2020-02-09 12:53 AM
It seems that H7A3ZI on Nucleo-H7A3ZI-Q is returning incorrect memory descriptor:
@Internal Flash /0x08000000/256*02Kg,@Option Bytes /0x1FF8000/01*128 e,@OTP Memory /0x1FF7000/01*1024 e
Which tells there are only 256 of 2K (512KB); it should have 2MB, right?
STM32CubeProgrammer is probably also complaining:
17:47:13:733 : received memory address is wrong or unsupported
17:47:13:734 : Status: errVENDOR, State: dfuERROR
17:47:13:734 : failed to read the requested memory content
17:47:13:734 : Could not read flash size at address 0x8fff80c
Any info on this?
2020-02-11 08:29 PM
Since H7A3 is capable of erasing at 8KB sector unit, I believe the correct descriptor for the internal flash would be
@Internal Flash /0x08000000/256*08Kg
instead of
@Internal Flash /0x08000000/256*02Kg
2020-02-18 06:28 AM
FYI, silicon revision for the problematic chip is 'Z'.
2020-05-12 11:19 AM
Another piece of information regarding DFU device descriptor for H7A.
There is an example project for H7A which implements standalone DFU in STM32Cube_FW_H7_V1.6.0, and the descriptor used is partially correct as it uses 8KB page instead of 2KB page from the embedded boot loader.
STM32Cube_FW_H7_V1.6.0/Projects/NUCLEO-H7A3ZI-Q/Applications/USB_Device/DFU_Standalone/USB_Device/App/usbd_dfu_flash.c: line 66
#define FLASH_DESC_STR "@Internal Flash /0x08000000/6*8Ka,122*8Kg"
I'm wondering if newer silicon revision fixes this problem.
2020-05-12 11:57 AM
>>I'm wondering if newer silicon revision fixes this problem.
I seem to recall the system memory being flash/otp so flaws in the loader would be addressed in programming/testing phase.