gcc -DHAVE_CONFIG_H -I. -I. -I.. -I.. -I.. -Wall -Wunused -Wchar-subscripts -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wpointer-arith -Wsign-promo -Wno-sign-compare -DORBIT2=1 -D_REENTRANT -I/usr/include/orbit-2.0 -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/bonobo-activation-2.0 -I/usr/include/libxml2 -DPLUGIN_DIR=\"/usr/lib/bonobo/plugin\" -DBONOBO_LIBDIR=\"/usr/lib\" -DBONOBO_DATADIR=\"/usr/share\" -DBONOBO_PIXMAPDIR=\"/usr/share/pixmaps\" -DBONOBO_BINDIR=\"/usr/bin\" -DBONOBO_LOCALSTATEDIR=\"/var\" -DBONOBO_LOCALEDIR=\"\" -DG_LOG_DOMAIN=\"Bonobo\" -DVERSION=\"1.102.0\" -O2 -march=i386 -mcpu=i686 -c Bonobo-skels.c -fPIC -DPIC -o .libs/Bonobo-skels.lo gcc: Internal error: Bus error (program cpp0) Please submit a full bug report. OK, reproducing this is not going to be fun. Basically, try to build ~hp/libbonobo-1.102.0-2.src.rpm using all the build prereqs from the unstable tree. I don't know how to come up with a better test case... I can't send you the preprocessed output as I would for a compiler bug for obvious reasons. I'm using gcc-2.96-98
OK, de-prioritize this. It only happens with parallel makes, because the Bonobo build is in some hosed state and headers are missing or something. So cpp has already encountered some other errors before it crashes, I think. cpp still shouldn't crash of course but it may not be worth the pain of installing all these packages on an SMP system to see what the issue is.
Actually if some other process truncates/modifies one of the headers used during preprocessing, Bus error may happen even without cpp0 being on fault (cpp0 uses mmap for larger headers and if somebody truncates the file and you access the part beyond the end, you get SIGBUS). So I'd be first of all interested on what is bonobo doing with the headers used by cpp0.
GCC 2.96-RH is too old now and I certainly don't remember seeing something like this with contemporary preprocessors.