Hello GCC Users, I have fundamental question with regard to GCC for PowerPC. Does it or does it not fully support EABI specified SDA sections. I use : $ powerpc-elf-gcc -v Reading specs from /usr/local/lib/gcc/powerpc-elf/3.4.0-macraigor1/specs Configured with: /rel/share/gnu/src/gcc-3.4.0-macraigor1/configure --host=i686-p c-cygwin --target=powerpc-elf --disable-nls --with-stabs --disable-install-libib erty --disable-install-libbfd --disable-dependency-tracking --enable-languages=c ,c++ --with-gnu-as --with-gnu-ld Thread model: single gcc version 3.4.0-macraigor1 There are two problems with this version of GCC: 1) The compiler is assigining a wrong relocation entry for a function. 2) Instead of using .sbss for uninitialized small variables, it uses the .comm section and thus the final link fails. I give here the example: powerpc-elf-ld: ./objs/main.o: the target (func3) of a R_PPC_EMB_SDA21 relocatio : n is in the wrong output section (.text) powerpc-elf-ld: ./objs/main.o: the target (.rodata.str1.4) of a R_PPC_EMB_SDA21 relocation is in the wrong output section (.rodata) powerpc-elf-ld: final link failed: Bad value make: *** [bin/pipeline_test.elf] Error 1 Here is the relocation info: 0000085c 0000136d R_PPC_EMB_SDA21 00000000 .rodata.str1.4 + 0 => Small Data addressed, but not in .sdata2 000007c4 00000d6d R_PPC_EMB_SDA21 000000d0 func3 + 0 => A function addressed through SDA !? >From Assembly: .section .rodata.str1.4,"aMS",@progbits,1 .align 2 .LC2: .string "abc" char vchar; short vshort; int vint; long vlong; char * vpchar; short * vpshort; int * vpint; long * vplong; unsigned char vuchar; unsigned short vushort; unsigned int vuint; unsigned long vulong; unsigned char * vpuchar; unsigned short * vpushort; unsigned int * vpuint; unsigned long * vpulong; unsigned char * * vppuchar; unsigned short * * vppushort; unsigned int * * vppuint; unsigned long * * vppulong; .comm vchar,1,1 .comm vshort,2,2 .comm vint,4,4 .comm vlong,4,4 .comm vpchar,4,4 .comm vpshort,4,4 .comm vpint,4,4 .comm vplong,4,4 00000020 C vbfield 00000001 C vchar 00000005 C vdarray 0000001e C vdblarray 00000004 C vint 00000004 C vlong 00000004 C vpchar 00000004 C vpint 00000004 C vplong 00000004 C vppuchar My compiler options: CCOPTS = -O2 -meabi -G 256 -msdata=eabi -mbig -mregnames -mstrict-align -mupdate -mmultiple -nostdinc -nostdlib Is there any problem in this version of GCC ? Has it been fixed ? Or am missing some information ? Many thanks for your help! Regards, Ravishankar
