Turns out, the latest GCC does not allow null pointers, and we were checking if anyone was writing to address 0.
Maybe we can remove the null pointer check now, since gcc already implements it? (edit: after a few tests, -fdelete-null-pointer-checks does not appear to have much effect - it inserts an undefined instruction only if we try to dereference a pointer hardcoded to 0, but not a pointer passed as parameter, that might be 0, so I would leave these checks)
Any other obvious issues introduced with the new gcc?
Boots now on M, didn't test 5D3 yet, but previously they both failed.
yeah, no probs over here too
Just tried the latest 5.2.1 release from official GCC ARM Embedded launchpad. No problems with that on 70D, too. Once this PR gets hot again I suggest using the latest release