rfg@ics.uci.edu (Ron Guilmette) (10/22/89)
In article <8910131908.AA02951@mizar.DoCS.UU.SE> ewerlid@MIZAR.DOCS.UU.SE (Ove Ewerlid) writes: > >Around about the time GCC-1.34 appeared I installed it on an Intel Hypercube >(V/386 Rel 3.0 30386 alas not 3.2) > >I did experience problems but managed to circumscribe them by using two >C-compilers. (one PCC-based and a GREENHILLS produkt) >None of them managed to compile GCC alone but together they did it. > >When I saw your message I installed GCC-1.36 using GCC-1.34 without >any trouble at all. Hard-params barfed but that was due to CC not GCC. >(and as I understand it the error was not fatal for the floats) > >GCC-1.36 later bootstraped itself. > >I used the default assembler and linker. We now have GCC 1.36 running on the HyperCube (aka iPSC2). I believe that the OS version that we have is System V.3.2. To get GCC going, we used the "i386-sysv" configuration (not the i386-sysv-gas one). We used the "native" assembler and linker. We used the PCC derived (AT&T) C compiler for the bootstrap rather than the GreenHills compiler because we found that the GreenHills compiler was producing bogus assembler code that the assembler refused to eat. The PCC compiler is called "/lib/cc" while the GreenHills compiler is called "/bin/cc". Using the i386-sysv configuration and the /lib/cc compiler as the bootstrap compiler, we had zero problems. // rfg