[PATCH v2 0/3] RISC-V: The series to supporting mapping symbols

[Nelson Chu <nelson.chu@sifive.com>]

Hi Guys,

See in advance what are the difference with v1,
* Support data mapping symbols with the data size, if they are added
  by cons_worker, e.g. $d2, $d4.
* Support alignment mapping symbols, $a and $a1.  If the instructions
  are used for alignment, then we dump them with the prefix align.<insn>.
* Since the above two new features, we don't need to record all
  mapping symbols of the fragments.  Instead, only record the
  first and last mapping symbols should be enough.
* Extend .insn directive to support hardcode encoding.

These are all binutils changes, but not sure if the changes will break
the gdb disassembler.  Unfortunately I am not familiar with gdb, hope
gdb experts can give some suggestions, or warnings if something may be
broken.

Besides, the binutils readelf testcases in the second patch are affected
by the first patch, so I send them as the same series.  But in fact the
first patch can be regarded as an independent issue.  Therefore, I think
we can commit it separately first, if everything looks good.

The elf/linux toolchain regressions all passed, looks fine so far.

Thanks
Nelson