Kenneth Jonsson (kj.delete@this.localhost.org) on August 17, 2012 11:54 am wrote:
> I compiled with
> -O2 (recompiled now and double-checked that -O2 and no -O0 was passed to the
> compiler) and I excluded r13 (stack pointer, which "objdump" show as "sp"), r14
> (link register, shown as "lr") and r15 (program counter, listed as "pc"). The
> compiler is gcc 4.3.3, so not the latest and greatest. I got all this
> information via objdump, sed, sort and uniq and I'm not a expert on sed. But I
> opened the full output from "objdump -d" in emacs to double-check that those
> registers weren't use, only thing I saw with "r11" in was "cr11".

I don't think you caught what Wilco is telling you. Old versions of objdump gave the names "sl", "fp", and "ip" for r10, r11, and r12 respectively. That's almost definitely the reason why you're not seeing them. He also rightfully mentioned that you're unfairly culling R14 ("lr"), since not only is it used as a GPR but it's actually very commonly used for temporaries.

Calling GCC 4.3.3 not the latest and greatest is also a bit of an understatement, GCC 4.4, 4.5, and 4.6 all list major improvements in ARM code quality. This isn't surprising because it was over this period that improvements from Code Sourcery were merged.

I made this blog post several years listing some dynamic usage statistics for GBA games, for ARM and Thumb code separately. Of course the compiler used for these games would have been absolutely ancient, but they probably also had more hand-coded assembly, especially for ARM code. You can read it here:

http://gpsp-dev.blogspot.com/2007/03/register-usage-statistics.html