Hello.
I try to get the nema-gfx working on a STM32H7S78-DK.
But there's a bug in the NEMA library, or a bug in the ST toolchain.
When I try to call the library function nema_cl_create() in my function nema_gfx_init(), according to the STM32H7RS GPU2D document (page 6), stm32h7rs-system-gpu2d.pdf
the stack is overwritten with values, so the return address on stack (0x7002AEFB) is no longer available (0x00000000) and the cpu goes in hard fault.
I suspect it's a bug in the NEMA library.
The correct version for the H7 series was used.
The error occurs with both versions: libnemagfx-float-abi-hard.a and libnemagfx.a.
I looked inside the function in the disassembler, and at the end of the function, registers r0-r3 are written to the stack, but much too far! (See images)
Stack bevore calling nema_cl_create. Return-adress: ok
Stack after calling nema_cl_create. Return-adress: 0x00000000
Dissass. inside der nema_cl_create: overwriting behind actual stack end
The library is integrated into the STM32CubeIDE as usual via Settings -> MCU/MPU GCC Linker -> Libraries.
Are there any additional settings required for the libraries in the STM32CubeIDE project to prevent the stack from being overwritten?
