Clive Two.Zero

Prefetch issues on the STM3240G-EVAL

Blog Post created by Clive Two.Zero on Feb 23, 2017

An issue came up over at the Keil Forum the other day, and reminds us to be cautious and careful about working with some early silicon that might be on the boards we use.

scary STM32F4 compiler issue? 


The STM3240G-EVAL boards, mine included, have an STM32F457IGH6 part, an Engineering Sample (ES) that didn't get into wide availability, and your tools might not even have a selection for. It appears to be an A-stepping of the F407, uses the same Cortex-M4 core and patch level as the F407, but exhibits a failure with the ART Prefetch path. I suspect this is similar to a critical path issue seen on the F2 parts, which seemed particularly sensitive to a code sequence the GNU/GCC compilers emitted, and was the subject of a number of forum posts a few years back.