"R. Diez" <rdiezmail-gcc@xxxxxxxx> writes: > /home/rdiez/rdiez/Code/ClientCardFirmware/ArtistClientMpc850/Scripts/BuildMpc8xxToolchain/Tmp/cross-toolchain-mpc8xx-eabi-tmp/gcc-4.6.1/libiberty/regex.c:51:25: fatal error: sys/types.h: No such file or directory > compilation terminated. > make[3]: *** [regex.o] Fehler 1 > make[2]: *** [all-target-libiberty] Fehler 2 > make[1]: *** [all] Fehler 2 > make: *** [/home/rdiez/rdiez/Code/ClientCardFirmware/ArtistClientMpc850/Scripts/BuildMpc8xxToolchain/Tmp/cross-toolchain-mpc8xx-eabi-tmp/gcc-4.6.1-mpc8xx-eabi/phase1/ArtistMake1Sentinel] Fehler 2 libiberty/regex.c hasn't changed. I don't think building target-libiberty has changed. So one thing to do would be to go back and look at your 4.5.3 build and see how it worked. > I have a long and complicated makefile that builds all the different bits and pieces, it's the same makefile that used to work fine for the older GCC versions. The steps are as follows: > 1) Unpack the gcc-4.6.1 tarball > 2) Unpack the binutils-2.21 tarball > 3) Configure binutils for the embedded PowerPC target > 4) Build binutils > 5) Install binutils to some local path > 6) Build the first phase of GCC. That's what fails. > 7) The next steps would be to build newlib and the second phase of GCC. I would not expect that building the first phase of gcc would build target-libiberty. So what do you mean when you say "build the first phase of GCC?" I would expect this to mean something like "make all-gcc", but your make command appear to be trying to build all of gcc before newlib has been built. This is a case where unpacking newlib into your gcc source directory, under the name "newlib", and configuring gcc using --with-newlib, may work well. Or you can continue the procedure you are using, but run "make all-gcc" and "make install-gcc" for the first phase. Ian
