Fix wolfBoot_update_trigger sector calculation

[MulattoKid]

wolfBoot_update_trigger is called to initiate an update on the next boot. It does this by modifying the metadata in the last sector (selecting one of the two if NVM_FLASH_WRITEONCE is defined).

There's a check to handle the case where the PART_UPDATE_ENDFLAGS isn't aligned with the WOLFBOOT_SECTOR_SIZE. However, this check was wrong, as the modulo of these two macros is 0 when aligned, so the check should be != 0.

See ticket #19133 for more information.

[MulattoKid]

I've tested this, and performing an update now works. Without this change the second and third to last sectors are chosen to work with, instead of the last and second to last sectors, resulting in an invalid SHA integrity check when our image spans the entire available region (excluding the two sectors reserved by wolfBoot at the end of the BOOT and UPDATE partitions).

I'm really curious to understand if I'm thinking correctly about this, and there's been a bug in this check for a long time, or if I'm misunderstanding something. AFAIK wolfBoot reserves the last two sectors (each of size WOLFBOOT_SECTOR_SIZE) of each partition, but from testing it overwrites more (as explained in the ticket).

Edit: I see there's a unit test failing, so there's something strange going on here...

[dgarske]

Unit test says:

83%: Checks: 6, Failures: 1, Errors: 0
unit-common.c:53:F:verify_integrity:test_verify_integrity:0: Assertion 'address + len <= (33 * 1024)' failed: address + len == 94631014433515, (33 * 1024) == 33792
make: *** [Makefile:37: run] Error 1

If the fix is valid it's possible a unit test may need updated. @jpbland1 please review.

[dgarske]

Closing in favor of #531

[MulattoKid]

Thanks for looking into it!

[MulattoKid]

Just noting that #531 seems to fix the issue