I updated ncurses to version 5.4-r6 on 2005-06-05. Since then, I cannot run top at all and nano often refuses to start. I cannot explain why nano work sometimes. Reproducible: Sometimes Steps to Reproduce: 1. top; nano Actual Results: The follwing was printed on the console: Jun 6 18:09:40 [kernel] grsec: signal 11 sent to /bin/nano[nano:32435] uid/euid:0/0 gid/egid:0/0, parent /bin/bash[bash:4683] uid/euid:0/0 gid/egid:0/0 Jun 6 18:09:40 [kernel] grsec: denied resource overstep by requesting 4096 for RLIMIT_CORE against limit 0 for /bin/nano[nano:32435] uid/euid:0/0 gid/egid:0/0, parent /bin/bash[bash:4683] uid/euid:0/0 gid/egid:0/0 Segmentation fault Expected Results: Nano and top should start correctly. Gentoo Base System version 1.4.16 Portage 2.0.51.19 (uclibc/x86/hardened, gcc-3.3.5-20050130, uclibc-0.9.27-r0, 2.4.30-hardened-r1 i686) ================================================================= System uname: 2.4.30-hardened-r1 i686 AMD Athlon(tm) XP 2500+ Python: dev-lang/python-2.3.5 [2.3.5 (#1, May 21 2005, 20:49:38)] dev-lang/python: 2.3.5 sys-apps/sandbox: [Not Present] sys-devel/autoconf: 2.13, 2.59-r6 sys-devel/automake: 1.9.5, 1.7.9-r1, 1.5, 1.6.3, 1.4_p6, 1.8.5-r3 sys-devel/binutils: 2.15.92.0.2-r10 sys-devel/libtool: 1.5.16 virtual/os-headers: 2.4.22-r1 ACCEPT_KEYWORDS="x86" AUTOCLEAN="yes" CFLAGS="-pipe -O2 -march=pentium3 -mcpu=athlon-xp -fomit-frame-pointer" CHOST="i686-pc-linux-uclibc" CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3/share/config /usr/lib/X11/xkb /usr/share/config /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/env.d" CXXFLAGS="-pipe -O2 -march=pentium3 -mcpu=athlon-xp -fomit-frame-pointer -fconserve-space" DISTDIR="/usr/portage/distfiles" FEATURES="autoaddcvs autoconfig buildpkg ccache digest distlocks noclean nodoc noinfo sandbox sfperms strict userpriv usersandbox" GENTOO_MIRRORS="http://gentoo.osuosl.org http://www.ibiblio.org/pub/Linux/distributions/gentoo http://adelie.polymtl.ca/ http://gentoo.seren.com/gentoo ftp://gentoo.mirrors.tds.net/gentoo" LANG="fr_CA" LC_ALL="fr_CA" LDFLAGS="-Wl,-O1 -Wl,--sort-common" MAKEOPTS="-j2 -s" PKGDIR="/usr/local/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.namerica.gentoo.org/gentoo-portage" USE="x86 16bit 3dfx 3dnow 3dnowext a52 aac apm asm async bidi bitmap-fonts blender-game bmp bootsplash bzip2 cap cdparanoia cdr chroot cjk cracklib crypt curl dga dhcp dlloader droproot dv dvd dvdr dvdread encode exif expat faad flac ftp gatos gd-external gif glep gpm gtk2 hal hardened hardenedphp immqt-bc innodb ipv6 jack jit jpeg justify kdeenablefinal kdexdeltas lcms libclamav lm_sensors logitech-mouse lzw mad md5sum mdb minimal mmx mmx2 mozdevelop mozp3p mozsvg mp3 mpeg4 msn multicall native ncurses noplugin ogg oggvorbis openal oscar oss pcre pdflib pic png posix readline sharedmem speex spell sql sqlite sse ssl subversion svg svgz sysvipc tcpmd5 theora threads truetype truetype-fonts uclibc ucs2 unicode usb userlocales utf8 v4l vanilla vim-pager voodoo3 vorbis xine xml2 xpm xrandr xscreensaver xv xvmc zlib userland_GNU kernel_linux elibc_uclibc" Unset: ASFLAGS, CBUILD, CTARGET, LINGUAS
Created attachment 60806 [details] strace output for top
Created attachment 60807 [details] strace output for nano (crashing)
Created attachment 60808 [details] strace output for nano (not crashing)
I also tried to downgrade to ncurses 5.4-r5 and it did not work. Maybe it is related to the GCC update I performed on the same day.
drop the unicode stuff and rebuild ncurses/nano/etc...
I emerged ncurses without unicode, then nano and it still segfaulted most of the time. I consider my system as broken. I will format the root partition this weekend and go through the 3 stages again :-(. This time, I will enforce the following rule: anything that has bootstrap or build in their IUSE must NEVER be re-emerged again.
ok i found the issue nano passes (size_t)-1 to strncmp() as the length parameter in its internal history routines default uClibc build enables arch-specific optimized string functions the i386-optimized strncmp function starts off with a 'js count' count here is an unsigned 0xFFFFFFFF ... however, the asm interprets that as a signed value and hence the sign bit is set ... so the first thing strncmp() does is check the sign bit and if it is set, exit and return 0 since nano relies on the value of strncmp to determine what to do with its memory and strncmp is always returning true, it starts to dereference memory incorrectly and kill itself workaround atm is to disable arch-specific optimized string functions
fixed in cvs
yay, it's fixed /me humps spanky
