Known bug in remote serial protocol, or something else?

[Juha Aaltonen]

I was trying to debug my "standalone gdbstub" when I encountered a
strange problem.

I loaded a debuggee program and started executing it, and it crashed
in a weird address.
When I disassembled the code (ARMv7 target) I found out that some
instructions were corrupted.
I couldn't figure it out about binary transfer (Looked like UTF-8, but
didn't quite match), so I disabled
the 'X'-command, and the 'M'-command seemed to do even worse - telling
by the debug messages
(set debug serial 1, set debug remote 1).

The gdb client I'm using is gdb-multiarch 7.7.1 on Debian 8 (x86_64) -
the one in the official repo.

I'm happy to supply further info, if it could be of any help -
including the debuggee ELF, but maybe this
mailing list is not the place...