tccgen.c:
- fix ldouble asm hack
- fix a VLA problem on Win64 (also x86_64-gen.c)
- patch_type(): make sure that no symbol ever changes
  from global to static

tcc.c:
- tcc -vv: print libtcc1.a path also on win32

tccpe.c, tcctools.c:
- use unix LF mode to for .def output files (that is for
  creating reproducible output trees)

Makefile:
- suppress some warnings when makeinfo is missing
- call 'which install' only on win32

tests/Makefile:
- change PATH only on WINNT systems (i.e. not if cross-compiling
  on linux for win32)
- asm-c-connect.test: slim output and do diff

tccrun.c tccpe.c *-link.c:
- integrate former 'pe_relocate_rva()' into normal relocation
  This also fixes linkage of the unwind data on WIN64 for -run
  (reported by Janus Lynggaard Thorborg)

tccasm.c, tests/tcctest.c:
- fix dot (sym_index of -1 crashed in put_elf_reloc)
- massage .set a bit (see test)

other:
- #define SECTION_ABS removed
- ST_DATA Section *strtab_section: removed
- put_extern_sym2(): take int section number

Conflicts:
	tccelf.c
	tccpe.c

Conflicts:
	tccelf.c

This properly fixes what 870271ea tried to fix.  Absolute memory
references can't use %rip relative addressing, and additionally,
if the address doesn't fit 32bit (signed) it must be loaded via
movabs.  No good testcase added, it would require catching signals
and still be unreliable.

This reverts commit 870271ea071971002fa556e09e1873db316fa1a9.

The commit is broken, you can't unconditionally emit a PC-relative
relocation without a symbol.  And if there's a symbol the addend
need to be in the relocation, not the section.

Comments only, no change to functionality

Dereferencing of absolute pointers is broken on x86_64, eg:

*(int*)NULL does not segfault but returns -4 instead
*(char*)(-10L << 20) does not return 0x55 (vsyscall page, push rbp)

Eliminate VT_ENUM as a basic type.  Instead use the integral
types VT_InT/VT_LLONG with an additional flag to indicate
that it is an enum.

tcc.h:
* cleanup struct 'Sym'
* include some 'Attributes' into 'Sym'
* in turn get rid of VT_IM/EXPORT, VT_WEAK
* re-number VT_XXX flags
* replace some 'long' function args by 'int'

tccgen.c:
* refactor parse_btype()

- configure
  * use aarch64 instead of arm64

- Makefile
  * rename the custom include file to "config-extra.mak"
  * Also avoid "rm -r /*" if $(tccdir) is empty

- pp/Makefile
  * fix .expect generation with gcc

- tcc.h
  * cleanup #defines for _MSC_VER

- tccgen.c:
  * fix const-propagation for &,|
  * fix anonymous named struct (ms-extension) and enable
    -fms-extension by default

- i386-gen.c
  * clear VT_DEFSIGN

- x86_64-gen.c/win64:
  * fix passing structs in registers
  * fix alloca (need to keep "func_scratch" below each alloca area on stack)
    (This allows to compile a working gnu-make on win64)

- tccpp.c
  * alternative approach to 37999a4fbf3487b363fd0c2f9b7ab622401b202a
    This is to avoid some slowdown with ## token pasting.
  * get_tok_str() : return <eof> for TOK_EOF
  * -funsigned-char: apply to "string" literals as well

- tccpe/tools.c: -impdef: support both 32 and 64 bit dlls anyway

If there were more than 6 integer arguments before the ellipsis, or
there were used more than 8 slots used until the ellipsis (e.g. by
a large intermediate struct) we generated wrong code.  See testcase.

This fixes two ABI testcases involving large arguments when there
are still registers available for later args.