Bugzilla Bug 23651

I am installing 1.4rc4 on an HP PC, Pentium MMX 200. 

I get all the way to running the bootstrap script, and then gcc 3.2.2 fails to
build. 

@shlib_slibdir_qual@/@shlib_base_name@.so; ln -s @shlib_so_name@.so.1
$(DESTDIR)$(slibdir)@shlib_slibdir_qual@/@shlib_base_name@.so' \
SHLIB_EXT='.so' \
SHLIB_MULTILIB='' \
SHLIB_MKMAP='/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/mkmap-symver.awk' \
SHLIB_MKMAP_OPTS='' \
SHLIB_MAPFILES='/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/libgcc-std.ver
/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/config/libgcc-glibc.ver' \
SHLIB_NM_FLAGS='-pg' \
MULTILIB_OSDIRNAMES='' \
mkinstalldirs='/bin/sh
/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/mkinstalldirs' \
  /bin/sh mklibgcc > tmp-libgcc.mk
mv tmp-libgcc.mk libgcc.mk
HEADERS="ansidecl.h  i386/i386.h i386/att.h dbxelf.h elfos.h svr4.h linux.h
i386/linux.h defaults.h" DEFINES="POSIX" \
TARGET_CPU_DEFAULT="" \
/bin/sh /var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/mkconfig.sh tconfig.h
./xgcc -B./ -B/usr/i686-pc-linux-gnu/bin/ -isystem
/usr/i686-pc-linux-gnu/include -isystem /usr/i686-pc-linux-gnu/sys-include -O2
-DIN_GCC    -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
-isystem ./include  -I. -I. -I/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc
-I/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/.
-I/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/config
-I/var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/../include  -g0
-finhibit-size-directive -fno-inline-functions -fno-exceptions
-fno-omit-frame-pointer \
   -c /var/tmp/portage/gcc-3.2.2/work/gcc-3.2.2/gcc/crtstuff.c -DCRT_BEGIN \
  -o crtbegin.o
make[2]: *** [crtbegin.o] Segmentation fault
make[2]: Leaving directory `/var/tmp/portage/gcc-3.2.2/work/build/gcc'
make[1]: *** [stage1_build] Error 2
make[1]: Leaving directory `/var/tmp/portage/gcc-3.2.2/work/build/gcc'
make: *** [bootstrap-lean] Error 2

!!! ERROR: sys-devel/gcc-3.2.2 failed.
!!! Function src_compile, Line 307, Exitcode 2
!!! (no error message)

Without it, I cannot proceed to the next step of "emerge system".

Given the machine (and my assumption that it may be resource limited) have you
checked that you are not running out of RAM/swap space, or that your
/var/tmp/portage directory is not filling the free space on your hard-drive?

cdimage gentoo # free
             total       used       free     shared    buffers     cached
Mem:         61692      54240       7452          0      11252      34712
-/+ buffers/cache:       8276      53416
Swap:       262072       4220     257852

I'm not sure how much was being used during the build. I suppose it's possible, but shouldn't it do something more graceful than simply segfaulting?

cdimage gentoo # mount
proc on /proc type proc (rw)
none on /dev type devfs (rw)
tmpfs on /mnt/.init.d type tmpfs (rw,mode=0644,size=2048k)
tmpfs on /dev/shm type tmpfs (rw)
usbdevfs on /proc/bus/usb type usbdevfs (rw)
/dev/cdroms/cdrom0 on /mnt/cdrom type iso9660 (ro)
/mnt/cdrom/livecd.cloop on /mnt/cloop type ext2 (ro)
tmpfs on / type tmpfs (rw)
/dev/hda3 on /mnt/gentoo type reiserfs (rw)
/dev/hda1 on /mnt/gentoo/boot type ext3 (rw)
proc on /mnt/gentoo/proc type proc (rw)
cdimage gentoo # df -h /dev/hda3
Filesystem            Size  Used Avail Use% Mounted on
/dev/hda3             1.7G  364M  1.3G  22% /mnt/gentoo

No problems for disk space. 

Any ideas on this issue? I've since installed RedHat 8.0 on the box with no
problems. I was going to try Gentoo 1.4rc4 again on another box, just to
compare. 

I have the same problem.... - Install gentoo 2rc4 on K6 .... (stage3) on trying
to compile the kernel i get segmentation fault, after removing the match=
parameter from CFLAGS i was able to get gcc 3.2.3 and new glibc ... but i have
the same problems now.... - cant emerge -u world - segmentation fault - cant
emerge lynx - cant compile kernel the segemntaton faults are everitime on the
same positions, but there is no pattern....  the portage dir and memory isnt
overloaded .... cu Mike 

I run several ~x86 gentoo servers and with latest updates I am getting random
segfaults on openssh, x crashes, etc.  Something bad is going on.  I have no
way to pinpoint the culprit though.  Many times rebooting helps for a while. 
Also updating to the latest openssh seems to help.  Seems to be some sort of a
binary incompatibility.

Probable reason: hardware problems. Have you tried running a 12-hour memtest86
session on these boxes?

> Probable reason: hardware problems. Have you tried running a 12-hour memtest86
> session on these boxes?

I've run memtest86, albiet not the 12-hour test. Please note that the box
in question has successfully run FreeBSD, Debian, and now RedHat. I recently
did a full kernel compile on it with no problems, using gcc3. 

Resolving as INVALID, as I honestly believe you simply have low memory (~60MB
it looks).  Even though you have almost four times as much swap as memory,
gcc can still behave strangely with low memory.  I've seen this issue on
sparc and mips as well.  gcc also will do strange things when the box in
question has low memory.  Sometimes it segfaults, sometimes it ICE's (Internal
Compiler Error), sometimes cc1 goes awry...it's a strange beast.

I recommend a minimum of 128MB of ram, and with 2.4 kernels, the rule of
thumb is Swap = 2xMemory.  You might also try removing -pipe from your CFLAGS
if it is present, as this makes gcc consume a little extra memory because
it calls gcc and the assembler together so gcc can pipe it's processed data
directly into the assembler.

Lastly, if you're really desperate, cd into the portage WORKDIR and constantly
invoke 'make' until the compile finishes.  gcc can sometimes be forced to
finish this way as by sheer luck, it will sometimes compile a source file
that previously failed due to low memory.