Re: [Xen-ia64-devel] Please try PV-on-HVM on IPF - Kasai Takanori - com.xensource.lists.xen-ia64-devel

23 messages in com.xensource.lists.xen-ia64-develRe: [Xen-ia64-devel] Please try PV-on...

From	Sent On	Attachments
DOI Tsunehisa	16 Oct 2006 05:30	.patch, .patch, .patch
Doi....@jp.fujitsu.com	17 Oct 2006 05:29
Doi....@jp.fujitsu.com	17 Oct 2006 05:31
Tristan Gingold	17 Oct 2006 05:45
Doi....@jp.fujitsu.com	17 Oct 2006 06:12
Tristan Gingold	17 Oct 2006 06:34
You, Yongkang	17 Oct 2006 07:27
You, Yongkang	17 Oct 2006 07:36
You, Yongkang	17 Oct 2006 18:59
Alex Williamson	17 Oct 2006 21:16
Doi....@jp.fujitsu.com	18 Oct 2006 00:55
You, Yongkang	18 Oct 2006 01:02
Doi....@jp.fujitsu.com	18 Oct 2006 01:15
You, Yongkang	18 Oct 2006 01:17
Doi....@jp.fujitsu.com	18 Oct 2006 01:37
Kasai Takanori	18 Oct 2006 22:32
You, Yongkang	19 Oct 2006 02:41
Kasai Takanori	20 Oct 2006 02:51
Kasai Takanori	15 Nov 2006 20:09
You, Yongkang	15 Nov 2006 21:08
Kasai Takanori	15 Nov 2006 22:11
You, Yongkang	15 Nov 2006 22:55
Kasai Takanori	16 Nov 2006 20:15

Subject:	Re: [Xen-ia64-devel] Please try PV-on-HVM on IPF
From:	Kasai Takanori (kasa...@jp.fujitsu.com)
Date:	10/20/2006 02:51:03 AM
List:	com.xensource.lists.xen-ia64-devel

Hi Yangkang,

You (yongkang.you) said:

The trace back message will happen when I just inserted vnif driver into VTI, when I used 2 NICs for VTI (1 is IOEMU, the other is VNIF).

I've looked `netif_release_rx_bufs: fix me for copying receiver.' message sometimes at vnif detaching. But I've not met domain-vti crash.

The crashing happened if I also use IOEMU NIC for VTI domain. The VTI vif config should like: vif = [ 'type=ioemu, bridge=xenbr0', '' ]

Thanks. I'll try it.

I was able to confirm this problem occurred in the same environment. (changeset :11810 ) We continue and investigate this problem.

There are two files called by net/core/dev.c. One is a kernel's file, and another is added for VNIF. The added file for VNIF should be called when VNIF is initialized. However, when IOEMU NIC is defined in the config file for VTI domain, kernel's file is called. I have not understood the cause yet.

--- log file --- # kernel: netfront: Initializing virtual ethernet driver. # kernel: vif vif-0: 2 parsing device/vif/0/mac # kernel: netif_release_rx_bufs: fix me for copying receiver. # net.agent[6461]: remove event not handled # net.agent[6463]: remove event not handled # kernel: netfront: device eth2 has copying receive path. # kernel: kernel BUG at net/core/dev.c:3073! <<<<--- This --->>>> # kernel: xenwatch[6329]: bugcheck! 0 [1]

--- linux-2.6-xen-sparse/net/core/dev.c --- # void netdev_run_todo(void) # { 〜 # if (list_empty(&net_todo_list)) <==== 3073 line : disagreement # goto out; 〜

--- linux-2.6.16.29/net/core/dev.c --- # int unregister_netdevice(struct net_device *dev) # { # struct net_device *d, **dp; 〜 # BUG_ON(dev->reg_state != NETREG_REGISTERED); <=== 3073 line : agreement 〜

You (yongkang.you) said: 2. After insert VNIF driver, VTI kernel often calls trace when running applications (kudzu, vi etc.).

This problem doesn't occur as long as we have operated up to now. How many times has this problem occurred in you up to now? Could you teach to me a little more in detail?

You (yongkang.you) said:

Thanks for tracking the issue. I redo the testing, but I also didn't meet the kudzu call trace again. I will track it in future. If I find more problems, I will let you know.

Thank you.

Best Regards, - Takanori Kasai