Schindler Karl-Michael wroteI tried to do the fpc tests and encountered a bug. The file "tests/
test/cg/obj/darwin/powerpc/tcext5.o" was missing. I created it with
gcc -c cext5.c from tests/test/cg/obj/tcext5.c. Then the test run
went through. However, I am not sure how cext5.c should be compiled.
It's been in svn for more than a month already:
~/fpc/tests/test/cg/obj jonas$ svn log -v darwin/powerpc/tcext5.o ------------------------------------------------------------------------ r3166 | jonas | 2006-04-08 00:17:45 +0200 (Sat, 08 Apr 2006) | 6 lines Changed paths:
M /trunk/compiler/powerpc/cpupara.pas
A /trunk/tests/test/cg/obj/darwin/powerpc/tcext5.o A /trunk/tests/test/cg/obj/tcext5.c
A /trunk/tests/test/cg/tcalext5.pp
* fixed skipping of parameter registers for ppc/aix in case a parameter
has to be passed on the stack before all registesters are used up
* fixed calculation of stack offsets for ppc/aix abi for small records,
arrays and ordinals passed on the stack after all registers are used
+ test for all of the above
------------------------------------------------------------------------
The i386 version is still missing though.
The notes in /trunk/tests/test/cg/obj/readme.txt also need an
update for darwin both processors.
It doesn't really matter for Darwin. Any gcc is ok, and you don't need special options.
Jonas