[PATCH,4.6/4.5.1,PR42776] Implement LTO for Windows (PE-COFF) targets.

[Dave Korn]

    Hi list,

  The attached patch generalizes the interface between lto1 and its object
file reader, and provides an object file format-specific reader for COFF
(currently only validated against and enabled for Windows platforms) as an
alternative to the current ELF object file reader.

  The configury provided to make the object format reader selectable will
facilitate an eventual solution to PR43729 (no LTO on Mach-O) and the COFF
reader will be easily extensible (if not already sufficient) for any other
COFF targets that want to enable LTO.

ChangeLog:

	PR lto/42776
	* configure.ac (--enable-lto): Refactor handling so libelf tests
	are only performed inside then-clause of ACX_ELF_TARGET_IFELSE,
	and allow LTO to be explicitly enabled on non-ELF platforms that
	are known to support it inside else-clause.
	* configure: Regenerate.

gcc/ChangeLog

	PR lto/42776
	* configure.ac (gcc_cv_as_section_has_align): Set if installed
	binutils supports extended .section directive needed by LTO, or
	warn if older binutils found.
	(LTO_BINARY_READER): New AC_SUBST'd variable.
	(LTO_USE_LIBELF): Likewise.
	* gcc/config.gcc (lto_binary_reader): New target-specific configure
	variable.
	* gcc/Makefile.in (LTO_BINARY_READER): Import AC_SUBST'd autoconf var.
	(LTO_USE_LIBELF): Likewise.
	* configure: Regenerate.

	* collect2.c (is_elf): Rename from this ...
	(is_elf_or_coff): ... to this, and recognize and allow i386 COFF
	 object files in addition to ELF-formatted ones.
	(scan_prog_file): Caller updated.  Also allow for LTO info marker
	symbol to be prefixed or not by an extra underscore.

	* config/i386/t-cygming (winnt.o): Also depend on LTO_STREAMER_H.
	* config/i386/winnt.c: Also #include lto-streamer.h
	(i386_pe_asm_named_section): Specify 1-byte section alignment for
	LTO named sections.
	(i386_pe_asm_output_aligned_decl_common): Add comment.
	(i386_pe_maybe_record_exported_symbol): Allow decl arg to be NULL.

lto/ChangeLog

	PR lto/42776
	* Make-lang.in (LTO_OBJS): Use LTO_BINARY_READER instead of
	hardcoding 'lto-elf.o'.
	($(LTO_EXE)): Use LTO_USE_LIBELF instead of hardcoding '-lelf'.

	* lto.h (lto_elf_file_open): Rename prototype from this ...
	(lto_obj_file_open): ... to this.
	(lto_elf_file_close): Likewise ...
	(lto_obj_file_close): ... and likewise.
	(lto_elf_build_section_table): Likewise ...
	(lto_obj_build_section_table): ... and likewise.
	(lto_elf_begin_section): Likewise ...
	(lto_obj_begin_section): ... and likewise.
	(lto_elf_append_data): Likewise ...
	(lto_obj_append_data): ... and likewise.
	(lto_elf_end_section): Likewise ...
	(lto_obj_end_section): ... and likewise.
	* lto.c (lto_file_read): Update references to the above.
	(lto_wpa_write_files): Likewise.
	(lto_read_all_file_options): Likewise.
	(read_cgraph_and_symbols): Likewise.
	* gcc/lto/lto-lang.c (LANG_HOOKS_BEGIN_SECTION): Likewise.
	(LANG_HOOKS_APPEND_DATA): Likewise.
	(LANG_HOOKS_END_SECTION): Likewise.
	* lto-elf.c (lto_elf_file_open): Rename from this ...
	(lto_obj_file_open): ... to this, updating any references.
	(lto_elf_file_close): Likewise ...
	(lto_obj_file_close): ... and likewise.
	(lto_elf_build_section_table): Likewise ...
	(lto_obj_build_section_table): ... and likewise.
	(lto_elf_begin_section): Likewise ...
	(lto_obj_begin_section): ... and likewise.
	(lto_elf_append_data): Likewise ...
	(lto_obj_append_data): ... and likewise.
	(lto_elf_end_section): Likewise ...
	(lto_obj_end_section): ... and likewise.
	* lto-coff.h: New file.
	* lto-coff.c: Likewise.

gcc/testsuite/ChangeLog

	PR lto/42776
	* lib/lto.exp (lto_prune_vis_warns): New function.
	(lto-link-and-maybe-run): Call it.

  Bootstrapped on i686-pc-cygwin against r.158195, tests in progress; repeated
previous runs have shown that all lto tests (except for a couple which rely on
ELF symbol aliasing) now pass without regressions.  It's also been tested,
according to contributors on the PR, against MinGW, where it works, and on
Linux, where it doesn't break existing ELF LTO any.  I've also quickly tested
(in a non-bootstrap configuration) that the configury correctly warns when I
roll back my binutils to a version that lacks the necessary support for LTO on
COFF (an extended form of the .section directive that allows specifying
alignments below the default section alignment).

  Assuming no regressions when the tests complete, OK for HEAD, and for 4.5
branch when it reopens after the .0 release?

    cheers,
      DaveK
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pr42776-final.diff
Type: text/x-c
Size: 56422 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20100413/3cb1997e/attachment.bin>