On Fri, 2016-05-20 at 19:37 +0200, Andreas Schwab wrote: +2016-05-20 Andreas Schwab <[email protected]> > + > + * Makefile.am (modules): Add m68k. > + (libebl_pic): Add libebl_m68k_pic.a. > + (m68k_SRCS, libebl_m68k_pic_a_SOURCES) > + (am_libebl_m68k_pic_a_OBJECTS): Define. > + * m68k_init.c: New file. > + * m68k_symbol.c: New file. > + * m68k_regs.c: New file. > + * m68k_retval.c: New file. > + * m68k_corenote.c: New file. > + * m68k_reloc.def: New file. > + * linux-core-note.c (ALIGN_INT): Only define if not defined.
This looks good. Just a couple of comments because I know nothing about m68k. Is there a canonical ELF/gabi processor spec for m68k? If so it would be good to add an URL in e.g. m68k_regs.c so others can easily lookup the register mappings described there. If the only "official" documentation is in GCC or binutils/gdb that would also be good to mention. How does the make check result look on m68k with this patch applied? Any suspicious FAILs or UNSUPPORTED results? Could you provide testcases/files for run-readelf-mixed-corenote.sh, run-allregs.sh, run-strip-test?.sh and/or run-strip-reloc.sh. That way people can run make check on a non-m68k setup and see whether m68k ELF files are handled properly. If you are ambitious you could also add support for backtraces by implementing m68k_init_reg.c set_initial_registers_tid, m68k_cfi.c abi_cfi, frame_nregs. None of the above is mandatory to get the m68k support in, but it would help maintaining the support and making sure it doesn't accidentally break in the future. Thanks, Mark
