Possible bug with gcc 2.95.2 on Hitachi h8

[Ross Crawford <rcrawford at csi dot com>]

Hi, all.

I'd like to report a problem. I'm using legOS, an open-source alternative
firmware for the Lego Mindstorms, based on the h8/300 CPU. The gcc
compiler seems to produce incorrect assembler code for the source file in
the attached tarball, memcpy.c (legOS' implementation of the standard libc
function).

Included is:
memcpy.c (source file)
cmdline (compiler command line used)
gcc/gcc.log (output from gcc compilation)
gcc/memcpy.i
gcc/memcpy.s
egcs/gcc.log
egcs/memcpy.i
egcs/memcpy.s

The reason I include the egcs/* files is that egcs 1.1.2 compiles the same
code correctly, and these are the resulting files for comparison.

Please let me know if there's any more info you need - I'm not familiar
with compiler internals at all, and not really familiar with the h8 yet.

And thanks for a great product!!

Regards

ROSCO

Bug Info