kernel bug in netback.c

kernel bug in netback.c

[William Dauchy]

Hello,

Using xen-4.1 and dom0 3.4.97, I'm getting the following trace. I was
wondering if the 3.4.x stable tree was not missing any patch.

------------[ cut here ]------------
kernel BUG at drivers/net/xen-netback/netback.c:1737!
invalid opcode: 0000 [#1] SMP
Modules linked in: ebt_arp sd_mod crc_t10dif ebt_ip6 ebt_dnat ebt_ip
ebtable_broute ebt_snat ebtable_nat ebtable_filter ebtables bridge stp
llc ipmi_devintf iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi
scsi_mod ipmi_si ipmi_msghandler aoe(O) rtc_cmos button

Pid: 1407, comm: netback/9 Tainted: G           O
3.4.97-dom0-a801fd8-i386 #68          /0D61XP
EIP: 0061:[<c124ea69>] EFLAGS: 00010287 CPU: 9
EIP is at xen_netbk_idx_release+0x109/0x120
EAX: 0000006f EBX: f5093584 ECX: 00000000 EDX: 0000006f
ESI: 0000006f EDI: 0000006f EBP: ed473ee8 ESP: ed473ecc
 DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0069
CR0: 8005003b CR2: 08acb2f8 CR3: 27013000 CR4: 00002660
DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
DR6: ffff0ff0 DR7: 00000400
Process netback/9 (pid: 1407, ti=ed472000 task=eeb0b500 task.ti=ed472000)
Stack:
 0000007f ffffffff e70eac00 00000071 f5093584 f5094e00 ffffffff ed473fb8
 c1250564 f3ec1da0 00000086 00000840 ed473fa8 c1059d04 e8ab4d40 0000007d
 ed473fa0 ed473f8c ed473f6c f3128300 e8ab4c40 eeb0b500 eeb0b500 f5094de8
Call Trace:
 [<c1250564>] xen_netbk_kthread+0x954/0xe20
 [<c1059d04>] ? idle_balance+0xd4/0x120
 [<c1049cc0>] ? wake_up_bit+0x30/0x30
 [<c124fc10>] ? xen_netbk_tx_build_gops+0xe90/0xe90
 [<c1049964>] kthread+0x74/0x80
 [<c10498f0>] ? kthread_freezable_should_stop+0x50/0x50
 [<c13c11f6>] kernel_thread_helper+0x6/0x10
Code: 8b 44 93 0c c7 40 04 00 00 00 00 8b 44 93 0c e8 de f7 e3 ff 8b
7d e4 c7 44 bb 0c 00 00 00 00 83 c4 10 5b 5e 5f c9 c3 0f 0b eb fe <0f>
0b eb fe 0f 0b eb fe eb 0d 90 90 90 90 90 90 90 90 90 90 90
EIP: [<c124ea69>] xen_netbk_idx_release+0x109/0x120 SS:ESP 0069:ed473ecc
---[ end trace b51200cc1b02fe07 ]---
vif vif-50-0: 2 reading script
p47: port 5(vif50.0) entered disabled state


-- 
William

Re: kernel bug in netback.c

[Wei Liu]

On Sat, Jul 19, 2014 at 10:41:37AM +0200, William Dauchy wrote:
> Hello,
> 
> Using xen-4.1 and dom0 3.4.97, I'm getting the following trace. I was
> wondering if the 3.4.x stable tree was not missing any patch.
> 
> ------------[ cut here ]------------
> kernel BUG at drivers/net/xen-netback/netback.c:1737!

I gusse you have some other patches applied? There is no BUG_ON at
vanilla 3.4.97 L1737.

> invalid opcode: 0000 [#1] SMP
> Modules linked in: ebt_arp sd_mod crc_t10dif ebt_ip6 ebt_dnat ebt_ip
> ebtable_broute ebt_snat ebtable_nat ebtable_filter ebtables bridge stp
> llc ipmi_devintf iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi
> scsi_mod ipmi_si ipmi_msghandler aoe(O) rtc_cmos button
> 
> Pid: 1407, comm: netback/9 Tainted: G           O
> 3.4.97-dom0-a801fd8-i386 #68          /0D61XP

And this version number contains a hash doesn't exist in vanilla kernel.
So it suggests the kernel has some custom patches applied.

You can look at the stable patches in 3.10.y tree which I think should
apply to 3.4.y kernel. Of course you might need to adjust them as
needed.

Wei.

> EIP: 0061:[<c124ea69>] EFLAGS: 00010287 CPU: 9
> EIP is at xen_netbk_idx_release+0x109/0x120
> EAX: 0000006f EBX: f5093584 ECX: 00000000 EDX: 0000006f
> ESI: 0000006f EDI: 0000006f EBP: ed473ee8 ESP: ed473ecc
>  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0069
> CR0: 8005003b CR2: 08acb2f8 CR3: 27013000 CR4: 00002660
> DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
> DR6: ffff0ff0 DR7: 00000400
> Process netback/9 (pid: 1407, ti=ed472000 task=eeb0b500 task.ti=ed472000)
> Stack:
>  0000007f ffffffff e70eac00 00000071 f5093584 f5094e00 ffffffff ed473fb8
>  c1250564 f3ec1da0 00000086 00000840 ed473fa8 c1059d04 e8ab4d40 0000007d
>  ed473fa0 ed473f8c ed473f6c f3128300 e8ab4c40 eeb0b500 eeb0b500 f5094de8
> Call Trace:
>  [<c1250564>] xen_netbk_kthread+0x954/0xe20
>  [<c1059d04>] ? idle_balance+0xd4/0x120
>  [<c1049cc0>] ? wake_up_bit+0x30/0x30
>  [<c124fc10>] ? xen_netbk_tx_build_gops+0xe90/0xe90
>  [<c1049964>] kthread+0x74/0x80
>  [<c10498f0>] ? kthread_freezable_should_stop+0x50/0x50
>  [<c13c11f6>] kernel_thread_helper+0x6/0x10
> Code: 8b 44 93 0c c7 40 04 00 00 00 00 8b 44 93 0c e8 de f7 e3 ff 8b
> 7d e4 c7 44 bb 0c 00 00 00 00 83 c4 10 5b 5e 5f c9 c3 0f 0b eb fe <0f>
> 0b eb fe 0f 0b eb fe eb 0d 90 90 90 90 90 90 90 90 90 90 90
> EIP: [<c124ea69>] xen_netbk_idx_release+0x109/0x120 SS:ESP 0069:ed473ecc
> ---[ end trace b51200cc1b02fe07 ]---
> vif vif-50-0: 2 reading script
> p47: port 5(vif50.0) entered disabled state
> 
> 
> -- 
> William
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel

Re: kernel bug in netback.c

[William Dauchy]

On Sun, Jul 20, 2014 at 9:11 PM, Wei Liu <wei.liu2@citrix.com> wrote:
> I gusse you have some other patches applied? There is no BUG_ON at
> vanilla 3.4.97 L1737.

yes indeed; I forgot to double check that.
In my code it does correspond to:
BUG_ON(netbk->pending_ring[pending_index(head)] != pending_idx);
in function xen_netbk_idx_release

> And this version number contains a hash doesn't exist in vanilla kernel.
> So it suggests the kernel has some custom patches applied.

yes this a hash in my own tree.
I'll have a look at 3.10 patches.

-- 
William