Bump gdb to FSF gdb 10.1, with our gdb sim patched in.

[jim-wilson]

Tested with build and make check for rv32i/ilp32 elf and rv64gc/lp64d elff, and with a check-sim in build-gdb-newlib. I had to fix DECLARE_CSR to have 5 args instead of 2. I had to fix the sim testsuite which looks like it maybe never worked. Otherwise it just worked. Doesn't work for riscv linux target, but not expected to.

This is expected to fix MacOS build errors, issues 745 and pull request 769.

This also gives us gdbserver.

[tyfkda]

I come here from #769 .

@jim-wilson Thank you to update the submodule.
I run make command with this change on Mac Catalina,
it seems compile error reported in #745 are solved,
but other error ocurred.

In file included from /Users/admin/tmp/riscv-gnu-toolchain/riscv-gdb/sim/riscv/interp.c:32:
/Users/admin/tmp/riscv-gnu-toolchain/riscv-gdb/sim/riscv/../../bfd/elf-bfd.h:3085:10: error: 
      implicitly declaring library function 'strncmp' with type 'int (const char *, const char *,
      unsigned long)' [-Werror,-Wimplicit-function-declaration]
  return strncmp (name, ".ctf", 4) == 0 && (name[4] == 0 || name[4] == '.');
         ^
/Users/admin/tmp/riscv-gnu-toolchain/riscv-gdb/sim/riscv/../../bfd/elf-bfd.h:3085:10: note: include
      the header <string.h> or explicitly provide a declaration for 'strncmp'
2 warnings and 1 error generated.
make[4]: *** [interp.o] Error 1
make[3]: *** [all] Error 1
make[2]: *** [all-sim] Error 2
make[1]: *** [all] Error 2
make: *** [stamps/build-gdb-newlib] Error 2

Is string.h not included?

[jim-wilson]

OK I see the problem now. The CTF changes to elf-bfd.h means that anything including elf-bfd.h now needs to include string.h. Upgrading gdb fixes all upstream code, but doesn't fix the local riscv sim. This didn't show up in pr #769 because it accidentally dropped the local riscv sim by switching to the binutils branch. I pushed another patch to my branch to fix the sim/riscv/interp.c file and updated my pull request to include it.

[tyfkda]

@jim-wilson I succeeded building the toolchain. Thank you!

[kito-cheng]

LGTM, verified with gcc testsuite, the only fail cases are already reported in riscvarchive/riscv-binutils-gdb#235