emerge -u world keeps failing while trying to build glibc-2.3.3_pre20040117, although it doesn't seem to fail in the same place every time. I am running a SS20 with 4x hypersparcs, 256MB RAM, and 256MB swap enabled. See 'emerge info' below. After searching for similar bugs, I did add "-fstack-protector" to my CFLAGS in make.conf; no improvement. Reproducible: Always Steps to Reproduce: 1. emerge -u world 2. 3. Actual Results: emerge terminates as follows: gcc ../sysdeps/posix/sigblock.c -c -std=gnu99 -O2 -Wall -Winline -Wstrict-prototypes -Wwrite-strings -fcall-used-g6 -freorder-blocks -fstack-protector -mcpu=supersparc -pipe -fPIC -I../include -I. -I/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/signal -I.. -I../libio -I/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere -I../sysdeps/sparc/sparc32/elf -I../linuxthreads/sysdeps/unix/sysv/linux/sparc/sparc32 -I../linuxthreads/sysdeps/unix/sysv/linux/sparc -I../linuxthreads/sysdeps/unix/sysv/linux -I../linuxthreads/sysdeps/pthread -I../sysdeps/pthread -I../linuxthreads/sysdeps/unix/sysv -I../linuxthreads/sysdeps/unix -I../linuxthreads/sysdeps/sparc/sparc32 -I../linuxthreads/sysdeps/sparc -I../sysdeps/unix/sysv/linux/sparc/sparc32 -I../sysdeps/unix/sysv/linux/sparc -I../sysdeps/unix/sysv/linux -I../sysdeps/gnu -I../sysdeps/unix/common -I../sysdeps/unix/mman -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix/sparc -I../sysdeps/unix -I../sysdeps/posix -I../sysdeps/sparc/sparc32/fpu -I../sysdeps/sparc/sparc32 -I../sysdeps/wordsize-32 -I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754/dbl-64 -I../sysdeps/sparc/sparc32/soft-fp -I../sysdeps/sparc/fpu -I../sysdeps/sparc -I../sysdeps/ieee754 -I../sysdeps/generic/elf -I../sysdeps/generic -nostdinc -isystem /usr/lib/gcc-lib/sparc-unknown-linux-gnu/3.2.3/include -isystem /usr/include -D_LIBC_REENTRANT -include ../include/libc-symbols.h -DPIC -DSHARED -o /usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/signal/sigblock.os -MD -MP -MF /usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/signal/sigblock.os.dt make[2]: *** [/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/signal/kill.os] Error 139 make[2]: *** Waiting for unfinished jobs.... make[2]: Leaving directory `/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/signal' make[1]: *** [signal/subdir_lib] Error 2 make[1]: Leaving directory `/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2' make: *** [all] Error 2 !!! ERROR: sys-libs/glibc-2.3.3_pre20040117 failed. !!! Function src_compile, Line 495, Exitcode 2 !!! (no error message) Expected Results: Finish emerge-ing glibc, of course. :^) Portage 2.0.50_pre16 (default-sparc-1.4, gcc-3.2.3, glibc-2.3.3_pre20031222-r0, 2.4.24-sparc) ================================================================= System uname: 2.4.24-sparc sparc sun4m Gentoo Base System version 1.4.3.12 ccache version 2.3 [enabled] Autoconf: sys-devel/autoconf-2.59 Automake: sys-devel/automake-1.7.8 ACCEPT_KEYWORDS="sparc ~sparc" AUTOCLEAN="yes" CFLAGS="-mcpu=supersparc -O2 -pipe -fstack-protector" CHOST="sparc-sun4m-linux-gnu" COMPILER="gcc3" CONFIG_PROTECT="/etc /usr/X11R6/lib/X11/xkb /usr/kde/2/share/config /usr/kde/3/share/config /usr/share/config /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/env.d" CXXFLAGS="-mcpu=supersparc -O2 -pipe -fstack-protector" DISTDIR="/usr/portage/distfiles" FEATURES="buildpkg ccache digest sandbox" GENTOO_MIRRORS="http://www.gtlib.cc.gatech.edu/pub/gentoo" MAKEOPTS="-j5" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/usr/portage/tmp/" PORTDIR="/usr/portage" PORTDIR_OVERLAY="" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="X berkdb crypt cups curl ethereal foomaticdb gd gdbm gnome gtk imlib jpeg libwww mbox mozilla mysql ncurses nocd pam pdflib perl png ppds python qt readline samba sasl slang sparc spell ssl tcltk tcpd tiff truetype xml xml2 zlib"
Does this consistently die at exactly the same place with the same errors?
I recommend trying to build glibc in uniprocessor mode. sparc32 SMP support (especially with 4x hypersparcs) is flakey at best in 2.4. If it still fails, try the more stable 2.3.2-r9 version of glibc.
1. I have tried it three times; the first two times, I think it ran for over 12 hours (glibc is a VERY slow build on this system :^/ ) before failing. The third, it failed much sooner. I don't THINK it's failing at the same place or even with the same error, but unfortunately I don't have the results from the previous build attempts handy for comparison. 2. I would LOVE to be able to stick with the previous version of glibc; emerge insists on trying to build this new one before anything else. Can I tell it to leave glibc alone for now (without a version-specific mask)? Failing that, how do I run in uniprocessor mode? Reboot with/out certain options?
Changing severity to normal. This version is still ~sparc and this kind of problems can be expected. By the way, I built this glibc version fine on a Ultrasparc system. Both as 32bit and 64bit version.
Here's the failure point from the latest run. I commented out MAKEOPTS="-j5" in make.conf, but for some reason I still noticed "make -j2" going by during the build. As mentioned earlier, I have 256MB physical and 256MB swap. The swap was virtually untouched during the entire build. /usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/elf/ld-linux.so.2 --library-path /usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/math:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/elf:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/dlfcn:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/nss:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/nis:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/rt:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/resolv:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/crypt:/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/linuxthreads /usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere/timezone/zic -d /usr/portage/tmp//portage/glibc-2.3.3_pre20040117/image//usr/share/zoneinfo -L /dev/null -y ./yearistype southamerica make[2]: *** [/usr/portage/tmp//portage/glibc-2.3.3_pre20040117/image//usr/share/zoneinfo/America/Buenos_Aires] Illegal instruction make[2]: *** Waiting for unfinished jobs.... make[2]: Leaving directory `/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/timezone' make[1]: *** [timezone/subdir_install] Error 2 make[1]: Leaving directory `/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2' make: *** [install] Error 2 make: Leaving directory `/usr/portage/tmp/portage/glibc-2.3.3_pre20040117/work/glibc-2.3.2/buildhere' !!! ERROR: sys-libs/glibc-2.3.3_pre20040117 failed. !!! Function src_install, Line 512, Exitcode 2 !!! (no error message)
Try with a uniprocessor kernel as Joshua said in a previous comment. And also, *PLEASE* don't raise the severity since glibc-2.3.3_pre is a "pre" as it's implied, and won't go to stable until at least glibc 2.3.3 final is out. Thanks.
I apologize for inadvertently resetting the severity. When I posted my comment, the page I was on had the severity set at "blocker"; I was surprised myself to find that the severity had been set back when I posted.
Well, I compiled a SP kernel, booted it, and emerged glibc-2.3.3_pre20040117 . This time it completed successfully. I guess I've learned that I'll have to boot into SP any time a new glibc is released.
Hey, I just wanted to add a note... I had left the MAKEOPTS="-j5" line commented out in make.conf since the last time I used the UP kernel to build glibc (see my previous comment). In the process of trying to emerge a couple packages by name (to avoid glibc; was going to boot back into the UP kernel for that build), emerge decided to go ahead and build glibc-2.3.3_pre20040207 between the other two. I decided to let it go ahead and see what would happen (where it would fail). It ran for over a day, and to my great surprise, it actually finished emerging all three of the packages! So I don't know if glibc's latest version has been fixed for SMP kernels, or if the default setting of MAKEOPTS="-j2" made it work... but it did. :^)
