Component: flash parameter storage on stm32h7. Fixes #15331.

[taylornelms15]

As per the discussion in #15331, fixed issue where stm32h7 chips
use hardware ECC bits in program memory that disallow overwriting
32-byte flash line that has already been written. As such,
this change allows for a variant implementation of the flashfs system
that uses more space in the flash entry header in order to
allow an entire line to be reserved for erasing an entry.

Signed-off-by: Taylor Nelms [email protected]

Describe problem solved by this pull request

Fixed #15331 allowing for onboard-flash parameter storage on STM32H7 chips

Describe your solution

I added padding to the flash entry structure such that "erasing" an entry could be done by writing an entire flash line, avoiding the issue with hardware ECC bits causing issues in overwriting the flag on each entry.

Describe possible alternatives

I looked into other flash filesystems as potential drop-in alternatives, but did not find any that, out-of-the-box, met the limitations of the STM32H7.

Test data / coverage

Tests were admittedly minimal; application target was custom board for Robotic Research, LLC applications. Parameter reading/writing confirmed functional across reboots when configured to use two or one sectors of flash program memory.

Additional context

No additional context.

[dagar]

FYI @spiderkeys @taileron

[davids5]

Thank you @taylornelms15 I am OOO until 7/5 I will look at it as soon as I can when I get back

[davids5]

@taylornelms15 - This looks good. I raised a minor nit in naming to match the reference manual.

What testing was done?

[taylornelms15]

Also @davids5 let me know if there's anything I'm messing up procedurally with this request, I haven't gone through the pull request process in the past so I'm willing to take notes on process mistakes.

[davids5]

@taylornelms15 - you have some CI failures

../../src/lib/parameters/flashparams/flashfs32.c:716:16: error: 'FLASH_FLAG_LINE_SIZE' undeclared (first use in this function); did you mean 'FLASH_FLAG_ROW_SIZE'?
  716 |  h_flag_t data[FLASH_FLAG_LINE_SIZE] = {0xffff};
      |                ^~~~~~~~~~~~~~~~~~~~
      |                FLASH_FLAG_ROW_SIZE

Once you fix it. Please rebase on main and fixup the commits
git checkout main
git pull <remote px4> main
git checkout pr-flash-params-32byte-aligned

git rebase i main

Then change the pick to f for the all the commits not including the top one.

then
git push -f <remote for taylornelms15> pr-flash-params-32byte-aligned

This will fixup the commits. making one clean commit with the comment from 5813198

[davids5]

@taylornelms15 - will you be following up on this?

[taylornelms15]

@davids5 I thought I put the edits into the code and pushed it; did it not go through? I'm happy to follow up on whatever parts are missing.

[davids5]

@taylornelms15 - Sorry my bad. I missed the last commit. Let see how CI likes it.

[taylornelms15]

I can't tell from the logs whether the continuous-integration/jenkins/pr-head CI failure is something on my end or not; definitely let me know if I can make any other modifications to close this thing out

[taylornelms15]

@davids5 would love to wrap this up so we can use it in actual environments; is there anything else I should change? I can't tell if something on my end is blocking continuous-integration/jenkins/pr-head

[davids5]

Having a build using the Flash parameters on an H7 and run on the test rack would help test this. There have been some issues with the test rack that are unrelated. @dagar - do you have a way of testing this?

[matthewoots]

@taylornelms15 Thank you for your fix on the H743 flash params I've tested the flash32.c and your commit works for my H743 board, in fact I tried on an H7 AIO board without an SD card slot and it works like a charm using Matek H743-slim board target as a base. Can this be pushed to master?