Hello all,

various sources (aliexpress, really) where one can buy a P4-Evalboard state that the chip is still has some rough edges (calibration, etc.).

Is that still true? If yes, is there any expectation/information, when newer revisions will be available?

Thanks!

GNU extensions? You could try to pass "-pedantic" to g++. That would prohibit extensions to the Standard.

OK, understood, thanks!

Hello Sprite, thanks for the pointers. Indeed, some data can be dumped with "mon flash read_bank"; for just printing to the screen "mon flash mdb" seems to work fine, too. I still seem to miss something: suppose I have a custom partition table with an entry for data at offset 0xF000 and suppose that...

Nothing?
Printing flash memory should be possible (for internal memory, but also for external memory because it is mapped into address space).

Maybe the MMU interferes during runtime?

Hi,

is it, in principle, possible to read flash memory in GDB?

I have tried to print the partition table at address 0x8000. But only zeros are returned.
The chip in question is an ESP32S3 with 32MB external flash memory.

Maybe some lowlevel openocd command can be used?

Thanks!

Just add `mon xtensa set_permissive 1`

Thanks. Can confirm that this works around the issue: memory is now readable and writable.

I modified the "hello-world" example to make the issue visible. It allocates a buffer and sets the first byte. Then it continues to dump a region from the buffer. Note, that the modified example uses `heap_caps_malloc` but it behaves the same with plain malloc. /* Hello World Example This example co...

Sorry to check the obvious things too late, but with the latest release version of openocd the issue persists. And further infos/test results: The allocated buffer is part of a struct, and the struct is in internal memory `0x3fcb1298` From GDB I cannot set any value in the malloc'd buffer: "Cannot a...

Sorry to hijack the thread, but there is an interesting question that I did not see an explicit answer for. This "encryption-aware" concept really does not make any sense. You are suggesting if you have a working "plain text" app installed, and then you turn on the built-in encryption process, the w...