Konstantin Khlebnikov wrote on 19/03/15 02:38:
On 11.01.2015 14:26, Arthur Marsh wrote:
Hi, I hit the following when resetting my ADSL modem, which dropped the
Ethernet link on this pc using the current Linus' git head kernel
compiled for X86-64 in 32 bit mode:
Ethernet controller is identified as:
00:12.0 Ethernet controller: VIA Technologies, Inc. VT6102 [Rhine-II]
(rev 7c)
[ 0.000000] Initializing cgroup subsys cpuset
[ 0.000000] Initializing cgroup subsys cpu
[ 0.000000] Initializing cgroup subsys cpuacct
[ 0.000000] Linux version 3.19.0-rc3+ (root@am64) (gcc version 4.9.2
(Debian 4.9.2-10) ) #1453 SMP PREEMPT Sat Jan 10 19:21:40 ACDT 2015
[62178.076871] via-rhine 0000:00:12.0 eth0: Reset not complete yet.
Trying harder.
[62178.077380] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[62358.924028] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[62590.593390] ------------[ cut here ]------------
[62590.593803] kernel BUG at mm/rmap.c:399!
[62590.594140] invalid opcode: 0000 [#1] PREEMPT SMP
[62590.594583] Modules linked in: dm_mod cpuid snd_hrtimer nfc
cpufreq_stats cpufreq_conservative cpufreq_powersave cpufreq_userspace
bnep binfmt_misc nfnetlink_queue nfnetlink_log nfnetlink bluetooth
rfkill nls_utf8 nls_cp437 vfat fat hwmon_vid tun snd_emu10k1_synth
snd_emux_synth snd_seq_midi_emul snd_seq_virmidi snd_seq_midi_event
snd_seq cuse fuse lp uas usb_storage ppdev radeon snd_emu10k1
snd_util_mem snd_hwdep snd_rawmidi snd_seq_device snd_ac97_codec snd_pcm
ttm snd_timer drm_kms_helper psmouse snd evdev pcspkr serio_raw
soundcore i2c_viapro ac97_bus k8temp emu10k1_gp gameport drm
i2c_algo_bit asus_atk0110 parport_pc parport button shpchp processor
thermal_sys ext4 mbcache crc16 jbd2 sr_mod cdrom ata_generic sg sd_mod
eata firewire_ohci firewire_core crc_itu_t ahci libahci via_rhine mii
pata_via
[62590.596016] uhci_hcd ehci_pci ehci_hcd usbcore usb_common libata
scsi_mod
[62590.596016] CPU: 0 PID: 16909 Comm: midori Not tainted 3.19.0-rc3+
#1453
[62590.596016] Hardware name: System manufacturer System Product
Name/A8V-MX, BIOS 0503 12/06/2005
[62590.596016] task: f45bd530 ti: e585a000 task.ti: e585a000
[62590.596016] EIP: 0060:[<c1157614>] EFLAGS: 00010286 CPU: 0
[62590.596016] EIP is at unlink_anon_vmas+0x134/0x1a0
[62590.596016] EAX: f3b107c0 EBX: ed2763d4 ECX: 00000018 EDX: e4ac69a0
[62590.596016] ESI: ffffffff EDI: ed2763dc EBP: e585bebc ESP: e585bea0
[62590.596016] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
[62590.596016] CR0: 8005003b CR2: b3cf4054 CR3: 017d7000 CR4: 000007d0
[62590.596016] Stack:
[62590.596016] 0002d114 ed2763a0 ed2763dc f3b107c0 e2d49f50 ed2763a0
a7bee000 e585bee0
[62590.596016] c114bb51 00000000 a7800000 00000000 e585beec e732cf50
ea5ef2c0 00000000
[62590.596016] e585bf3c c1154492 00000000 ea5ef2c0 a5c00000 bfaa3000
00000001 e5930000
[62590.596016] Call Trace:
[62590.596016] [<c114bb51>] free_pgtables+0x81/0xf0
[62590.596016] [<c1154492>] exit_mmap+0x82/0x120
[62590.596016] [<c104afb3>] mmput+0x43/0xf0
[62590.596016] [<c10501a9>] do_exit+0x259/0xa00
[62590.596016] [<c1286bfa>] ? ___preempt_schedule+0x8/0xe
[62590.596016] [<c10509c2>] do_group_exit+0x32/0x90
[62590.596016] [<c1050a31>] SyS_exit_group+0x11/0x20
[62590.596016] [<c14ceee0>] sysenter_do_call+0x12/0x12
[62590.596016] Code: 42 08 00 01 10 00 c7 42 0c 00 02 20 00 e8 b5 1f 01
00 8b 43 08 8d 48 f8 8d 43 08 39 c6 74 38 8b 43 04 89 da 8b 58 4c 85 db
74 bc <0f> 0b 66 90 89 55 f0 e8 40 fe ff ff 8b 55 f0 eb b3 8b 45 e8 c7
[62590.596016] EIP: [<c1157614>] unlink_anon_vmas+0x134/0x1a0 SS:ESP
0068:e585bea0
[62590.871873] ---[ end trace 03349ef15ff73606 ]---
[62590.871881] Fixing recursive fault but reboot is needed!
This and other mmamp related problems appear to have surfaced in the
Linus' git head kernel in the last few days.
I'm happy to supply further information or run tests to help identify
the source of the problem.
More likely this is already fixed in v3.19-rc4 by:
b800c91a0517071156e772d4fb329ad33590da62
(mm: fix corner case in anon_vma endless growing prevention)
and there is one fix for rare error on error path in linux-mm:
(mm: fix anon_vma->degree underflow in anon_vma endless growing prevention
Thanks, I haven't seen this problem recently.
Arthur.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>