[Bug bootstrap/28026] Cross compiles involving Darwin fail

[geoffk at gcc dot gnu dot org]

--- Comment #6 from geoffk at gcc dot gnu dot org  2006-10-03 20:47 ---
In

I report that cross-compiling from powerpc-darwin to i386-darwin succeeds.


-- 

geoffk at gcc dot gnu dot org changed:

   What|Removed |Added

 Status|UNCONFIRMED |RESOLVED
 Resolution||WORKSFORME


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

[Bug bootstrap/28026] Cross compiles involving Darwin fail

[w dot northcott at unsw dot edu dot au]

--- Comment #5 from w dot northcott at unsw dot edu dot au  2006-06-23 
05:57 ---
(In reply to comment #3)
> I think the problem with the FSF build is that the SDK is actually
> *-darwin-8.6.1 but your host system is running Tiger which is *-darwin-8.6.0. 

Thanks for the thoughgt, but I don't quite get your drift.

The test is 
if test $hostos = $targetos -o $targetos = darwin
In the Apple build this succeeds because both target and host are set to
darwin8.
In the FSF build the same test fails because the target is set to ...darwin8
and the host to ...darwin8.6.whatever.  Since the test fails, the build
succeeds because CROSS_COMPILE is still defined.  It is a case of two wrongs
making a right, but it is not nice.

Bill


-- 


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

[Bug bootstrap/28026] Cross compiles involving Darwin fail

[geoffk at gcc dot gnu dot org]

--- Comment #4 from geoffk at gcc dot gnu dot org  2006-06-22 23:38 ---
(In reply to comment #3)
> I think the problem with the FSF build is that the SDK is actually
> *-darwin-8.6.1 but your host system is running Tiger which is *-darwin-8.6.0. 

I mean, of course, *-apple-darwin8.6.1.


-- 


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

[Bug bootstrap/28026] Cross compiles involving Darwin fail

[geoffk at gcc dot gnu dot org]

--- Comment #3 from geoffk at gcc dot gnu dot org  2006-06-22 23:36 ---
I think the problem with the FSF build is that the SDK is actually
*-darwin-8.6.1 but your host system is running Tiger which is *-darwin-8.6.0. 
If you give configure the full three-entry version numbers it will realise you
are trying to build a cross compiler to a different Darwin version and will not
assume that the system libraries are what you want.


-- 


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

[Bug bootstrap/28026] Cross compiles involving Darwin fail

[w dot northcott at unsw dot edu dot au]

--- Comment #2 from w dot northcott at unsw dot edu dot au  2006-06-14 
20:28 ---
(In reply to comment #1)
> What problem did you see when ld wasn't in build_gcc?

collect2: cannot find ld
when trying to link cross-compilers.  

This is not seen whn building on a fat library machine because then there are
no cross-compiler builds.
> 
> Also, if the compiler is configured with isysroot, if gcc.c is taught to
> default to using it, then I think the change to configure.ac isn't needed. 
> This also helps with point #3 I think.
> 

The logic that comes out of my tests runs like this:
1.  Unless CROSS_COMPILE is defined --with-sysroot and its derivative
TARGET_SYSTEM_ROOT are ignored.  So the mod to configure.ac is absolutely
essential

2 --with-sysroot would be enough on most platforms.  the -isysroot in the
CFLAGS is needed to prevent the issue with libsystem referencing the system
libmathCommon as in my email.

Bill


-- 


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

[Bug bootstrap/28026] Cross compiles involving Darwin fail

[mrs at apple dot com]

--- Comment #1 from mrs at apple dot com  2006-06-14 17:11 ---
What problem did you see when ld wasn't in build_gcc?

Also, if the compiler is configured with isysroot, if gcc.c is taught to
default to using it, then I think the change to configure.ac isn't needed. 
This also helps with point #3 I think.


-- 


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