cancel
Showing results for 
Search instead for 
Did you mean: 

Fixed version of Cortex in STM32H7 ?

Gpeti
Senior II

Hello,

I've heard that some bugs have been found on Cortex M7 related to cache management, and that the STM32H7 will embed the fixed version of Cortex design. But I can't find when this updated version of STM32H7 will be available. Also, where (in what document) is the ARM Cortex M7 design revision specified ?

Regards

11 REPLIES 11

​OK I've found the application note, I was confused by a ST web page, sorry for that.

For the rest of your post... It is a common criticism done to ST's software, that I don't completely share but I don't want to feed the troll on this topic. I just find that, generally speaking and based on many of my previous experiences, many developpers are lacking humility...

It's exactly the opposite. Western society has gained too much humility for lazy, dumb, incapable people, low quality things, general degradation and even their own destruction. OK, but this is technical forum, so, talking about HAL, here we go...

Critical deficiencies:

  • Not interrupt safe.
  • Not thread safe and has no thread safe layer.
  • Has buffer overrun/underrun bugs.
  • Lacks processor and compiler memory barriers where needed.
  • Wrong cache clean/invalidate usage on cores with cache memory.

Architectural and performance deficiencies:

  • Bad interrupt performance, because of the same unsafe enable/disable.
  • Full-duplex peripherals used only as half-duplex.
  • Lacks asynchronous APIs for receive functionality.
  • Lacks zero-copy buffering where it should be.
  • Doesn't have necessary interrupt callback per peripheral instance.
  • Does have unnecessary bunch of callbacks per interrupt event type.
  • Interrupt callbacks implemented as weak functions, prohibiting building compiled libraries.
  • Code is not modular and can't be built for multiple boards and/or configurations.

And here is proof to my "trolling":

https://community.st.com/s/question/0D50X0000AhNBoWSQW/actually-working-stm32-ethernet-and-lwip-demonstration-firmware

It's a few months of work for a team of one (me) in my free time on average 2-3 evenings per week. Try it and compare to ST's multiple code monkey team in multiple years generated non-working absolute crap-ware.