[Xen-users] Xen PCI Passthrough error

[Fabien Gaud <fabien.gaud@xxxxxxxx>]

Hello,

I'm trying to get PCI passthrough working on my Dell R905 machine.
This machine is equipped with four e1000e cards. I'm running Xen 4.1
on Debian Squeeze with a 2.6.32 kernel.

When I start a DomU with a directly assigned NIC, I got "0000:00:00.0:
eth0: Detected Tx Unit Hang". And after many of these messages, I got
:

[   14.816116] WARNING: at
/build/buildd-linux-2.6_2.6.32-31-amd64-vrfdM4/linux-2.6-2.6.32/debian/build/source_amd64_xen/net/sched/sch_generic.c:261
dev_watchdog+0xe2/0x194()
[   14.816151] NETDEV WATCHDOG: eth0 (e1000e): transmit queue 0 timed out
[   14.816167] Modules linked in: evdev snd_pcm snd_timer snd
soundcore snd_page_alloc pcspkr ext3 jbd mbcache e1000e xen_blkfront
[   14.816247] Pid: 0, comm: swapper Not tainted 2.6.32-5-xen-amd64 #1
[   14.816264] Call Trace:
[   14.816275]  <IRQ>  [<ffffffff81272bee>] ? dev_watchdog+0xe2/0x194
[   14.816310]  [<ffffffff81272bee>] ? dev_watchdog+0xe2/0x194
[   14.816332]  [<ffffffff8104ef10>] ? warn_slowpath_common+0x77/0xa3
[   14.816354]  [<ffffffff81066fd5>] ? run_posix_cpu_timers+0x25/0x6ea
[   14.816376]  [<ffffffff81272b0c>] ? dev_watchdog+0x0/0x194
[   14.816395]  [<ffffffff8104ef98>] ? warn_slowpath_fmt+0x51/0x59
[   14.816416]  [<ffffffff8102ddcc>] ? pvclock_clocksource_read+0x3a/0x8b
[   14.816439]  [<ffffffff8100ec1a>] ? xen_vcpuop_set_next_event+0x0/0x60
[   14.816460]  [<ffffffff8100ec66>] ? xen_vcpuop_set_next_event+0x4c/0x60
[   14.816484]  [<ffffffff8107112f>] ? tick_dev_program_event+0x2d/0x95
[   14.816505]  [<ffffffff81272ae0>] ? netif_tx_lock+0x3d/0x69
[   14.816526]  [<ffffffff8125d584>] ? netdev_drivername+0x3b/0x40
[   14.816547]  [<ffffffff81272bee>] ? dev_watchdog+0xe2/0x194
[   14.816569]  [<ffffffff8100eb00>] ? xen_timer_interrupt+0x17d/0x18d
[   14.816592]  [<ffffffff8105b5ff>] ? run_timer_softirq+0x1c9/0x268
[   14.816615]  [<ffffffff81054cab>] ? __do_softirq+0xdd/0x1a6
[   14.816636]  [<ffffffff81012cac>] ? call_softirq+0x1c/0x30
[   14.816656]  [<ffffffff8101422b>] ? do_softirq+0x3f/0x7c
[   14.816675]  [<ffffffff81054b1b>] ? irq_exit+0x36/0x76
[   14.816696]  [<ffffffff811f2269>] ? xen_evtchn_do_upcall+0x33/0x42
[   14.816716]  [<ffffffff81012cfe>] ? xen_do_hypervisor_callback+0x1e/0x30
[   14.816733]  <EOI>  [<ffffffff810093aa>] ? hypercall_page+0x3aa/0x1001
[   14.816763]  [<ffffffff810093aa>] ? hypercall_page+0x3aa/0x1001
[   14.816783]  [<ffffffff8100ec1a>] ? xen_vcpuop_set_next_event+0x0/0x60
[   14.816804]  [<ffffffff8100e6bb>] ? xen_safe_halt+0xc/0x15
[   14.816825]  [<ffffffff8100bfc7>] ? xen_idle+0x37/0x40
[   14.816843]  [<ffffffff81010eb1>] ? cpu_idle+0xa2/0xda
[   14.816864]  [<ffffffff81509cdd>] ? start_kernel+0x3dc/0x3e8
[   14.816884]  [<ffffffff8150bc93>] ? xen_start_kernel+0x586/0x58a
[   14.816902] ---[ end trace 199b78cce11f6233 ]---



Note that I've got the same issue with Xen 4.0. I have also tried to
upgrade the e1000e driver to the latest version available (1.3.10a)
without any success.

Does anybody have an idea ?

My problem is closely related to this one :
http://lists.xensource.com/archives/html/xen-devel/2010-05/msg01066.html.
A solution (a patch) was proposed but, unfortunately, is not publicly
available.

Thanks in advance,
Fabien Gaud.

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users