, because it has 1024 not 48 bytes.
Added a few more families very likely compatible with STM32F1 flash driver. They all have UID/F_SIZE unreadable under RDP1, so need part ID tables. I started to notice a pattern "part ID <=> flash density", but the optimization would be handled in a follow-up PR, if ever.
Please review again.

Replaced the option byte size dispatch check from TOPT 32-bit writes (shared with Gigadevices) to direct strncmp. This triggers correctly, 48 bytes (24 halfwords) for AT32F403A/407, but still 16 bytes (8 halfwords) for GD32F103 etc. Did not register cmd_option for older F403 or newer F413, as I cannot test on them. Did not register it for F415 either, because it has 1024 not 48 bytes. Added a few more families very likely compatible with STM32F1 flash driver. They all have UID/F_SIZE unreadable under RDP1, so need part ID tables. I started to notice a pattern "part ID <=> flash density", but the optimization would be handled in a follow-up PR, if ever. Please review again.