On 2010年11月04日 06:01, Eric Dumazet wrote:
quoted text
> Le mercredi 03 novembre 2010 à 14:21 -0700, Randy Dunlap a écrit :
>> Maybe this isn't normal usage:  just modprobe cls_cgroup && rmmod cls_cgroup:
>>
>>
>> [  107.806607] ------------[ cut here ]------------
>> [  107.810180] kernel BUG at /local/linsrc/lnx-2637-rc1/kernel/cgroup.c:3855!
>> [  107.810180] invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC
>> [  107.822274] last sysfs file: /sys/devices/pci0000:00/0000:00:1d.1/usb3/3-1/3-1.3/devnum
>> [  107.824889] CPU 0 
>> [  107.832854] Modules linked in: cls_cgroup(-) ipt_MASQUERADE iptable_nat nf_nat af_packet nfsd lockd nfs_acl auth_rpcgss exportfs sco bridge stp llc bnep l2cap crc16 bluetooth rfkill sunrpc ipt_REJECT nf_conntrack_ipv4 nf_defrag_ipv4 iptable_filter ip_tables ip6t_REJECT xt_tcpudp nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables x_tables ipv6 p4_clockmod freq_table speedstep_lib binfmt_misc dm_mirror dm_region_hash dm_log dm_multipath scsi_dh dm_mod kvm uinput mousedev joydev snd_intel8x0 snd_ac97_codec ac97_bus usbmouse snd_seq snd_seq_device usbkbd usbhid snd_pcm ppdev hid tg3 led_class snd_timer dcdbas sr_mod snd iTCO_wdt cdrom iTCO_vendor_support sg rtc_cmos pcspkr soundcore i2c_i801 rng_core snd_page_alloc rtc_core parport_pc shpchp evdev rtc_lib parport 8250_pnp pci_hotplug mac_hid unix ide_pci_generic ide_core ata_generic pata_acpi ata_piix sd_mod crc_t10dif ext3 jbd mbcache uhci_hcd ohci_hcd ssb mmc_core pcmcia pcmcia_core firmware
_c!
quoted text
>>  lass ehci_hcd usbcore nls_base i915 drm_kms_helper intel_agp button intel_gtt video thermal_sys hwmon output [last unloaded: mperf]
>> [  107.933458] 
>> [  107.933458] Pid: 3400, comm: rmmod Not tainted 2.6.37-rc1 #7 0HH807/OptiPlex GX620               
>> [  107.937800] RIP: 0010:[<ffffffff810e6c9d>]  [<ffffffff810e6c9d>] cgroup_unload_subsys+0x64/0x1c8
>> [  107.937800] RSP: 0018:ffff88006c107ea8  EFLAGS: 00010202
>> [  107.937800] RAX: 0000000000000000 RBX: ffffffffa0009d50 RCX: 0000000000000000
>> [  107.937800] RDX: ffffffff81a3a5f0 RSI: ffff88006c107dc8 RDI: ffff88006c107e48
>> [  107.937800] RBP: ffff88006c107ec8 R08: ffffffff81a3a5f0 R09: 000000000000039a
>> [  107.937800] R10: 0000000000000001 R11: ffff88006c107e48 R12: 0000000000000000
>> [  107.937800] R13: 00007fff2664ffc0 R14: 0000000000000000 R15: 0000000000000001
>> [  107.937800] FS:  00007f52809e46f0(0000) GS:ffff88007c600000(0000) knlGS:0000000000000000
>> [  107.937800] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
>> [  107.937800] CR2: 0000003fb5a7bf20 CR3: 000000006c1d8000 CR4: 00000000000006f0
>> [  107.937800] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>> [  107.937800] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
>> [  107.937800] Process rmmod (pid: 3400, threadinfo ffff88006c106000, task ffff880075a33000)
>> [  107.937800] Stack:
>> [  107.937800]  ffff88006c107ec8 ffffffffa000a0e0 0000000000000000 00007fff2664ffc0
>> [  107.937800]  ffff88006c107ed8 ffffffffa0009819 ffff88006c107f78 ffffffff810d3cb0
>> [  108.048442]  ffffffffa000a0e0 0000000000000880 ffff88006c107f14 ffffffff8155036b
>> [  108.057485] Call Trace:
>> [  108.065148]  [<ffffffffa0009819>] exit_cgroup_cls+0x45/0x4e [cls_cgroup]
>> [  108.070071]  [<ffffffff810d3cb0>] sys_delete_module+0x2d6/0x368
>> [  108.085255]  [<ffffffff8155036b>] ? lockdep_sys_exit_thunk+0x35/0x67
>> [  108.093771]  [<ffffffff81007075>] ? xen_zap_pfn_range+0x53/0x139
>> [  108.101589]  [<ffffffff815502f5>] ? trace_hardirqs_on_thunk+0x3a/0x3f
>> [  108.111624]  [<ffffffff8100ea72>] system_call_fastpath+0x16/0x1b
>> [  108.119099] Code: 05 51 8d 71 01 0f 0b eb fe 31 f6 48 c7 c7 a0 a5 a3 81 48 ff 05 45 8d 71 01 e8 42 83 46 00 83 7b 58 07 7f 0b 48 ff 05 43 8d 71 01 <0f> 0b eb fe 48 ff 05 40 8d 71 01 48 8d bb 30 01 00 00 48 63 43 
>> [  108.145840] RIP  [<ffffffff810e6c9d>] cgroup_unload_subsys+0x64/0x1c8
>> [  108.152902]  RSP <ffff88006c107ea8>
>> [  108.161767] ---[ end trace 659fde6f8f5f2810 ]---
>>
>>
>>
>> kernel config file is attached (almost allmodconfig).
>> There may be some CONFIG options that are not helping...
>>
>> ---
> 
> commits 8e039d84b323c450 
> (cgroups: net_cls as module)
> 
> followed by commit f845172531f
> (cls_cgroup: Store classid in struct sock)
> 
> are the problem :
> 
> if CONFIG_NET_CLS_CGROUP is not defined
> 
> exit_cgroup_cls() does :
> 
> #ifndef CONFIG_NET_CLS_CGROUP
> 	net_cls_subsys_id = -1;  <<< -1
> 	synchronize_rcu();
> #endif
> 	cgroup_unload_subsys(&net_cls_subsys);
> 
> 
> but net_cls_subsys_id is an alias of net_cls_subsys.subsys_id
> 
> so putting -1 in it triggers BUG_ON() on line 3855 of kernel/cgroup.c
> 
> BUG_ON(ss->subsys_id < CGROUP_BUILTIN_SUBSYS_COUNT);
> 
> Herbert, I'll let you fix it ?
>

Exactly what I was going to reply. This bug report also reveals
another bug..

I'll post fixes for the 2 bugs in minutes.
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/