| Summary: | www-client/mozilla-firefox-3.5.5. crashes on startup: jemalloc.c:6224: Failed assertion: "((alignment - 1) & alignment) == 0 && alignment >= sizeof(void *)" | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Aniruddha Shankar <k> |
| Component: | Current packages | Assignee: | Mozilla Gentoo Team <mozilla> |
| Status: | RESOLVED TEST-REQUEST | ||
| Severity: | normal | CC: | Arsen.Shnurkov, c.apeltauer, daniel.vrcic, dmartin, fcoiffie, m1k0, mg, n-roeser, polidevk.polidevk, qbasicer, robert.bradbury |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | https://bugzilla.mozilla.org/show_bug.cgi?id=526152 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | backtrace | ||
|
Description
Aniruddha Shankar
2009-11-07 12:59:47 UTC
Created attachment 209535 [details]
backtrace
*** Bug 292271 has been marked as a duplicate of this bug. *** Downgraded to mozilla-firefox-3.5.4 + net-libs/xulrunner-1.9.1.4 and they work ok so the bug is 3.5.5/1.9.1.5 specific. (In reply to comment #3) > Downgraded to mozilla-firefox-3.5.4 + net-libs/xulrunner-1.9.1.4 and they work > ok so the bug is 3.5.5/1.9.1.5 specific. > Could you try with 3.6_beta1 from the overlay? That way we can check if we need to investigate this further or just stabilize 3.5.4. Please update with ccache disabled. We can not ensure that ccache injects bad translation into jemalloc. The jemalloc code has not changed since 3.5.3 with respects to your failure. I had the same problem. Removing "debug" from the USE flags and "-g" from CFLAGS and reemerging xulrunner and firefox worked for me. I'll leave the status change for someone competent to determine. I had the exact same error and removing debug from xulrunner and firefox fixed it. Portage 2.1.7.5 (default/linux/x86/10.0/desktop, gcc-4.4.2, glibc-2.11-r0, 2.6.30-gentoo-r8-qb i686) ================================================================= System uname: Linux-2.6.30-gentoo-r8-qb-i686-Genuine_Intel-R-_CPU_T2500_@_2.00GHz-with-gentoo-2.0.1 Timestamp of tree: Tue, 17 Nov 2009 18:30:01 +0000 ccache version 2.4 [enabled] app-shells/bash: 4.0_p35 dev-java/java-config: 2.1.9-r1 dev-lang/python: 2.6.4, 3.1.1-r1 dev-util/ccache: 2.4-r8 dev-util/cmake: 2.8.0 sys-apps/baselayout: 2.0.1 sys-apps/openrc: 0.5.2-r2 sys-apps/sandbox: 2.2 sys-devel/autoconf: 2.13, 2.63-r1 sys-devel/automake: 1.7.9-r1, 1.9.6-r2, 1.10.2, 1.11 sys-devel/binutils: 2.20 sys-devel/gcc-config: 1.4.1 sys-devel/libtool: 2.2.6b virtual/os-headers: 2.6.30-r1 ACCEPT_KEYWORDS="x86 ~x86" ACCEPT_LICENSE="* -@EULA" CBUILD="i686-pc-linux-gnu" CFLAGS="-march=native -mfpmath=sse -O2 -pipe -fomit-frame-pointer -fno-ident" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/env.d /etc/env.d/java/ /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/revdep-rebuild /etc/sandbox.d /etc/terminfo /etc/texmf/language.dat.d /etc/texmf/language.def.d /etc/texmf/updmap.d /etc/texmf/web2c /etc/udev/rules.d" CXXFLAGS="-march=native -mfpmath=sse -O2 -pipe -fomit-frame-pointer -fno-ident" DISTDIR="/usr/portage/distfiles" FEATURES="assume-digests ccache distlocks fixpackages news parallel-fetch protect-owned sandbox sfperms strict unmerge-logs unmerge-orphans userfetch" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/linux/distributions/gentoo" LANG="en_GB.UTF-8" LDFLAGS="-Wl,-O1" LINGUAS="en en_GB" MAKEOPTS="-j2" PKGDIR="/usr/portage/packages" PORTAGE_CONFIGROOT="/" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage/layman/kde-sunset" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="X a52 aac aalib acl acpi alsa amr apache2 apm ares aspell autoipd avahi avi bash-completion bcmath berkdb bl bluetooth bonjour branding bzip2 cairo cdinstall cdparanoia cdr cg cli consolekit cpudetection cracklib crypt cups cups-arts curl dbus devil dillo divx4linux doc double-precision dri dso dts dv dvb dvd dvdr dvdread eds emboss encode esd evo exif extrafilters extras faac faad fam fame fat ffmpeg fftw firefox flac foomaticdb fortran fuse gdbm gif gimp git glitz glut gnome gphoto2 gpm gstreamer gtk hal hfs iconv ieee1394 imap imlib innodb ipod ipv6 jadetex java jpeg jpeg2k kde lame laptop latex ldap libnotify lm_sensors lua mad mdnsresponder-compat mikmod mime mmap mmx modules mono moonlight mp2 mp3 mp4 mpeg mplayer msn mudflap musicbrainz mysql ncurses network nls nptl nptlonly nsplugin ntfs nvidia offensive ogg openal openexr opengl openmp pam pcre pdf perl phonon png ppds pppd python qt-static qt3 qt3support qt4 quicktime readline reflection samba sdl session slang smp spell spl sqlite sqlite3 sse sse2 ssl startup-notification subtitles subversion svg sysfs tcpd theora threads thunar tiff timidity tk truetype unicode usb v4l v4l2 visualization vorbis wifi win32codecs wxwindows x x264 x86 xcb xcomposite xft xml xorg xpm xulrunner xv xvid zeroconf zlib" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1 emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mmap_emul mulaw multi null plug rate route share shm softvol" APACHE2_MODULES="actions alias auth_basic authn_alias authn_anon authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache dav dav_fs dav_lock deflate dir disk_cache env expires ext_filter file_cache filter headers include info log_config logio mem_cache mime mime_magic negotiation rewrite setenvif speling status unique_id userdir usertrack vhost_alias" CAMERAS="adc65 agfa_cl20 aox barbie canon casio_qv clicksmart310 digigr8 digita dimagev dimera3500 directory enigma13 fuji gsmart300 hp215 iclick jamcam jd11 kodak_dc120 kodak_dc210 kodak_dc240 kodak_dc3200 kodak_ez200 konica konica_qm150 largan lg_gsm mars mustek panasonic_coolshot panasonic_dc1000 panasonic_dc1580 panasonic_l859 pccam300 pccam600 polaroid_pdc320 polaroid_pdc640 polaroid_pdc700 ptp2 ricoh ricoh_g3 samsung sierra sipix_blink sipix_blink2 sipix_web2 smal sonix sony_dscf1 sony_dscf55 soundvision spca50x sq905 stv0674 stv0680 sx330z template toshiba_pdrm11" ELIBC="glibc" INPUT_DEVICES="keyboard mouse evdev" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="en en_GB" USERLAND="GNU" VIDEO_CARDS="nvidia" Unset: CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, FFLAGS, INSTALL_MASK, LC_ALL, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS problem still valid on mozilla-firefox-3.5.6 and xulrunner-1.9.1.6 As I got a reproduceable crash with firefox, I recompiled it with "debug" flag to get a backtrace. But, now, it crashes when I start it. Is it possible to debug Firefox without the "debug" USE flag ? Same thing has happened with my firefox/xulrunner 3.5.7. Removing the "debug" use flag got rid of it.. *** Bug 301064 has been marked as a duplicate of this bug. *** Well, I'm not sure how/where gentoo gets the sources from, but I've been compiling Firefox from the CVS source off and on for a couple of years. Updating my Mozilla distributed CVS sources as of 100117, and building firefox-3.0.18pre from source, with --disable-crashreporter (which has problems compiling and building libraries) and --enable-static (which doesn't appear to "really" produce a "static" build but gets closer) and --disable-libxul (avoiding the whole xulrunner topic which is what I think is the source of the problem here -- old versions of Gentoo Firefox I believe had a -xulrunner USE option which seems to have disappeared) and I do get a working firefox installed (in /usr/local/lib/firefox-3.0.18pre. And this is on top of glib-2.22.3 (so I don't think glib has anything to do with this problem). I strongly suspect that the problem is inconsistent linking between Firefox versions and Xulrunner versions. For example right now my system (which has both Firefox and Xulrunner unlocked) seems to be bouncing around between xulrunner-1.9.1.6 and -1.9.1.7 with no indication that Firefox-3.5.6 may not work with either of them. This is an ebuild restriction problem in that Firefox may need to restrict xulrunner to "known-to-work" versions. Minimally, upgrading xulrunner after upgrading firefox may require a relinking of the firefox side of the equation (without Firefox altering the xulrunner directories). I'm going to play around with my configure directives a bit more and if I get a working semi-static version with debugging & optimization working I'll post the scripts. Those who want to try this may want to start downloading the Firefox CVS sources (the Mozilla instructions for this seem to work pretty well), but the source transfers will take a several hours over a DSL line. Be forewarned that the sources + the build seems to take up ~2.5 GB. It would be nice if someone who knows the thinking at Mozilla would confirm whether 3.5.6 is ~= 3.0.18pre. It was my impression that around 3.0.15, they simply started relabeling things as 3.5.x. (Or is there a completely different branch of the CVS tree that I'm unaware of?) Robert, cvs is no longer used to maintain current branches. They are not the same tree at all.This actually appears to be a preload failure of libc.so if you want to test `LD_PRELOAD="/lib/libc.so.6" firefox` and post your results back to the bug. Reproducable with net-libs/xulrunner-1.9.2-r1 with debug use flag set and www-client/mozilla-firefox-3.6-r1. (In reply to comment #14) > Reproducable with net-libs/xulrunner-1.9.2-r1 with debug use flag set and > www-client/mozilla-firefox-3.6-r1. Same here with a slightly different situation, it actually doesn't depend whether I have enabled or disabled debug, custom-optimization or sqlite use flag. In all cases that I have tried, firefox crashed with segfault upon startup. Besides use flag toggling, I have tried downgrading to the lastest stable, changing CFLAGS to some generic options like -march=i686, pulling up and building from the lastest ebuilds from mozilla overlay (even 3.7 alpha), neither of those changed anything... Actually the only firefox flavor that is currently working on my system is official binary packet. Daniel, I currently have two versions of Firefox working under Gentoo. One is downloaded and built using the standard (several year old) CVS procedures. That one claims to be 3.0.18pre (which I highly suspect is someplace between 3.5 and 3.6 in "official" terms). The other is a 3.6 version downloaded and built from the Mozilla source distribution. In both cases the compile and build options have been tuned by myself over the last couple of years (and are close to but not exactly the Gentoo) build options. In particular there are no "xul" options involved (the browsers are "standalone" -- part of my ongoing quest to produce a static linked browser which has no dependencies but simply works out of the box. But in summary, the problem does not appear to be in the Firefox/Mozilla source -- the problem would appear to be either in how Gentoo is putting it together or how they are integrating it with the rest of the system. That would be apparent from the initial diagnostic which is one of strangest startup errors I've ever seen (there are probably only a handful of people on the planet who understand the loader code -- and they generally don't frequent this bug database). The shell scripts to build an "as distributed" Mozilla archive are not difficult so if you want them let me know. Jory, thank you for the information (I had assumed that there was a cvs pitfall somewhere along the lines). Not a big deal since I'm comfortable dealing with large .gz files (for whatever version is current). The LD_PRELOAD options, though I understand them are a little bit less clear. It would seem easier to simply reboot the system. But in summary, it would appear that the simple solution to "running firefox" problems is "reboot the system". Not my choice since I keep my system up for weeks and maintain browser sessions over such a time frame -- but if it solves the problem and does not require a complete rebuild of FireFox -- go for it. (Just to be fair in tradeoffs -- rebuilding FireFox requires about 2-3 CPU hours on a Pentium IV Prescott @ 2.8 GHz) (In reply to comment #16) > The shell scripts to build an "as distributed" Mozilla archive are not > difficult so if you want them let me know. Few minutes later after I had confirmed this pesky gentoo bug, I managed to build a binary tarball from official source by myself. It was a simple and quite straightforward procedure, something that I wasn't aware of. ...so no, I don't need those scripts, but anyway, thank you very much for your kind reply. Here's the mozilla thread on this bug: https://bugzilla.mozilla.org/show_bug.cgi?id=526152 they also have a patch, which I tested, and can confirm works. Same issue for me. net-libs/xulrunner-1.9.1.8 www-client/mozilla-firefox-3.5.8 I have debug in the USE and -ggdb option in CFLAGS dmartin@mobi ~ $ firefox jemalloc.c:5832: Failed assertion: "((alignment - 1) & alignment) == 0 && alignment >= sizeof(void *)" Program firefox (pid = 17984) received signal 6. Stack: gsignal+0x00000035 [/lib/libc.so.6 +0x00032315] abort+0x00000181 [/lib/libc.so.6 +0x00033811] memalign+0x00000087 [firefox +0x00021CFA] UNKNOWN [/lib64/ld-linux-x86-64.so.2 +0x0001044D] __tls_get_addr+0x00000050 [/lib64/ld-linux-x86-64.so.2 +0x000109C0] UNKNOWN [/usr/lib64/xulrunner-1.9.1/libxul.so +0x01563675] NS_InitXPCOM3_P+0x000000FD [/usr/lib64/xulrunner-1.9.1/libxul.so +0x015061F7] UNKNOWN [/usr/lib64/xulrunner-1.9.1/libxul.so +0x004A514C] UNKNOWN [/usr/lib64/xulrunner-1.9.1/libxul.so +0x004A52DC] UNKNOWN [/usr/lib64/xulrunner-1.9.1/libxul.so +0x004A74BF] XRE_main+0x0000126F [/usr/lib64/xulrunner-1.9.1/libxul.so +0x004A9745] UNKNOWN [firefox +0x00002BF8] __libc_start_main+0x000000E6 [/lib/libc.so.6 +0x0001EA26] UNKNOWN [firefox +0x00002029] Sleeping for 300 seconds. Type 'gdb firefox 17984' to attach your debugger to this thread. The firefox bug (#526152) pretty much explains what the problem is. One could presumably apply the patch as others suggest. Looking at assert.h also suggests that one could get around it by defining NDEBUG in CFLAGS before emerging xulrunner and/or mozilla-firefox (though this would nullify all assertions in the build). That approach at least avoids having to muck around with patches and rebuilding Manifest files. (In reply to comment #21) > The firefox bug (#526152) pretty much explains what the problem is. One could > presumably apply the patch as others suggest. Looking at assert.h also > suggests that one could get around it by defining NDEBUG in CFLAGS before > emerging xulrunner and/or mozilla-firefox (though this would nullify all > assertions in the build). That approach at least avoids having to muck around > with patches and rebuilding Manifest files. > Patch is avaliable via mozilla overlay, we will move it to tree with next revision bump. 3.5.x is coming to end of life, I expect to get ready for 3.6 to become stable in next couple of weeks. *** Bug 303935 has been marked as a duplicate of this bug. *** (In reply to comment #22) > (In reply to comment #21) > > The firefox bug (#526152) pretty much explains what the problem is. One could > > presumably apply the patch as others suggest. Looking at assert.h also > > suggests that one could get around it by defining NDEBUG in CFLAGS before > > emerging xulrunner and/or mozilla-firefox (though this would nullify all > > assertions in the build). That approach at least avoids having to muck around > > with patches and rebuilding Manifest files. > > > > Patch is avaliable via mozilla overlay, we will move it to tree with next > revision bump. 3.5.x is coming to end of life, I expect to get ready for 3.6 to > become stable in next couple of weeks. > This bug is still crawling in mozilla-firefox-3.6-r4! x86_64 here. (In reply to comment #24) > (In reply to comment #22) > > (In reply to comment #21) > > > The firefox bug (#526152) pretty much explains what the problem is. One could > > > presumably apply the patch as others suggest. Looking at assert.h also > > > suggests that one could get around it by defining NDEBUG in CFLAGS before > > > emerging xulrunner and/or mozilla-firefox (though this would nullify all > > > assertions in the build). That approach at least avoids having to muck around > > > with patches and rebuilding Manifest files. > > > > > > > Patch is avaliable via mozilla overlay, we will move it to tree with next > > revision bump. 3.5.x is coming to end of life, I expect to get ready for 3.6 to > > become stable in next couple of weeks. > > > > This bug is still crawling in mozilla-firefox-3.6-r4! > x86_64 here. > This is fixed in 3.6-r5. Sorry for the delay. I will call for xulrunner-1.9.2/firefox-3.6 stabilization in 2 weeks. (In reply to comment #25) > (In reply to comment #24) > > (In reply to comment #22) > > > (In reply to comment #21) > > > > The firefox bug (#526152) pretty much explains what the problem is. One could > > > > presumably apply the patch as others suggest. Looking at assert.h also > > > > suggests that one could get around it by defining NDEBUG in CFLAGS before > > > > emerging xulrunner and/or mozilla-firefox (though this would nullify all > > > > assertions in the build). That approach at least avoids having to muck around > > > > with patches and rebuilding Manifest files. > > > > > > > > > > Patch is avaliable via mozilla overlay, we will move it to tree with next > > > revision bump. 3.5.x is coming to end of life, I expect to get ready for 3.6 to > > > become stable in next couple of weeks. > > > > > > > This bug is still crawling in mozilla-firefox-3.6-r4! > > x86_64 here. > > > > This is fixed in 3.6-r5. Sorry for the delay. I will call for > xulrunner-1.9.2/firefox-3.6 stabilization in 2 weeks. > Thank you! |
