[Bug bootstrap/28026] New: Cross compiles involving Darwin fail

["w dot northcott at unsw dot edu dot au" <gcc-bugzilla at gcc dot gnu dot org>]

This primarily affect Apple's gcc-5xxx but some of the issues are also in code
which is part of FSF gcc-4.x.x.  There may be some connection with PR 26814.

The symptom is that building i386 hosted or i386 target compilers on powerpc
Darwin using the SDK fails.  The problems would probably also affect attempts
to build Darwin targetted compilers on Linux or other OS.

I have following problems in the code with possible fixes:
1.  This affects both Apple and FSF code.  In gcc/configure.ac around line 1637
there is the following
        # Darwin crosses can use the host system's libraries and headers,
        # because of the fat library support.  Of course, it must be the
        # same version of Darwin on both sides.  Allow the user to
        # just say --target=foo-darwin without a version number to mean
        # "the version on this system".
            *-*-darwin*,*-*-darwin*)
                hostos=`echo $host | sed 's/.*-darwin/darwin/'`
                targetos=`echo $target | sed 's/.*-darwin/darwin/'`
                if test $hostos = $targetos -o $targetos = darwin ; then
                    CROSS=
                    SYSTEM_HEADER_DIR='$(NATIVE_SYSTEM_HEADER_DIR)'
                    with_headers=yes
                fi
                ;;

This might have been correct when it was added in 2003 but is not true for all
Darwin versions now.  I suggest a specific test for fat libraries:
        # Darwin crosses can use the host system's libraries and headers,
        # because of the fat library support.  Of course, it must be the
        # same version of Darwin on both sides.  Allow the user to
        # just say --target=foo-darwin without a version number to mean
        # "the version on this system".
            *-*-darwin*,*-*-darwin*)
                hostos=`echo $host | sed 's/.*-darwin/darwin/'`
                targetos=`echo $target | sed 's/.*-darwin/darwin/'`
                if test x`lipo -info /usr/lib/libSystem.b.dylib | grep i386 |
grep ppc` != x
                then
                        if test $hostos = $targetos -o $targetos = darwin ;
then
                                CROSS=
                                SYSTEM_HEADER_DIR='$(NATIVE_SYSTEM_HEADER_DIR)'
                                with_headers=yes
                        fi
                fi
                ;;
This could easily be extended if Steve decides to announce in August that the
new XServe will use Itanic. (only kidding)

2. The remaining problems only affect Apple's build system.
The configure invocations need a
"--with-sysroot=/Developer/SDKs/MacOSX10.4u.sdk" option.

This can be added to the CONFIGFLAGS definition in build-gcc around line 109 or
see below.

3. The CFLAGS for many of the compiles needs to include
"-isysroot=/Developer/SDKs/MacOSX10.4u.sdk".

This can be added in the invocation of build-gcc by defining RC_NONARCH_CFLAGS
but it would be better to deal with both 2 and 3 with a new envvar like
BUILD_SDK.

4. The targetted build tools need to include ld.

This can be achieved by adding ld to the list of tools in build-gcc around line
157.

I have tested building Apple gcc-5341 with these mods and the build succeeds on
both PowerPC and Intel Macs.

Bill Northcott


-- 
           Summary: Cross compiles involving Darwin fail
           Product: gcc
           Version: 4.0.1
            Status: UNCONFIRMED
          Severity: minor
          Priority: P3
         Component: bootstrap
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: w dot northcott at unsw dot edu dot au
 GCC build triplet: powerpc-apple-darwin8 i386-apple-darwin8
  GCC host triplet: powerpc-apple-darwin8 i386-apple-darwin8
GCC target triplet: powerpc-apple-darwin8 i386-apple-darwin8


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28026