Hi, I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d (just VT). I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I could see an usb device in domU. But it does not know which device is. Which are the possibilites to do that? I prefer to use hvm machines, but if it is not possible, with PV machines it can work? Thanks in advance -- Sergio Roberto Charpinel Jr. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote:> Hi, > > I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d > (just VT). >USB passthrough doesn''t require VT-d, it''s only required for PCI device passthrough to HVM guests.> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I > could see an usb device in domU. But it does not know which device is. > > Which are the possibilites to do that? I prefer to use hvm machines, > but if it is not possible, with PV machines it can work? >With PV guest you could passthrough the whole USB controller (the PCI device).. -- Pasi _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
He said it was a HVM guest. I don''t think you can passthrough to a hvm domain anything except usb storage with pvusb, correct? On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: >> Hi, >> >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d >> (just VT). >> > > USB passthrough doesn''t require VT-d, it''s only required for PCI device > passthrough to HVM guests. > >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I >> could see an usb device in domU. But it does not know which device is. >> >> Which are the possibilites to do that? I prefer to use hvm machines, >> but if it is not possible, with PV machines it can work? >> > > With PV guest you could passthrough the whole USB controller (the PCI device).. > > -- Pasi > > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pasi, Chris, Thanks for helping. When I use the usb and usbdevice options, I see this in my domU: domU:~# lsusb Bus 001 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub It''s a webcam, so I can''t use it.. it do not create the /dev/video0 since i cant assing the driver to it. Is it possible? Thanks. 2010/2/26 Pasi Kärkkäinen <pasik@iki.fi>:> On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: >> Hi, >> >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d >> (just VT). >> > > USB passthrough doesn''t require VT-d, it''s only required for PCI device > passthrough to HVM guests. > >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I >> could see an usb device in domU. But it does not know which device is. >> >> Which are the possibilites to do that? I prefer to use hvm machines, >> but if it is not possible, with PV machines it can work? >> > > With PV guest you could passthrough the whole USB controller (the PCI device).. > > -- Pasi > >-- Sergio Roberto Charpinel Jr. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote:> He said it was a HVM guest. I don''t think you can passthrough to a hvm > domain anything except usb storage with pvusb, correct? >pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. I''m not totally sure what devices are supported with pvusb and with the normal usb passthrough. -- Pasi> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > > On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: > >> Hi, > >> > >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d > >> (just VT). > >> > > > > USB passthrough doesn''t require VT-d, it''s only required for PCI device > > passthrough to HVM guests. > > > >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I > >> could see an usb device in domU. But it does not know which device is. > >> > >> Which are the possibilites to do that? I prefer to use hvm machines, > >> but if it is not possible, with PV machines it can work? > >> > > > > With PV guest you could passthrough the whole USB controller (the PCI device).. > > > > -- Pasi > > > > > > _______________________________________________ > > Xen-users mailing list > > Xen-users@lists.xensource.com > > http://lists.xensource.com/xen-users > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pasi, I know, my understanding was that pvusb was simply for exposing usb mass storage as a block device to the domain. I''ve never actually used it but I believe you can do this with a hvm domain without anything special. If anyone can actually confirm this it would be good to know :) As far as his original question, he needs vt-d to passthrough any physical device to a hvm domain. - chris On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: >> He said it was a HVM guest. I don''t think you can passthrough to a hvm >> domain anything except usb storage with pvusb, correct? >> > > pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. > > I''m not totally sure what devices are supported with pvusb and with the normal usb passthrough. > > -- Pasi > >> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >> > On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: >> >> Hi, >> >> >> >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d >> >> (just VT). >> >> >> > >> > USB passthrough doesn''t require VT-d, it''s only required for PCI device >> > passthrough to HVM guests. >> > >> >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I >> >> could see an usb device in domU. But it does not know which device is. >> >> >> >> Which are the possibilites to do that? I prefer to use hvm machines, >> >> but if it is not possible, with PV machines it can work? >> >> >> > >> > With PV guest you could passthrough the whole USB controller (the PCI device).. >> > >> > -- Pasi >> > >> > >> > _______________________________________________ >> > Xen-users mailing list >> > Xen-users@lists.xensource.com >> > http://lists.xensource.com/xen-users >> > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Fri, Feb 26, 2010 at 01:22:52PM -0500, chris wrote:> Pasi, > > I know, my understanding was that pvusb was simply for exposing usb > mass storage as a block device to the domain. I''ve never actually used > it but I believe you can do this with a hvm domain without anything > special. If anyone can actually confirm this it would be good to know > :) >There are two separate things: 1. the normal usb passthrough; this works for hvm guests without anything special by specifying usb=1 and usbdevice = [] 2. pvusb, which requires separate usbback driver in dom0, and xen+tools that has pvusb support, and usbfront driver in the guest. -- Pasi> As far as his original question, he needs vt-d to passthrough any > physical device to a hvm domain. > > - chris > > On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > > On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: > >> He said it was a HVM guest. I don''t think you can passthrough to a hvm > >> domain anything except usb storage with pvusb, correct? > >> > > > > pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. > > > > I''m not totally sure what devices are supported with pvusb and with the normal usb passthrough. > > > > -- Pasi > > > >> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > >> > On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: > >> >> Hi, > >> >> > >> >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d > >> >> (just VT). > >> >> > >> > > >> > USB passthrough doesn''t require VT-d, it''s only required for PCI device > >> > passthrough to HVM guests. > >> > > >> >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I > >> >> could see an usb device in domU. But it does not know which device is. > >> >> > >> >> Which are the possibilites to do that? I prefer to use hvm machines, > >> >> but if it is not possible, with PV machines it can work? > >> >> > >> > > >> > With PV guest you could passthrough the whole USB controller (the PCI device).. > >> > > >> > -- Pasi > >> > > >> > > >> > _______________________________________________ > >> > Xen-users mailing list > >> > Xen-users@lists.xensource.com > >> > http://lists.xensource.com/xen-users > >> > > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
How is usb passthrough performance wise with a HVM domain? Is this something thats emulated? I''m curious how it would compare performance wise to passthrough of a usb controller card. - chris On Fri, Feb 26, 2010 at 1:33 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Fri, Feb 26, 2010 at 01:22:52PM -0500, chris wrote: >> Pasi, >> >> I know, my understanding was that pvusb was simply for exposing usb >> mass storage as a block device to the domain. I''ve never actually used >> it but I believe you can do this with a hvm domain without anything >> special. If anyone can actually confirm this it would be good to know >> :) >> > > There are two separate things: > > 1. the normal usb passthrough; this works for hvm guests without anything special by specifying usb=1 and usbdevice = [] > 2. pvusb, which requires separate usbback driver in dom0, and xen+tools that has pvusb support, and usbfront driver in the guest. > > -- Pasi > >> As far as his original question, he needs vt-d to passthrough any >> physical device to a hvm domain. >> >> - chris >> >> On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >> > On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: >> >> He said it was a HVM guest. I don''t think you can passthrough to a hvm >> >> domain anything except usb storage with pvusb, correct? >> >> >> > >> > pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. >> > >> > I''m not totally sure what devices are supported with pvusb and with the normal usb passthrough. >> > >> > -- Pasi >> > >> >> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >> >> > On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: >> >> >> Hi, >> >> >> >> >> >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d >> >> >> (just VT). >> >> >> >> >> > >> >> > USB passthrough doesn''t require VT-d, it''s only required for PCI device >> >> > passthrough to HVM guests. >> >> > >> >> >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I >> >> >> could see an usb device in domU. But it does not know which device is. >> >> >> >> >> >> Which are the possibilites to do that? I prefer to use hvm machines, >> >> >> but if it is not possible, with PV machines it can work? >> >> >> >> >> > >> >> > With PV guest you could passthrough the whole USB controller (the PCI device).. >> >> > >> >> > -- Pasi >> >> > >> >> > >> >> > _______________________________________________ >> >> > Xen-users mailing list >> >> > Xen-users@lists.xensource.com >> >> > http://lists.xensource.com/xen-users >> >> > >> > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
The problem is that When I use the usb and usbdevice options, I see this in my domU: domU:~# lsusb Bus 001 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub It''s a webcam, so I can''t use it.. it do not create the /dev/video0 since i cant assing the driver to it. Do you know what I can do? Thanks 2010/2/26 chris <tknchris@gmail.com>:> How is usb passthrough performance wise with a HVM domain? Is this > something thats emulated? I''m curious how it would compare performance > wise to passthrough of a usb controller card. > > - chris > > On Fri, Feb 26, 2010 at 1:33 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >> On Fri, Feb 26, 2010 at 01:22:52PM -0500, chris wrote: >>> Pasi, >>> >>> I know, my understanding was that pvusb was simply for exposing usb >>> mass storage as a block device to the domain. I''ve never actually used >>> it but I believe you can do this with a hvm domain without anything >>> special. If anyone can actually confirm this it would be good to know >>> :) >>> >> >> There are two separate things: >> >> 1. the normal usb passthrough; this works for hvm guests without anything special by specifying usb=1 and usbdevice = [] >> 2. pvusb, which requires separate usbback driver in dom0, and xen+tools that has pvusb support, and usbfront driver in the guest. >> >> -- Pasi >> >>> As far as his original question, he needs vt-d to passthrough any >>> physical device to a hvm domain. >>> >>> - chris >>> >>> On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >>> > On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: >>> >> He said it was a HVM guest. I don''t think you can passthrough to a hvm >>> >> domain anything except usb storage with pvusb, correct? >>> >> >>> > >>> > pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. >>> > >>> > I''m not totally sure what devices are supported with pvusb and with the normal usb passthrough. >>> > >>> > -- Pasi >>> > >>> >> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >>> >> > On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: >>> >> >> Hi, >>> >> >> >>> >> >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d >>> >> >> (just VT). >>> >> >> >>> >> > >>> >> > USB passthrough doesn''t require VT-d, it''s only required for PCI device >>> >> > passthrough to HVM guests. >>> >> > >>> >> >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I >>> >> >> could see an usb device in domU. But it does not know which device is. >>> >> >> >>> >> >> Which are the possibilites to do that? I prefer to use hvm machines, >>> >> >> but if it is not possible, with PV machines it can work? >>> >> >> >>> >> > >>> >> > With PV guest you could passthrough the whole USB controller (the PCI device).. >>> >> > >>> >> > -- Pasi >>> >> > >>> >> > >>> >> > _______________________________________________ >>> >> > Xen-users mailing list >>> >> > Xen-users@lists.xensource.com >>> >> > http://lists.xensource.com/xen-users >>> >> > >>> > >> >-- Sergio Roberto Charpinel Jr. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Fri, Feb 26, 2010 at 02:00:19PM -0500, chris wrote:> How is usb passthrough performance wise with a HVM domain? Is this > something thats emulated? I''m curious how it would compare performance > wise to passthrough of a usb controller card. >I believe pvusb is much faster than the normal usb passthrough.. 02/2009: http://www.xen.org/files/xensummit_oracle09/PVUSB.pdf 11/2009: http://www.xen.org/files/xensummit_intel09/PVUSBStatusUpdate.pdf Those might have more information about pvusb. -- Pasi> - chris > > On Fri, Feb 26, 2010 at 1:33 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > > On Fri, Feb 26, 2010 at 01:22:52PM -0500, chris wrote: > >> Pasi, > >> > >> I know, my understanding was that pvusb was simply for exposing usb > >> mass storage as a block device to the domain. I''ve never actually used > >> it but I believe you can do this with a hvm domain without anything > >> special. If anyone can actually confirm this it would be good to know > >> :) > >> > > > > There are two separate things: > > > > 1. the normal usb passthrough; this works for hvm guests without anything special by specifying usb=1 and usbdevice = [] > > 2. pvusb, which requires separate usbback driver in dom0, and xen+tools that has pvusb support, and usbfront driver in the guest. > > > > -- Pasi > > > >> As far as his original question, he needs vt-d to passthrough any > >> physical device to a hvm domain. > >> > >> - chris > >> > >> On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > >> > On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: > >> >> He said it was a HVM guest. I don''t think you can passthrough to a hvm > >> >> domain anything except usb storage with pvusb, correct? > >> >> > >> > > >> > pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. > >> > > >> > I''m not totally sure what devices are supported with pvusb and with the normal usb passthrough. > >> > > >> > -- Pasi > >> > > >> >> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > >> >> > On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: > >> >> >> Hi, > >> >> >> > >> >> >> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d > >> >> >> (just VT). > >> >> >> > >> >> > > >> >> > USB passthrough doesn''t require VT-d, it''s only required for PCI device > >> >> > passthrough to HVM guests. > >> >> > > >> >> >> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I > >> >> >> could see an usb device in domU. But it does not know which device is. > >> >> >> > >> >> >> Which are the possibilites to do that? I prefer to use hvm machines, > >> >> >> but if it is not possible, with PV machines it can work? > >> >> >> > >> >> > > >> >> > With PV guest you could passthrough the whole USB controller (the PCI device).. > >> >> > > >> >> > -- Pasi > >> >> > > >> >> > > >> >> > _______________________________________________ > >> >> > Xen-users mailing list > >> >> > Xen-users@lists.xensource.com > >> >> > http://lists.xensource.com/xen-users > >> >> > > >> > > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Pvusb should work for most things. You might be thinking of the USB frontend in gplpv which only works with a very limited number of storage devices (and probably not at all since the last update) James Sent from my iPhone On 27/02/2010, at 5:25, "chris" <tknchris@gmail.com> wrote:> Pasi, > > I know, my understanding was that pvusb was simply for exposing usb > mass storage as a block device to the domain. I've never actually used > it but I believe you can do this with a hvm domain without anything > special. If anyone can actually confirm this it would be good to know > :) > > As far as his original question, he needs vt-d to passthrough any > physical device to a hvm domain. > > - chris > > On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wro > te: >> On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: >>> He said it was a HVM guest. I don't think you can passthrough to a >>> hvm >>> domain anything except usb storage with pvusb, correct? >>> >> >> pvusb is different from the 'normal' usb=1 / usbdevice = [] stuff.. >> >> I'm not totally sure what devices are supported with pvusb and with >> the normal usb passthrough. >> >> -- Pasi >> >>> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> >>> wrote: >>>> On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. >>>> wrote: >>>>> Hi, >>>>> >>>>> I want to use a webcam in a HVM guest, but my dom0 doesnt >>>>> support VT-d >>>>> (just VT). >>>>> >>>> >>>> USB passthrough doesn't require VT-d, it's only required for PCI >>>> device >>>> passthrough to HVM guests. >>>> >>>>> I tried passing the parameters usb=1 and usbdevice = [ '..' ] >>>>> and I >>>>> could see an usb device in domU. But it does not know which >>>>> device is. >>>>> >>>>> Which are the possibilites to do that? I prefer to use hvm >>>>> machines, >>>>> but if it is not possible, with PV machines it can work? >>>>> >>>> >>>> With PV guest you could passthrough the whole USB controller (the >>>> PCI device).. >>>> >>>> -- Pasi >>>> >>>> >>>> _______________________________________________ >>>> Xen-users mailing list >>>> Xen-users@lists.xensource.com >>>> http://lists.xensource.com/xen-users >>>> >> > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I would be curious to see benchmarks comparing usb device passthrough with vt-d passthrough of a usb controller + device. Would be interesting to see what differences if any there are. If I get some time I will try to test this. - chris On Fri, Feb 26, 2010 at 5:00 PM, James Harper <james.harper@bendigoit.com.au> wrote:> Pvusb should work for most things. You might be thinking of the USB frontend > in gplpv which only works with a very limited number of storage devices (and > probably not at all since the last update) > > James > > Sent from my iPhone > > On 27/02/2010, at 5:25, "chris" <tknchris@gmail.com> wrote: > >> Pasi, >> >> I know, my understanding was that pvusb was simply for exposing usb >> mass storage as a block device to the domain. I''ve never actually used >> it but I believe you can do this with a hvm domain without anything >> special. If anyone can actually confirm this it would be good to know >> :) >> >> As far as his original question, he needs vt-d to passthrough any >> physical device to a hvm domain. >> >> - chris >> >> On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >>> >>> On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: >>>> >>>> He said it was a HVM guest. I don''t think you can passthrough to a hvm >>>> domain anything except usb storage with pvusb, correct? >>>> >>> >>> pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. >>> >>> I''m not totally sure what devices are supported with pvusb and with the >>> normal usb passthrough. >>> >>> -- Pasi >>> >>>> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >>>>> >>>>> On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: >>>>>> >>>>>> Hi, >>>>>> >>>>>> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d >>>>>> (just VT). >>>>>> >>>>> >>>>> USB passthrough doesn''t require VT-d, it''s only required for PCI device >>>>> passthrough to HVM guests. >>>>> >>>>>> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I >>>>>> could see an usb device in domU. But it does not know which device is. >>>>>> >>>>>> Which are the possibilites to do that? I prefer to use hvm machines, >>>>>> but if it is not possible, with PV machines it can work? >>>>>> >>>>> >>>>> With PV guest you could passthrough the whole USB controller (the PCI >>>>> device).. >>>>> >>>>> -- Pasi >>>>> >>>>> >>>>> _______________________________________________ >>>>> Xen-users mailing list >>>>> Xen-users@lists.xensource.com >>>>> http://lists.xensource.com/xen-users >>>>> >>> >> >> _______________________________________________ >> Xen-users mailing list >> Xen-users@lists.xensource.com >> http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Fri, Feb 26, 2010 at 05:34:14PM -0500, chris wrote:> I would be curious to see benchmarks comparing usb device passthrough > with vt-d passthrough of a usb controller + device. Would be > interesting to see what differences if any there are. If I get some > time I will try to test this. >See here for benchmark results: http://www.xen.org/files/xensummit_intel09/PVUSBStatusUpdate.pdf pvusb performance seems to be pretty close to dom0 native usb performance. If using the normal qemu-dm usb passthrough it is limited to emulated usb 1.1, so that''ll be slow. -- Pasi> - chris > > On Fri, Feb 26, 2010 at 5:00 PM, James Harper > <james.harper@bendigoit.com.au> wrote: > > Pvusb should work for most things. You might be thinking of the USB frontend > > in gplpv which only works with a very limited number of storage devices (and > > probably not at all since the last update) > > > > James > > > > Sent from my iPhone > > > > On 27/02/2010, at 5:25, "chris" <tknchris@gmail.com> wrote: > > > >> Pasi, > >> > >> I know, my understanding was that pvusb was simply for exposing usb > >> mass storage as a block device to the domain. I''ve never actually used > >> it but I believe you can do this with a hvm domain without anything > >> special. If anyone can actually confirm this it would be good to know > >> :) > >> > >> As far as his original question, he needs vt-d to passthrough any > >> physical device to a hvm domain. > >> > >> - chris > >> > >> On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > >>> > >>> On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: > >>>> > >>>> He said it was a HVM guest. I don''t think you can passthrough to a hvm > >>>> domain anything except usb storage with pvusb, correct? > >>>> > >>> > >>> pvusb is different from the ''normal'' usb=1 / usbdevice = [] stuff.. > >>> > >>> I''m not totally sure what devices are supported with pvusb and with the > >>> normal usb passthrough. > >>> > >>> -- Pasi > >>> > >>>> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > >>>>> > >>>>> On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. wrote: > >>>>>> > >>>>>> Hi, > >>>>>> > >>>>>> I want to use a webcam in a HVM guest, but my dom0 doesnt support VT-d > >>>>>> (just VT). > >>>>>> > >>>>> > >>>>> USB passthrough doesn''t require VT-d, it''s only required for PCI device > >>>>> passthrough to HVM guests. > >>>>> > >>>>>> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] and I > >>>>>> could see an usb device in domU. But it does not know which device is. > >>>>>> > >>>>>> Which are the possibilites to do that? I prefer to use hvm machines, > >>>>>> but if it is not possible, with PV machines it can work? > >>>>>> > >>>>> > >>>>> With PV guest you could passthrough the whole USB controller (the PCI > >>>>> device).. > >>>>> > >>>>> -- Pasi > >>>>> > >>>>> > >>>>> _______________________________________________ > >>>>> Xen-users mailing list > >>>>> Xen-users@lists.xensource.com > >>>>> http://lists.xensource.com/xen-users > >>>>> > >>> > >> > >> _______________________________________________ > >> Xen-users mailing list > >> Xen-users@lists.xensource.com > >> http://lists.xensource.com/xen-users > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I wonder why it''s limited to 1.1 i would imagine that is a hard coded limit but given a faster system shouldn''t the emulation be faster? Who could shed more light on why? Sent from my iPhone On Feb 26, 2010, at 5:44 PM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Fri, Feb 26, 2010 at 05:34:14PM -0500, chris wrote: >> I would be curious to see benchmarks comparing usb device passthrough >> with vt-d passthrough of a usb controller + device. Would be >> interesting to see what differences if any there are. If I get some >> time I will try to test this. >> > > See here for benchmark results: > http://www.xen.org/files/xensummit_intel09/PVUSBStatusUpdate.pdf > > pvusb performance seems to be pretty close to dom0 native usb > performance. > > If using the normal qemu-dm usb passthrough it is limited to > emulated usb 1.1, > so that''ll be slow. > > -- Pasi > >> - chris >> >> On Fri, Feb 26, 2010 at 5:00 PM, James Harper >> <james.harper@bendigoit.com.au> wrote: >>> Pvusb should work for most things. You might be thinking of the >>> USB frontend >>> in gplpv which only works with a very limited number of storage >>> devices (and >>> probably not at all since the last update) >>> >>> James >>> >>> Sent from my iPhone >>> >>> On 27/02/2010, at 5:25, "chris" <tknchris@gmail.com> wrote: >>> >>>> Pasi, >>>> >>>> I know, my understanding was that pvusb was simply for exposing usb >>>> mass storage as a block device to the domain. I''ve never actually >>>> used >>>> it but I believe you can do this with a hvm domain without anything >>>> special. If anyone can actually confirm this it would be good to >>>> know >>>> :) >>>> >>>> As far as his original question, he needs vt-d to passthrough any >>>> physical device to a hvm domain. >>>> >>>> - chris >>>> >>>> On Fri, Feb 26, 2010 at 10:03 AM, Pasi Kärkkäinen <pasik@iki.f >>>> i> wrote: >>>>> >>>>> On Fri, Feb 26, 2010 at 10:00:34AM -0500, chris wrote: >>>>>> >>>>>> He said it was a HVM guest. I don''t think you can passthrough >>>>>> to a hvm >>>>>> domain anything except usb storage with pvusb, correct? >>>>>> >>>>> >>>>> pvusb is different from the ''normal'' usb=1 / usbdevice = [] >>>>> stuff.. >>>>> >>>>> I''m not totally sure what devices are supported with pvusb and >>>>> with the >>>>> normal usb passthrough. >>>>> >>>>> -- Pasi >>>>> >>>>>> On Fri, Feb 26, 2010 at 9:49 AM, Pasi Kärkkäinen <pasik@ik >>>>>> i.fi> wrote: >>>>>>> >>>>>>> On Fri, Feb 26, 2010 at 09:49:38AM -0300, Sergio Charpinel Jr. >>>>>>> wrote: >>>>>>>> >>>>>>>> Hi, >>>>>>>> >>>>>>>> I want to use a webcam in a HVM guest, but my dom0 doesnt >>>>>>>> support VT-d >>>>>>>> (just VT). >>>>>>>> >>>>>>> >>>>>>> USB passthrough doesn''t require VT-d, it''s only required for >>>>>>> PCI device >>>>>>> passthrough to HVM guests. >>>>>>> >>>>>>>> I tried passing the parameters usb=1 and usbdevice = [ ''..'' ] >>>>>>>> and I >>>>>>>> could see an usb device in domU. But it does not know which >>>>>>>> device is. >>>>>>>> >>>>>>>> Which are the possibilites to do that? I prefer to use hvm >>>>>>>> machines, >>>>>>>> but if it is not possible, with PV machines it can work? >>>>>>>> >>>>>>> >>>>>>> With PV guest you could passthrough the whole USB controller >>>>>>> (the PCI >>>>>>> device).. >>>>>>> >>>>>>> -- Pasi >>>>>>> >>>>>>> >>>>>>> _______________________________________________ >>>>>>> Xen-users mailing list >>>>>>> Xen-users@lists.xensource.com >>>>>>> http://lists.xensource.com/xen-users >>>>>>> >>>>> >>>> >>>> _______________________________________________ >>>> Xen-users mailing list >>>> Xen-users@lists.xensource.com >>>> http://lists.xensource.com/xen-users >>>_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> > I wonder why it's limited to 1.1 i would imagine that is a hard coded > limit but given a faster system shouldn't the emulation be faster? Who > could shed more light on why? >Well if it talks to the device in '1.1 mode' then the limit will still apply. At a guess, maybe the 1.1 hardware interface is a bit simpler? I suspect that they already implemented USB 1.1 to support the mouse and tablet interface, and passthrough was not much harder, but 2.0/3.0 implementation would have been a lot more work. James _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
That makes sense however one would think USB 2.0 support would be a good feature to have. There''s quite a bit of people looking to utilize pci passthrough and I''d imagine USB passthrough would be similarly useful Sent from my iPhone On Feb 26, 2010, at 7:36 PM, "James Harper" <james.harper@bendigoit.com.au > wrote:>> >> I wonder why it''s limited to 1.1 i would imagine that is a hard coded >> limit but given a faster system shouldn''t the emulation be faster? >> Who >> could shed more light on why? >> > > Well if it talks to the device in ''1.1 mode'' then the limit will > still apply. > > At a guess, maybe the 1.1 hardware interface is a bit simpler? I > suspect that they already implemented USB 1.1 to support the mouse > and tablet interface, and passthrough was not much harder, but > 2.0/3.0 implementation would have been a lot more work. > > James_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Fri, Feb 26, 2010 at 10:26:29PM -0500, Chris wrote:> That makes sense however one would think USB 2.0 support would be a good > feature to have. There''s quite a bit of people looking to utilize pci > passthrough and I''d imagine USB passthrough would be similarly useful >That''s why pvusb was developed.. it allows usb 2.0 speeds/functionality :) -- Pasi> Sent from my iPhone > > On Feb 26, 2010, at 7:36 PM, "James Harper" > <james.harper@bendigoit.com.au> wrote: > >>> >>> I wonder why it''s limited to 1.1 i would imagine that is a hard coded >>> limit but given a faster system shouldn''t the emulation be faster? >>> Who >>> could shed more light on why? >>> >> >> Well if it talks to the device in ''1.1 mode'' then the limit will still >> apply. >> >> At a guess, maybe the 1.1 hardware interface is a bit simpler? I >> suspect that they already implemented USB 1.1 to support the mouse and >> tablet interface, and passthrough was not much harder, but 2.0/3.0 >> implementation would have been a lot more work. >> >> James_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
My understanding of PVUSB is that it was only for PV domains though correct? Looked through the links and can''t find anything confirming or denying that. - chris On Sat, Feb 27, 2010 at 7:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Fri, Feb 26, 2010 at 10:26:29PM -0500, Chris wrote: >> That makes sense however one would think USB 2.0 support would be a good >> feature to have. There''s quite a bit of people looking to utilize pci >> passthrough and I''d imagine USB passthrough would be similarly useful >> > > That''s why pvusb was developed.. it allows usb 2.0 speeds/functionality :) > > -- Pasi > >> Sent from my iPhone >> >> On Feb 26, 2010, at 7:36 PM, "James Harper" >> <james.harper@bendigoit.com.au> wrote: >> >>>> >>>> I wonder why it''s limited to 1.1 i would imagine that is a hard coded >>>> limit but given a faster system shouldn''t the emulation be faster? >>>> Who >>>> could shed more light on why? >>>> >>> >>> Well if it talks to the device in ''1.1 mode'' then the limit will still >>> apply. >>> >>> At a guess, maybe the 1.1 hardware interface is a bit simpler? I >>> suspect that they already implemented USB 1.1 to support the mouse and >>> tablet interface, and passthrough was not much harder, but 2.0/3.0 >>> implementation would have been a lot more work. >>> >>> James >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I''ve successfully passed through a wireless USB stick to a windows VM via the "qemu" usb passthrough. At USB 1.1 it wasn''t really that usable however. On 27 February 2010 15:15, chris <tknchris@gmail.com> wrote:> My understanding of PVUSB is that it was only for PV domains though > correct? Looked through the links and can''t find anything confirming > or denying that. > > - chris > > On Sat, Feb 27, 2010 at 7:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > > On Fri, Feb 26, 2010 at 10:26:29PM -0500, Chris wrote: > >> That makes sense however one would think USB 2.0 support would be a good > >> feature to have. There''s quite a bit of people looking to utilize pci > >> passthrough and I''d imagine USB passthrough would be similarly useful > >> > > > > That''s why pvusb was developed.. it allows usb 2.0 speeds/functionality > :) > > > > -- Pasi > > > >> Sent from my iPhone > >> > >> On Feb 26, 2010, at 7:36 PM, "James Harper" > >> <james.harper@bendigoit.com.au> wrote: > >> > >>>> > >>>> I wonder why it''s limited to 1.1 i would imagine that is a hard coded > >>>> limit but given a faster system shouldn''t the emulation be faster? > >>>> Who > >>>> could shed more light on why? > >>>> > >>> > >>> Well if it talks to the device in ''1.1 mode'' then the limit will still > >>> apply. > >>> > >>> At a guess, maybe the 1.1 hardware interface is a bit simpler? I > >>> suspect that they already implemented USB 1.1 to support the mouse and > >>> tablet interface, and passthrough was not much harder, but 2.0/3.0 > >>> implementation would have been a lot more work. > >>> > >>> James > > > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Sat, Feb 27, 2010 at 10:15:12AM -0500, chris wrote:> My understanding of PVUSB is that it was only for PV domains though > correct? Looked through the links and can''t find anything confirming > or denying that. >pvusb is not only for PV guests. It also works with HVM guests. For both PV and HVM guests it requires the pvusb frontend driver in the guest kernel. For Windows the GPLPV drivers have pvusb frontend, but it''s not in a working state atm, afaik. -- Pasi> - chris > > On Sat, Feb 27, 2010 at 7:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > > On Fri, Feb 26, 2010 at 10:26:29PM -0500, Chris wrote: > >> That makes sense however one would think USB 2.0 support would be a good > >> feature to have. There''s quite a bit of people looking to utilize pci > >> passthrough and I''d imagine USB passthrough would be similarly useful > >> > > > > That''s why pvusb was developed.. it allows usb 2.0 speeds/functionality :) > > > > -- Pasi > > > >> Sent from my iPhone > >> > >> On Feb 26, 2010, at 7:36 PM, "James Harper" > >> <james.harper@bendigoit.com.au> wrote: > >> > >>>> > >>>> I wonder why it''s limited to 1.1 i would imagine that is a hard coded > >>>> limit but given a faster system shouldn''t the emulation be faster? > >>>> Who > >>>> could shed more light on why? > >>>> > >>> > >>> Well if it talks to the device in ''1.1 mode'' then the limit will still > >>> apply. > >>> > >>> At a guess, maybe the 1.1 hardware interface is a bit simpler? I > >>> suspect that they already implemented USB 1.1 to support the mouse and > >>> tablet interface, and passthrough was not much harder, but 2.0/3.0 > >>> implementation would have been a lot more work. > >>> > >>> James > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
James, Any idea what version it broke at? Do you need help with testing? I''d be willing to help you if I can. - chris On Sat, Feb 27, 2010 at 10:31 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote:> On Sat, Feb 27, 2010 at 10:15:12AM -0500, chris wrote: >> My understanding of PVUSB is that it was only for PV domains though >> correct? Looked through the links and can''t find anything confirming >> or denying that. >> > > pvusb is not only for PV guests. It also works with HVM guests. > > For both PV and HVM guests it requires the pvusb frontend driver in the guest kernel. > For Windows the GPLPV drivers have pvusb frontend, but it''s not in a working state atm, afaik. > > -- Pasi > >> - chris >> >> On Sat, Feb 27, 2010 at 7:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: >> > On Fri, Feb 26, 2010 at 10:26:29PM -0500, Chris wrote: >> >> That makes sense however one would think USB 2.0 support would be a good >> >> feature to have. There''s quite a bit of people looking to utilize pci >> >> passthrough and I''d imagine USB passthrough would be similarly useful >> >> >> > >> > That''s why pvusb was developed.. it allows usb 2.0 speeds/functionality :) >> > >> > -- Pasi >> > >> >> Sent from my iPhone >> >> >> >> On Feb 26, 2010, at 7:36 PM, "James Harper" >> >> <james.harper@bendigoit.com.au> wrote: >> >> >> >>>> >> >>>> I wonder why it''s limited to 1.1 i would imagine that is a hard coded >> >>>> limit but given a faster system shouldn''t the emulation be faster? >> >>>> Who >> >>>> could shed more light on why? >> >>>> >> >>> >> >>> Well if it talks to the device in ''1.1 mode'' then the limit will still >> >>> apply. >> >>> >> >>> At a guess, maybe the 1.1 hardware interface is a bit simpler? I >> >>> suspect that they already implemented USB 1.1 to support the mouse and >> >>> tablet interface, and passthrough was not much harder, but 2.0/3.0 >> >>> implementation would have been a lot more work. >> >>> >> >>> James >> > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Sat, Feb 27, 2010 at 10:36:46AM -0500, chris wrote:> James, > > Any idea what version it broke at? Do you need help with testing? I''d > be willing to help you if I can. >Also remember you need to have the usbback driver in dom0.. pvops dom0 doesn''t have usbback yet. -- Pasi> - chris > > On Sat, Feb 27, 2010 at 10:31 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > > On Sat, Feb 27, 2010 at 10:15:12AM -0500, chris wrote: > >> My understanding of PVUSB is that it was only for PV domains though > >> correct? Looked through the links and can''t find anything confirming > >> or denying that. > >> > > > > pvusb is not only for PV guests. It also works with HVM guests. > > > > For both PV and HVM guests it requires the pvusb frontend driver in the guest kernel. > > For Windows the GPLPV drivers have pvusb frontend, but it''s not in a working state atm, afaik. > > > > -- Pasi > > > >> - chris > >> > >> On Sat, Feb 27, 2010 at 7:49 AM, Pasi Kärkkäinen <pasik@iki.fi> wrote: > >> > On Fri, Feb 26, 2010 at 10:26:29PM -0500, Chris wrote: > >> >> That makes sense however one would think USB 2.0 support would be a good > >> >> feature to have. There''s quite a bit of people looking to utilize pci > >> >> passthrough and I''d imagine USB passthrough would be similarly useful > >> >> > >> > > >> > That''s why pvusb was developed.. it allows usb 2.0 speeds/functionality :) > >> > > >> > -- Pasi > >> > > >> >> Sent from my iPhone > >> >> > >> >> On Feb 26, 2010, at 7:36 PM, "James Harper" > >> >> <james.harper@bendigoit.com.au> wrote: > >> >> > >> >>>> > >> >>>> I wonder why it''s limited to 1.1 i would imagine that is a hard coded > >> >>>> limit but given a faster system shouldn''t the emulation be faster? > >> >>>> Who > >> >>>> could shed more light on why? > >> >>>> > >> >>> > >> >>> Well if it talks to the device in ''1.1 mode'' then the limit will still > >> >>> apply. > >> >>> > >> >>> At a guess, maybe the 1.1 hardware interface is a bit simpler? I > >> >>> suspect that they already implemented USB 1.1 to support the mouse and > >> >>> tablet interface, and passthrough was not much harder, but 2.0/3.0 > >> >>> implementation would have been a lot more work. > >> >>> > >> >>> James > >> > > >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> > James, > > Any idea what version it broke at? Do you need help with testing? I''d > be willing to help you if I can. >What sort of device are you trying to get working? The broken-ness should be pretty easy to fix, but there is a fair bit of missing functionality. James _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> > On Sat, Feb 27, 2010 at 10:36:46AM -0500, chris wrote: > > James, > > > > Any idea what version it broke at? Do you need help with testing? I''d > > be willing to help you if I can. > > > > Also remember you need to have the usbback driver in dom0.. pvops dom0 > doesn''t have usbback yet. >It''s not that hard to port. There was some problem with freeing some data structure with interrupts disabled which causes a problem but otherwise it''s pretty straightforward. James _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I have no need at the moment to support a specific device. I recently I had built a machine for a client with virtual machines accessing physical scanners, which I ended up using vt-d / usb controller. In hindsight it looks like pvusb would be more ideal and would probably perform good enough to support this type of a setup. I''d be glad to help test some devices if it would help get this working. - chris On Sat, Feb 27, 2010 at 5:40 PM, James Harper <james.harper@bendigoit.com.au> wrote:>> >> On Sat, Feb 27, 2010 at 10:36:46AM -0500, chris wrote: >> > James, >> > >> > Any idea what version it broke at? Do you need help with testing? I''d >> > be willing to help you if I can. >> > >> >> Also remember you need to have the usbback driver in dom0.. pvops dom0 >> doesn''t have usbback yet. >> > > It''s not that hard to port. There was some problem with freeing some data structure with interrupts disabled which causes a problem but otherwise it''s pretty straightforward. > > James >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Where I can find the modules usbback and usbfront? Thanks 2010/2/27 chris <tknchris@gmail.com>:> I have no need at the moment to support a specific device. I recently > I had built a machine for a client with virtual machines accessing > physical scanners, which I ended up using vt-d / usb controller. In > hindsight it looks like pvusb would be more ideal and would probably > perform good enough to support this type of a setup. I''d be glad to > help test some devices if it would help get this working. > > - chris > > On Sat, Feb 27, 2010 at 5:40 PM, James Harper > <james.harper@bendigoit.com.au> wrote: >>> >>> On Sat, Feb 27, 2010 at 10:36:46AM -0500, chris wrote: >>> > James, >>> > >>> > Any idea what version it broke at? Do you need help with testing? I''d >>> > be willing to help you if I can. >>> > >>> >>> Also remember you need to have the usbback driver in dom0.. pvops dom0 >>> doesn''t have usbback yet. >>> >> >> It''s not that hard to port. There was some problem with freeing some data structure with interrupts disabled which causes a problem but otherwise it''s pretty straightforward. >> >> James >> >-- Sergio Roberto Charpinel Jr. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Mon, Mar 01, 2010 at 08:22:45AM -0300, Sergio Charpinel Jr. wrote:> Where I can find the modules usbback and usbfront? >I think they''re at least in http://xenbits.xen.org/linux-2.6.18-xen.hg If you have some extra time you could forward-port them to pv_ops dom0 kernel! -- Pasi> Thanks > > 2010/2/27 chris <tknchris@gmail.com>: > > I have no need at the moment to support a specific device. I recently > > I had built a machine for a client with virtual machines accessing > > physical scanners, which I ended up using vt-d / usb controller. In > > hindsight it looks like pvusb would be more ideal and would probably > > perform good enough to support this type of a setup. I''d be glad to > > help test some devices if it would help get this working. > > > > - chris > > > > On Sat, Feb 27, 2010 at 5:40 PM, James Harper > > <james.harper@bendigoit.com.au> wrote: > >>> > >>> On Sat, Feb 27, 2010 at 10:36:46AM -0500, chris wrote: > >>> > James, > >>> > > >>> > Any idea what version it broke at? Do you need help with testing? I''d > >>> > be willing to help you if I can. > >>> > > >>> > >>> Also remember you need to have the usbback driver in dom0.. pvops dom0 > >>> doesn''t have usbback yet. > >>> > >> > >> It''s not that hard to port. There was some problem with freeing some data structure with interrupts disabled which causes a problem but otherwise it''s pretty straightforward. > >> > >> James > >> > > > > > > -- > Sergio Roberto Charpinel Jr._______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users