arm-none-eabi/bin/ld: region RAM overflowed with stack

[Jim]

I'm getting this error, but oddly it's coming up somewhere in the MIDDLE of my build, and not always the same point.

I have modified the HEAP size to be 0 as I don't do any malloc-ing. But, I did that in gcc_startup_nrf51.s, which isn't compiled yet at this point in the build (nor is main.c).

I've started this particular project from the HRM example, which I am able to build under eclipse when the wind is blowing in the right direction.

I've attached my build output, showing the error coming up early in the build, and my .map file, which I am under-qualified to interpret properly.

Anyone got an idea?

10:30:50 **** Incremental Build of configuration D.txt

fizzy_s110_xxaa.map.txt

[Jim]

OK, I'm looking at the .map file and it looks like I've somehow inherited a whole load of libraries, courtesy of gcc. My suspicion is that they are related to the existence of "printf" statements in my code which are intended to send messages out through the UART, so I'm trying to figure out how to re-implement the functionality of retarget.c that I had been using. I'll post another question on this topic to avoid confusion...

[Jim]

I'm definitely picking up libraries I shouldn't. A lot of libc.a in the map file. Any way to control what gcc decides to include on it's own?

[Jim]

OK, I've resolve the mystery of the out-of-order linking. It's actually an Eclipse issue. The Eclipse console cannot properly interleave stdout and stderr from the build. I guess that's hard to do. I am resisting challenging that notion with all my might.

So, if you get confused by Eclipse build output, try breaking the build up with some intermediate targets.

Silly silly silly.