CPU stalling at XINTF memory access TMS320F28335

Hi E2E community,

I have an issue of a sporadically (approx. 1% of the cases) stalling CPU after powercycling with the TMS320F28335. As I figured out so far, the CPU stalls at the first access to the external RAM (write access) through the XINTF interface during initialization of a task. Unfortunately I cannot show here the source code.

XINTF configuration (zone 6):

Data size: 32 bit

Write buffer: 0

Lead Wait states: 1 (same for read and write)

Active Wait states: 3 (same for read and write)

Read Wait states: 1 (same for read and write)

XReady: disabled (0)

Hold: disabled (1)

XTIMCLK: 1 (divide clock by 2)

CLKOFF: 1 (clock out disabled)

X2TIMING:0

------------------------------------------------------

Systemclock = 128MHz

Hardware architecture:

Two 16-bit asynchronous RAM's are used. One RAM is for low 16 bit (XD(15;0)) and one for high 16 bit XD(16;31). The complete address bus is shared by both. Chipselect goes to both RAM's. XREADY, XHOLD or XHOLDA pins are not used, resp. used for other purposes as GPIO's. XINTF is using only memory zone 6 (0x100000-0x200000). Maximum access time of the RAM is 25ns.

Before, the ready mode was accidently enabled and it made the CPU stalling occur more often (XReady pin is not connected and therefore floating).

The datasheet is saying that either the Ready mode or the Hold mode can stall the CPU, but both are disabled.

The symptom occurs also on other boards. I haven't figured out any dependencies yet.

Does anyone has an idea?

Thanks in advance.