For some strange reason, there was no core dump, but dmesg reported this: [64226.321195] BUG: unable to handle kernel paging request at ffffb4ff [64226.321223] IP: [<c10b255f>] __destroy_inode+0x2b/0x62 [64226.321267] *pde = 01442067 *pte = 00000000 [64226.321289] Oops: 0000 [#1] [64226.321305] Modules linked in: mperf cpufreq_conservative cpufreq_userspace cpufreq_powersave cpufreq_stats parport_pc ppdev lp parport bnep rfcomm bluetooth rfkill uinput ecryptfs snd_cs5535audio snd_ac97_codec snd_pcm snd_page_alloc scx200_acb snd_seq i2c_core snd_seq_device ecb snd_timer cryptd snd aes_i586 soundcore ac97_bus aes_generic cs5535_mfd ac processor geode_aes geode_rng thermal_sys evdev microcode usbhid hid ext4 crc16 jbd2 mbcache msr sg sd_mod crc_t10dif ata_generic pata_cs5536 libata ohci_hcd ehci_hcd usbcore scsi_mod button usb_common 8139too 8139cp mii [last unloaded: scsi_wait_scan] [64226.321571] [64226.321589] Pid: 6204, comm: updatedb.mlocat Not tainted 3.2.0-4-486 #1 Debian 3.2.39-2 First International Computer, Inc. ION603/ION603 [64226.321626] EIP: 0060:[<c10b255f>] EFLAGS: 00210286 CPU: 0 [64226.321650] EIP is at __destroy_inode+0x2b/0x62 [64226.321669] EAX: ffffb4ff EBX: c8412cd8 ECX: f39e1e30 EDX: ffffb4fe [64226.321690] ESI: f5402698 EDI: c8412cd8 EBP: f371ec00 ESP: f39e1e48 [64226.321711] DS: 007b ES: 007b FS: 0000 GS: 00e0 SS: 0068 [64226.321733] Process updatedb.mlocat (pid: 6204, ti=f39e0000 task=f371ec00 task.ti=f39e0000) [64226.321750] Stack: [64226.321761] c8412cd8 c10b2869 c8413318 c10b0caf c8413318 f39e1ec4 00000000 c10b0fa8 [64226.321803] f39e1e84 c10aa27b 00000000 c10ab987 00000041 f5812080 c8be61d8 f5812080 [64226.321844] c8413318 00000000 f39e1ec4 ffffff9c ffffff9c f34ac000 c10aba07 f39e1ec4 [64226.321886] Call Trace: [64226.321912] [<c10b2869>] ? destroy_inode+0x1a/0x3e [64226.321937] [<c10b0caf>] ? dentry_kill+0x82/0x8f [64226.321962] [<c10b0fa8>] ? dput+0x8e/0x95 [64226.321992] [<c10aa27b>] ? path_put+0xb/0x13 [64226.322016] [<c10ab987>] ? path_lookupat+0x1f1/0x258 [64226.322041] [<c10aba07>] ? do_path_lookup+0x19/0x73 [64226.322066] [<c10acebf>] ? user_path_at_empty+0x3d/0x64 [64226.322194] [<f86e0136>] ? ext4_readdir+0x1a5/0x55e [ext4] [64226.322280] [<f86e0143>] ? ext4_readdir+0x1b2/0x55e [ext4] [64226.322306] [<c10acef1>] ? user_path_at+0xb/0xe [64226.322332] [<c10a67d0>] ? vfs_fstatat+0x35/0x5c [64226.322357] [<c10a681c>] ? vfs_stat+0x10/0x12 [64226.322381] [<c10a69d9>] ? sys_stat64+0xe/0x21 [64226.322408] [<c1028ff5>] ? ns_capable+0x29/0x3d [64226.322433] [<c10a9fee>] ? path_get+0xa/0x13 [64226.322462] [<c10be412>] ? set_fs_pwd+0x3b/0x42 [64226.322495] [<c10a3037>] ? sys_fchdir+0x54/0x67 [64226.322528] [<c1282c5c>] ? syscall_call+0x7/0xb [64226.322543] Code: 53 89 c3 e8 1a ef 00 00 85 c0 74 02 0f 0b 89 d8 e8 ac 1b 04 00 89 d8 e8 88 47 01 00 8b 43 10 8d 50 ff 83 fa fd 77 14 85 c0 74 10 <ff> 08 0f 94 c2 84 d2 74 07 31 d2 e8 ad d0 fa ff 8b 43 14 8d 50 [64226.322780] EIP: [<c10b255f>] __destroy_inode+0x2b/0x62 SS:ESP 0068:f39e1e48 [64226.322813] CR2: 00000000ffffb4ff [64226.322831] ---[ end trace 33adc6f3acd9b544 ]---
2013/3/31 Tollef Fog Heen <tfh...@debian.org> > ]] Martin-Éric Racine > > > 2013/3/31 Tollef Fog Heen <tfh...@debian.org>: > > > ]] Martin-Éric Racine > > > > > >> I'm assuming that it's CPU-specific, since that particular host has a > > >> Geode LX and the issue doesn not appear on any of my other hosts > > >> running other x86 variants. > > > > > > Interesting. Any chance you can get a core dump and backtrace from the > > > mlocate run so we can figure out whether this is a toolchain bug or > not? > > > > I welcome instructions on how to achieve this. The mlocate run only > > fails whenever executed from the cron job. > > Rebuild the package with DEB_BUILD_OPTIONS="debug nostrip" so you get > debug symlbols. Install it. > > Try echo "/tmp/core" > /proc/sys/kernel/core_pattern and then wait for > an error. Once you have that, you should have a core dump in /tmp, run > gdb /usr/bin/mlocate /tmp/core.$PID (or what the core file is named), > run bt full and mail that to the bug. > > -- > Tollef Fog Heen > UNIX is user friendly, it's just picky about who its friends are >