Re: net-next: Panic in switchdev_port_attr_set_work+0x50/0x60

On Mon, Jun 8, 2015 at 10:04 AM, Brenden Blanco [off-list ref] wrote:

Hi netdev!

I am running a net-next kernel, compiled this morning, git hash of
6508060232ee78200a2258aefeeb148ce9ed1526. The distro is Arch linux latest.
The
issue is reproducible 100% of the time. I create a new Libvirt VM with some
default options, no storage (i.e. 1 core, 1GB ram, PXE boot, 1 vNIC). As
soon as
I launch the VM, I hit this panic.

If, however, I remove the vNIC from the configuration, there is no issue.
The
vNIC is configured to attach to the default libvirt bridge (virbr0).

The below kernel panic is transcribed from a photo, so there is possibility
of
minor transcription errors. Apologies if so, and especially sorry that the
full
panic dump is not captured. I can try to reproduce and capture a more full
trace
if required.

=============== snip ==============
Workqueue: events switchdev_port_attr_set_work
task: ffff8808070b8a30 ti: ffff880807f10000 task.ti: ffff880807f10000
RIP: 0010:[<ffffffff815830b0>] [<ffffffff815830b0>]
switchdev_port_attr_set_work+0x50/0x60
RSP: 0018:ffff880807f13de8  EFLAGS: 00010282
RAX: 00000000ffffffa1 RBX: ffff8808078028a0 RCX: ffff88082ecd5818
RDX: 0000000000000000 RSI: ffff880807f13d78 RDI: ffff8807ff3f40c0
RBP: ffff880807f13df8 R08: ffff88082ecd5800 R09: 0000000000000018
R10: 00000000ffffffff R11: 0000000000021f2a R12: ffff88082ecd5800
R13: ffff88082ecda200 R14: 0000000000000000 R15: ffff8808078028a0
FS:  0000000000000000(0000) GS:ffff88082ecc0000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f7e79fc1000 CR3: 0000000002a0b000 CR4: 00000000001427e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Stack:
 ffff880807626840 ffff880807626840 ffff880807f13e48 ffffffff81091abb
 ffff88082ecd5800 0000000000000000 ffff88082ecd5818 ffff880807626870
 ffff88082ecd5800 0000000000000008 ffff88082ecd5818 ffff880807626840
Call Trace:
 [<ffffffff81091abb>] process_one_work+0x14b/0x470
 [<ffffffff81091e28>] worker_thread+0x48/0x4c0
 [<ffffffff81091de0>] ? process_one_work+0x470/0x470
 [<ffffffff81091de0>] ? process_one_work+0x470/0x470
 [<ffffffff81097978>] kthread+0xd8/0xf0
 [<ffffffff810978a0>] ? kthread_worker_fn+0x170/0x170
 [<ffffffff81594ca2>] ret_from_fork+0x42/0x70
 [<ffffffff810978a0>] ? kthread_worker_fn+0x170/0x170
Code: 75 29 e8 f4 62 f1 ff 48 8b 43 20 48 89 df 48 8b 80 48 04 00 00 65 ff
08 e8 ae 44 c4 ff 48 83 c4 08 5b 5d c3 0f 1f 80 00 00 00 00 <0f> 0b 0f 1f 40
00 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00
RIP  [<ffffffff815830b0>] switchdev_port_attr_set_work+0x50/0x60
 RSP <ffff880807f13de8>
---[ end trace x246aa5718676503 ]---
Kernel panic - not syncing: Fatal exception
Kernel Offset: disabled
=============== end snip ==============

Hi Brenden, I think I see the problem.  Would you try this patch on
net/switchdev/switchdev.c and report back?

@@ -103,7 +105,7 @@ static void switchdev_port_attr_set_work(struct

work_struct *work)

        rtnl_lock();
        err = switchdev_port_attr_set(asw->dev, &asw->attr);
-       BUG_ON(err);
+       BUG_ON(err && err != -EOPNOTSUPP);
        rtnl_unlock();