I did a world rebuild the other day, and the only package that failed to build (and continues to fail to build) is procps. Reproducible: Always Steps to Reproduce: 1. emerge any rev of gcc-3.4.3 2. emerge procps 3. watch it fail Actual Results: nephiliah root # emerge procps Calculating dependencies ...done! >>> emerge (1 of 1) sys-apps/procps-3.2.3-r1 to / >>> md5 src_uri ;-) procps-3.2.3.tar.gz >>> Unpacking source... >>> Unpacking procps-3.2.3.tar.gz to /var/tmp/portage/procps-3.2.3-r1/work >>> Source unpacked. gcc -D_GNU_SOURCE -I proc -I/usr/include/ncurses -MM -MG proc/alloc.c proc/devname.c proc/escape.c proc/ksym.c proc/pwcache.c proc/readproc.c proc/sig.c proc/slab.c proc/status.c proc/sysinfo.c proc/version.c proc/whattime.c > proc/.depend gcc -D_GNU_SOURCE -I proc -I/usr/include/ncurses -fno-common -ffast-math -W -Wall -Wshadow -Wcast-align -Wredundant-decls -Wbad-function-cast -Wcast-qual -Wwrite-strings -Waggregate-return -Wstrict-prototypes -Wmissing-prototypes -O1 -pipe -mcpu=G4 -mabi=altivec -malign-natural -ftracer -fsched-spec -fstrength-reduce -frerun-loop-opt -falign-loops -fpeel-loops -fno-delayed-branch -fschedule-insns -freduce-all-givs -fcse-follow-jumps -fexpensive-optimizations -Wdeclaration-after-statement -Wpadded -Wstrict-aliasing -frename-registers -fomit-frame-pointer -fno-inline-functions -m64 -c -o uptime.o uptime.c gcc -c -D_GNU_SOURCE -I proc -I/usr/include/ncurses -fno-common -ffast-math -W -Wall -Wshadow -Wcast-align -Wredundant-decls -Wbad-function-cast -Wcast-qual -Wwrite-strings -Waggregate-return -Wstrict-prototypes -Wmissing-prototypes -O1 -pipe -mcpu=G4 -mabi=altivec -malign-natural -ftracer -fsched-spec -fstrength-reduce -frerun-loop-opt -falign-loops -fpeel-loops -fno-delayed-branch -fschedule-insns -freduce-all-givs -fcse-follow-jumps -fexpensive-optimizations -Wdeclaration-after-statement -Wpadded -Wstrict-aliasing -frename-registers -fomit-frame-pointer -fno-inline-functions -m64 -DSHARED=1 -fpic proc/alloc.c -o proc/alloc.o gcc -c -D_GNU_SOURCE -I proc -I/usr/include/ncurses -fno-common -ffast-math -W -Wall -Wshadow -Wcast-align -Wredundant-decls -Wbad-function-cast -Wcast-qual -Wwrite-strings -Waggregate-return -Wstrict-prototypes -Wmissing-prototypes -O1 -pipe -mcpu=G4 -mabi=altivec -malign-natural -ftracer -fsched-spec -fstrength-reduce -frerun-loop-opt -falign-loops -fpeel-loops -fno-delayed-branch -fschedule-insns -freduce-all-givs -fcse-follow-jumps -fexpensive-optimizations -Wdeclaration-after-statement -Wpadded -Wstrict-aliasing -frename-registers -fomit-frame-pointer -fno-inline-functions -m64 -DSHARED=1 -fpic proc/devname.c -o proc/devname.o proc/alloc.c: In function `xcalloc': proc/alloc.c:21: internal compiler error: in print_operand_address, at config/rs6000/rs6000.c:9678 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://bugs.gentoo.org/> for instructions. {standard input}: Assembler messages: {standard input}:0: Warning: end of file not at end of a line; newline inserted {standard input}:34: Error: syntax error; found ` proc/devname.c:52: warning: padding struct size to alignment boundary Preprocessed source stored into /var/tmp/portage/procps-3.2.3-r1/temp/cc3lgJYv.out file, please attach this to your bugreport. make: *** [proc/alloc.o] Error 1 make: *** Waiting for unfinished jobs.... proc/devname.c: In function `load_drivers': proc/devname.c:107: internal compiler error: in print_operand_address, at config/rs6000/rs6000.c:9678 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://bugs.gentoo.org/> for instructions. {standard input}: Assembler messages: {standard input}:0: Warning: end of file not at end of a line; newline inserted {standard input}:40: Error: syntax error; found ` Preprocessed source stored into /var/tmp/portage/procps-3.2.3-r1/temp/ccNl4YVP.out file, please attach this to your bugreport. make: *** [proc/devname.o] Error 1 !!! ERROR: sys-apps/procps-3.2.3-r1 failed. !!! Function src_compile, Line 40, Exitcode 2 !!! (no error message) !!! If you need support, post the topmost build error, NOT this status message. Expected Results: The software should've actually built. :-) I'm actually writing an article on CFLAGS on ppc at this very moment, so I'm pretty sure it's not my CFLAGS. (Actually I'm very sure, 'cause even with CFALGS="" this still fails). But specifically, mine are: --- CFLAGS="-O1 -pipe -mcpu=G4 -mabi=altivec -malign-natural -ftracer -fsched-spec -fstrength-reduce -frerun-loop-opt -falign-loops -fpeel-loops -fno-delayed-branch -fschedule-insns -freduce-all-givs -fcse-follow-jumps -fexpensive-optimizations" --- ... this is only part of -O2 and -O3. If you notice, the build adds in "-m64" for some reason. Here is my current emerge info, but be aware I tried this with CFLAGS="" as already stated, and gcc-3.4.3 and gcc-3.4.3-r1 (testing gcc-3.4.3-r1 atm, and, as I said, _every_ other ebuild on the system built with it). ---Portage 2.0.51-r7 (default-linux/ppc/2004.3, gcc-3.4.3, glibc-2.3.4.20041102-r0, 2.6.9-gentoo-r8-patches1 ppc) ================================================================= System uname: 2.6.9-gentoo-r8-patches1 ppc 7447A, altivec supported Gentoo Base System version 1.4.16 Python: dev-lang/python-2.3.3 [2.3.3 (#1, Dec 1 2004, 23:32:43)] dev-lang/python: 2.3.3 sys-devel/autoconf: 2.59-r5 sys-devel/automake: 1.8.5-r1 sys-devel/binutils: 2.15.90.0.3-r3 sys-devel/libtool: 1.5.2-r7 virtual/os-headers: 2.6.8.1-r1 ACCEPT_KEYWORDS="ppc" AUTOCLEAN="yes" CFLAGS="-O1 -pipe -mcpu=G4 -mabi=altivec -malign-natural -ftracer -fsched-spec -fstrength-reduce -frerun-loop-opt -falign-loops -fpeel-loops -fno-delayed-branch -fschedule-insns -freduce-all-givs -fcse-follow-jumps -fexpensive-optimizations" CHOST="powerpc-unknown-linux-gnu" CONFIG_PROTECT="/etc /usr/X11R6/lib/X11/xkb /usr/kde/2/share/config /usr/kde/3/share/config /usr/share/config /usr/share/texmf/dvipdfm/config/ /usr/share/texmf/dvips/config/ /usr/share/texmf/tex/generic/config/ /usr/share/texmf/tex/platex/config/ /usr/share/texmf/xdvi/ /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/env.d" CXXFLAGS="-O1 -pipe -mcpu=G4 -mabi=altivec -malign-natural -ftracer -fsched-spec -fstrength-reduce -frerun-loop-opt -falign-loops -fpeel-loops -fno-delayed-branch -fschedule-insns -freduce-all-givs -fcse-follow-jumps -fexpensive-optimizations" DISTDIR="/usr/portage/distfiles" FEATURES="autoaddcvs autoconfig ccache cvs distlocks nostrip sandbox sfperms userpriv usersandbox" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/Linux/distributions/gentoo" MAKEOPTS="-j2" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/home/armando/Projects/GENTOO/gentoo-x86" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="X aalib alsa altivec berkdb bitmap-fonts cdr cups dvd f77 flac fortran gdbm gif gnustep gpm gstreamer gtk gtk2 guile imagemagick imlib ipv6 jack jpeg kerberos ldap libwww mad mysql ncurses nptl nptlonly oggvorbis opengl oss pam pdflib perl png postgres ppc python readline samba snmp speex sqlite ssl tetex tiff truetype unicode xml xml2 zlib" --- Marking the bug a "Major" only 'cause this is procps we're talking about -- it shouldn't be failing!
Created attachment 45214 [details, diff] Allows build to complete on ppc The procps Makefile is broken. This simple patch cures the symptoms (build failing) but not the bug (Makefile being broken).
This is the code that failing: --snip-- check_gcc = $(shell if $(CC) $(ALL_CPPFLAGS) $(ALL_CFLAGS) dummy.c $(ALL_LDFLAGS) $(1) -o /dev/null $(CURSES) > /dev/null 2>&1; then echo "$(1)"; else echo "$(2)"; fi ;) --snip-- ... and then, --snip-- m64 := $(call check_gcc,-m64,) ALL_CFLAGS += $(m64) --snip-- This may not even be particular to gcc-3.4.3, but possibly a kernel version, bash version, etceteras.
Just for fun, let's exercise the "64-bit test" the Makefile uses, in a fresh untarred directory of the sources: armando@nephiliah procps-3.2.3 $ for i in `ls -1 --color=never *.c`; do if gcc $i -m64 -o /dev/null > /dev/null 2>&1; then echo "$i: true"; else echo "$i: false"; fi; done dummy.c: true free.c: false minimal.c: true pgrep.c: false pmap.c: false skill.c: false slabtop.c: false sysctl.c: false tload.c: false top.c: false uptime.c: false vmstat.c: false w.c: false watch.c: false ... so that's 2 false positives out of 14, and one of the false positives is supposed to be the test for all this to begin with! *sigh* I really don't know why I find this bug so damned annoying and yet amusing.
procps-3.2.4 has a fix in the ebuild for this already, but procps-3.2.3-r1 is still marked as stable for ppc, and 3.2.4 is not
procps-3.2.4 marked ppc stable.
The Makefile code is derived from code found in the kernel Makefile. It had better work. It's interesting that your gcc randomly accepts options that don't work. With gcc on a non-Gentoo box, you'd get a reproducable message sent to stderr and a proper exit code. Here is the error message: cc1: invalid option `64' I don't know how you ended up with a gcc that doesn't bother to sanity check the command options, but you need to get that fixed ASAP. The kernel Makefile relys on very similar code.
