[LTO] Request for testing: Last merge from trunk before final merge

[Diego Novillo <dnovillo at google dot com>]

In preparation for the final merge into mainline.  I need to test
the branch on various platforms.  Richi is currently testing on
i586, ppc, ppc64, ia64, s390, s390x.

If anyone has free cycles I would appreciate results from other
ELF-capable targets.

$ svn co svn://gcc.gnu.org/svn/gcc/branches/lto
$ mkdir bld && cd bld
$ ../lto/configure --enable-lto && make

You will need to have libelf 0.8.12 installed
(http://www.mr511.de/software/libelf-0.8.12.tar.gz)


I have made some cleanups in this merge:

- Spurious changes that were not needed.

- Removed LTO testing from several testsuites to prevent
  excessive build slowdowns.  We essentially only test the lto
  directories, some dg tests, executable tests and not much else.

  In particular, we no longer test libstdc++ with lto enabled (it
  had 4 failures) and fortran (it had 143 failures).  At some
  point, I would like to bring LTO testing back to fortran, but I
  understand that this needs fixes in the FE.

  My plan is to set up a tester that forces LTO builds on every
  testsuite.  I will move unique failures to one of the LTO
  directories and create tracking bugs (help appreciated in doing
  this).

- We no longer do functional testing for LTO to decide whether to
  run LTO tests.  We just test whether LTO has been enabled.
  Before, we would silently refuse to run any LTO testing if the
  little program used to detect LTO was broken.

- Removed two hacks in the C++ mangler.  In free_lang_data, we
  were creating more mangled names than necessary.  As Jason
  pointed out, there is no need to generate mangled names for
  templates.  I added a new langhook that need_assembler_name_p
  calls if set.  The only front end that needs it (so far) is
  C++.

Bootstrapped and tested on x86_64.


Diego.