SPI debugging stops after around 125 bytes being send

[johantheking]

Hi STM forums,

I use a stm32f205rct6tr for this project

in my project i need to send 1024 clock signals to the slave, the message inside is not inportant.

but after sending around 125 bytes it stops debugging entirely, i checked it out with the oscilloscoop. And i'm not sure why it can't send more. plz take a looks at the screenshot and if there is more information needed let me know.

Thanks for the help.

[TDK]

1024 bytes is a lot for the stack. Might want to define that as a global variable, or to ensure you have defined enough stack space in the linker file (_Min_Stack_Size variable).

[Tesla DeLorean]

Perhaps it stalls or underflows because you're debugging. Don't park a peripheral view over an active peripheral that clears registers, or has a FIFO.

SPI is driven by output, not input. Use Transmit or TransmitReceive methods.

Use a static buffer, not a local/auto variable.

[David Littell]

Combining what both gentlemen have contributed, you don't need ontvangenbuffer[] at all.  You mentioned the data content of your SPI frame doesn't matter, only the SPI clocks.  So definitely use HAL_SPI_Transmit() but since the data doesn't matter the address of the buffer passed to ..._Transmit() can be *any* address, e.g., the base of RAM.  Don't forget to include a very descriptive comment!

[waclawek.jan]

Try  bitbanging the SPI.

JW

[johantheking]

Hi everyone,

I would like to thank everyone in this conversation for there suggestions on how to fix this issue. Unfortunately none of them worked. but while i was digging deeper to find where it caused the problem, i found out that after sending 125 bytes, in the file stm32f2xx_hal_spi.c the program keeps looping forever between line 1272 to 1307, check the picture pls. If there are anymore suggestions let me know.

Johan

[waclawek.jan]

Note, what @Tesla DeLorean said above:

Perhaps it stalls or underflows because you're debugging. Don't park a peripheral view over an active peripheral that clears registers, or has a FIFO.

JW

[Tesla DeLorean]

Check what's flagging when it gets stuck in the loop. Seems to assume RXNE / TXE and not any potentially for error / failure conditions.

[johantheking]

it appears to be that the RXNE is constantly enable in a loop, any idea on how to fix this, or what it may cause to be stuck on this loop?

[waclawek.jan]

> SPI_DIRECTION_2_LINES

I don't use Cube, does this mean Bidir or Rx-only modes of SPI? If yes, don't use them.

JW