Re: [Qemu-devel] Advice on some configuration parameters

Re: [Qemu-devel] Advice on some configuration parameters

[Fabio Fantoni]

[-- Attachment #1: Type: text/plain, Size: 2436 bytes --]

Il 12/02/2013 14:05, Fabio Fantoni ha scritto:
> I'm making patches to enable some qemu upstream features in xen that 
> are missing in libxl.
> I'm trying to do it just by giving arguments to qemu, and I want them 
> to be dynamic (e.g. without physical addresses if possible) and concise.
>
> I'm confused about usb configuration.
>
> Looking at what virt-manager does, it seems that you set the 
> usbredirection in channels like this:
> -chardev spicevmc,id=charredir0,name=usbredir -device 
> usb-redir,chardev=charredir0,id=redir0
>
> At the moment I'm defining the physical usb devices like this:
> -readconfig /etc/qemu/ich9-ehci-uhci.cfg
>
> If I try to set up virt-manager so that it doesn't point to a file, it 
> uses this configuration:
>
> -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
> -device 
> ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,addr=0x5
> -device ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
> -device ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2
>
> This seems to differ from the configuration file, e.g. you have 
> multifunzion=on only on the first uhci, besides I can't understand how 
> many physical ports are defined. I tried setting 9 usbredirection 
> channels in virt-manager but that configuration doesn't change and no 
> errors are reported, but I doubt that physical ports are being define 
> dynamically.
>
> Can you omit physical addresses?
> I searched the web but I can't find detailed informations about it. 
> Can someone explain in detail how you are supposed to configure usb 
> physical ports? I want to enable both usb redirection and usb 
> passthrough in a dynamic and concise manner.
>
> About spice vdagent at the moment I'm using this configuration that 
> seems to work:
> -device virtio-serial -chardev spicevmc,id=vdagent,name=vdagent 
> -device virtserialport,chardev=vdagent,name=com.redhat.spice.0
>
> But I noticed that virt-manager use this slightly different one:
> -chardev pty,id=charserial0 -device 
> isa-serial,chardev=charserial0,id=serial0 -chardev 
> spicevmc,id=charchannel0,name=vdagent -device 
> virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.spice.0
>
> Are there any problem with the more concise one I'm using?
>
> Thanks for any reply.
>
>
Are there someone can help me please?


[-- Attachment #2: Firma crittografica S/MIME --]
[-- Type: application/pkcs7-signature, Size: 4510 bytes --]

Re: [Qemu-devel] [Spice-devel] Advice on some configuration parameters

[Hans de Goede]

Hi,

On 03/01/2013 11:02 AM, Fabio Fantoni wrote:
> Il 12/02/2013 14:05, Fabio Fantoni ha scritto:
>> I'm making patches to enable some qemu upstream features in xen that are missing in libxl.
>> I'm trying to do it just by giving arguments to qemu, and I want them to be dynamic (e.g. without physical addresses if possible) and concise.
>>
>> I'm confused about usb configuration.
>>
>> Looking at what virt-manager does, it seems that you set the usbredirection in channels like this:
>> -chardev spicevmc,id=charredir0,name=usbredir -device usb-redir,chardev=charredir0,id=redir0
>>
>> At the moment I'm defining the physical usb devices like this:
>> -readconfig /etc/qemu/ich9-ehci-uhci.cfg
>>
>> If I try to set up virt-manager so that it doesn't point to a file, it uses this configuration:
>>
>> -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
>> -device ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,addr=0x5
>> -device ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
>> -device ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2
>>
>> This seems to differ from the configuration file, e.g. you have multifunzion=on only on the first uhci

That does not matter the multifunction bit needs to set only on 0x5.0x0.

> besides I can't understand how many physical ports are defined.

6, the same as with the "-readconfig /etc/qemu/ich9-ehci-uhci.cf" directive.

> I tried setting 9 usbredirection channels in virt-manager but that configuration doesn't change and no errors are reported, but I doubt that physical ports are being define dynamically.

Hmm, that would be a libvirt bug you should be able to add only 6
usb-redir channels, unless you add a second usb controller ...

Possibly libvirt is also adding a hub? That would work sort-of, it
would make channels 6-9 be usb-1 only, since we don't emulate
usb-2 hubs.

>>
>> Can you omit physical addresses?

In most cases yes, then qemu will pick one itself.

>> I searched the web but I can't find detailed informations about it. Can someone explain in detail how you are supposed to configure usb physical ports? I want to enable both usb redirection and usb passthrough in a dynamic and concise manner.

The easiest way to do is probably to create the setup you want using
virt-manager and then run the vm once, after that the xml file
should contain fixed addresses for all devices.

>>
>> About spice vdagent at the moment I'm using this configuration that seems to work:
>> -device virtio-serial -chardev spicevmc,id=vdagent,name=vdagent -device virtserialport,chardev=vdagent,name=com.redhat.spice.0
>>
>> But I noticed that virt-manager use this slightly different one:
>> -chardev pty,id=charserial0 -device isa-serial,chardev=charserial0,id=serial0 -chardev spicevmc,id=charchannel0,name=vdagent -device virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.spice.0
>>
>> Are there any problem with the more concise one I'm using?

No.

Regards,

Hans

Re: [Qemu-devel] [Spice-devel] Advice on some configuration parameters

[Fabio Fantoni]

[-- Attachment #1: Type: text/plain, Size: 10737 bytes --]

Il 05/03/2013 11:19, Hans de Goede ha scritto:
> Hi,
>
> On 03/01/2013 11:02 AM, Fabio Fantoni wrote:
>> Il 12/02/2013 14:05, Fabio Fantoni ha scritto:
>>> I'm making patches to enable some qemu upstream features in xen that 
>>> are missing in libxl.
>>> I'm trying to do it just by giving arguments to qemu, and I want 
>>> them to be dynamic (e.g. without physical addresses if possible) and 
>>> concise.
>>>
>>> I'm confused about usb configuration.
>>>
>>> Looking at what virt-manager does, it seems that you set the 
>>> usbredirection in channels like this:
>>> -chardev spicevmc,id=charredir0,name=usbredir -device 
>>> usb-redir,chardev=charredir0,id=redir0
>>>
>>> At the moment I'm defining the physical usb devices like this:
>>> -readconfig /etc/qemu/ich9-ehci-uhci.cfg
>>>
>>> If I try to set up virt-manager so that it doesn't point to a file, 
>>> it uses this configuration:
>>>
>>> -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
>>> -device 
>>> ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,addr=0x5
>>> -device 
>>> ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
>>> -device 
>>> ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2
>>>
>>> This seems to differ from the configuration file, e.g. you have 
>>> multifunzion=on only on the first uhci
>
> That does not matter the multifunction bit needs to set only on 0x5.0x0.
>
>> besides I can't understand how many physical ports are defined.
>
> 6, the same as with the "-readconfig /etc/qemu/ich9-ehci-uhci.cf" 
> directive.
>
>> I tried setting 9 usbredirection channels in virt-manager but that 
>> configuration doesn't change and no errors are reported, but I doubt 
>> that physical ports are being define dynamically.
>
> Hmm, that would be a libvirt bug you should be able to add only 6
> usb-redir channels, unless you add a second usb controller ...
>
> Possibly libvirt is also adding a hub? That would work sort-of, it
> would make channels 6-9 be usb-1 only, since we don't emulate
> usb-2 hubs.
>
>>>
>>> Can you omit physical addresses?
>
> In most cases yes, then qemu will pick one itself.
>
>>> I searched the web but I can't find detailed informations about it. 
>>> Can someone explain in detail how you are supposed to configure usb 
>>> physical ports? I want to enable both usb redirection and usb 
>>> passthrough in a dynamic and concise manner.
>
> The easiest way to do is probably to create the setup you want using
> virt-manager and then run the vm once, after that the xml file
> should contain fixed addresses for all devices.
>
>>>
>>> About spice vdagent at the moment I'm using this configuration that 
>>> seems to work:
>>> -device virtio-serial -chardev spicevmc,id=vdagent,name=vdagent 
>>> -device virtserialport,chardev=vdagent,name=com.redhat.spice.0
>>>
>>> But I noticed that virt-manager use this slightly different one:
>>> -chardev pty,id=charserial0 -device 
>>> isa-serial,chardev=charserial0,id=serial0 -chardev 
>>> spicevmc,id=charchannel0,name=vdagent -device 
>>> virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.spice.0
>>>
>>> Are there any problem with the more concise one I'm using?
>
> No.
>
> Regards,
>
> Hans
>

Thanks for reply, my goal is doing complete and good full feature 
implementation of spice in xen (now exactly on libxl).
The basic implementation was already present and working but without 
vdagent, qxl and usb redirection.

I did patch for qxl support, for now is working only as standard vga, 
with qxl driver not working in both windows and linux vm.
I already did many posts on xen-devel,qemu-devel and spice-devel but 
without solution by now.
I tried to found a solution myself for about 1 year without success. It 
seems (almost certain) there are changes/fixes to do on hvmloader (xen 
side) and it should be also necessary changes on qemu and/or spice side.
Unfortunately I don't have much knowledge on this parts and I not found 
nobody that can help me about for now.

I'm trying to have a complete feature list of qemu options (about spice) 
by doing a try and learn tests between kvm and xen.
I have some question:
Is -nodefaults necessary/useful for spice? xen do not use/want that for now
Is -no-shutdown necessary/useful for spice?
What is exactly seamless migration on spice? is necessary to be on?
Is -device usb-tablet necessary/useful for spice?


I tried to do usb redirection without cfg files but is not working.
These are all qemu option of W7 domU of my test:

libxl: debug: libxl_device.c:920:device_hotplug: calling hotplug script: 
/etc/xen/scripts/block add
libxl: debug: libxl_dm.c:1156:libxl__spawn_local_dm: Spawning 
device-model /usr/lib/xen/bin/qemu-system-i386 with arguments:
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
/usr/lib/xen/bin/qemu-system-i386
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -xen-domid
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   9
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
socket,id=libxl-cmd,path=/var/run/xen/qmp-libxl-9,server,nowait
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -mon
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
chardev=libxl-cmd,mode=control
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -name
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   W7
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -global
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: isa-fdc.driveA=
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -spice
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
port=6002,tls-port=0,addr=0.0.0.0,disable-ticketing,agent-mouse=on
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -vga
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   std
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -boot
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   order=dc
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -soundhw
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   hda
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -smp
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   2,maxcpus=2
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
e1000,id=nic0,netdev=net0,mac=00:16:3e:38:af:5f
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -netdev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
type=tap,id=net0,ifname=vif9.0-emu,script=no,downscript=no
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
ich9-usb-ehci1,id=ehci,bus=pci.0,addr=1d.7
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
ich9-usb-uhci1,masterbus=ehci.0,firstport=0,bus=pci.0,multifunction=on,addr=1d.5
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
ich9-usb-uhci2,masterbus=ehci.0,firstport=2,bus=pci.0,addr=1d.1
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
ich9-usb-uhci3,masterbus=ehci.0,firstport=4,bus=pci.0,addr=1d.2
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
spicevmc,name=usbredir,id=usbredirchardev1
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
usb-redir,chardev=usbredirchardev1,id=usbredirdev1,bus=ehci.0
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
spicevmc,name=usbredir,id=usbredirchardev2
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
usb-redir,chardev=usbredirchardev2,id=usbredirdev2,bus=ehci.0
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
spicevmc,name=usbredir,id=usbredirchardev3
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
usb-redir,chardev=usbredirchardev3,id=usbredirdev3,bus=ehci.0
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
spicevmc,name=usbredir,id=usbredirchardev4
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
usb-redir,chardev=usbredirchardev4,id=usbredirdev4,bus=ehci.0
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
spicevmc,name=usbredir,id=usbredirchardev5
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
usb-redir,chardev=usbredirchardev5,id=usbredirdev5,bus=ehci.0
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
spicevmc,name=usbredir,id=usbredirchardev6
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
usb-redir,chardev=usbredirchardev6,id=usbredirdev6,bus=ehci.0
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   virtio-serial
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -chardev
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
spicevmc,id=vdagent,name=vdagent
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -device
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
virtserialport,chardev=vdagent,name=com.redhat.spice.0
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -M
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   xenfv
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -m
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   2032
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm:   -drive
libxl: debug: libxl_dm.c:1158:libxl__spawn_local_dm: 
file=/mnt/vm/disks/W7.disk1.xm,if=ide,index=0,media=disk,format=raw,cache=writeback

Vdagent is working, I'll do a patch for insert it in libxl, 
usbredirection with cfg file is working without (with config above) not, 
what I do wrong?
Based on this qemu options is there something other that can be a 
problem with spice and need to be improved?

Thanks for any reply.


[-- Attachment #2: Firma crittografica S/MIME --]
[-- Type: application/pkcs7-signature, Size: 4510 bytes --]

Re: [Qemu-devel] Advice on some configuration parameters

[Paolo Bonzini]

Il 12/02/2013 14:05, Fabio Fantoni ha scritto:
> I'm making patches to enable some qemu upstream features in xen that are
> missing in libxl.
> I'm trying to do it just by giving arguments to qemu, and I want them to
> be dynamic (e.g. without physical addresses if possible) and concise.

They should use physical addresses, otherwise changing the VM hardware
in trivial ways may cause a waterfall effect and cause Windows to
reactivate.

> If I try to set up virt-manager so that it doesn't point to a file, it
> uses this configuration:
> 
> -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
> -device
> ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,addr=0x5
> 
> -device ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
> -device ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2
> 
> This seems to differ from the configuration file, e.g. you have
> multifunzion=on only on the first uhci

That doesn't matter.

> , besides I can't understand how many physical ports are defined.

There are 6 ports, controlled by either the EHCI for USB 2.0 devices
(the EHCI controls 6 ports) or by one of the three UHCI for USB 1.1
ports (one UHCI has 2 ports).

> I tried setting 9 usbredirection
> channels in virt-manager but that configuration doesn't change and no
> errors are reported, but I doubt that physical ports are being define
> dynamically.

QEMU creates hubs automatically I think in that case.

> About spice vdagent at the moment I'm using this configuration that
> seems to work:
> -device virtio-serial -chardev spicevmc,id=vdagent,name=vdagent -device
> virtserialport,chardev=vdagent,name=com.redhat.spice.0
> 
> But I noticed that virt-manager use this slightly different one:
> -chardev pty,id=charserial0 -device
> isa-serial,chardev=charserial0,id=serial0 -chardev
> spicevmc,id=charchannel0,name=vdagent -device
> virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.spice.0
> 
> Are there any problem with the more concise one I'm using?

It is the same.

Paolo

Re: [Qemu-devel] Advice on some configuration parameters

[Fabio Fantoni]

[-- Attachment #1: Type: text/plain, Size: 2551 bytes --]

Il 01/03/2013 13:45, Paolo Bonzini ha scritto:
> Il 12/02/2013 14:05, Fabio Fantoni ha scritto:
>> I'm making patches to enable some qemu upstream features in xen that are
>> missing in libxl.
>> I'm trying to do it just by giving arguments to qemu, and I want them to
>> be dynamic (e.g. without physical addresses if possible) and concise.
> They should use physical addresses, otherwise changing the VM hardware
> in trivial ways may cause a waterfall effect and cause Windows to
> reactivate.
>
>> If I try to set up virt-manager so that it doesn't point to a file, it
>> uses this configuration:
>>
>> -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
>> -device
>> ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,addr=0x5
>>
>> -device ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
>> -device ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2
>>
>> This seems to differ from the configuration file, e.g. you have
>> multifunzion=on only on the first uhci
> That doesn't matter.
>
>> , besides I can't understand how many physical ports are defined.
> There are 6 ports, controlled by either the EHCI for USB 2.0 devices
> (the EHCI controls 6 ports) or by one of the three UHCI for USB 1.1
> ports (one UHCI has 2 ports).
>
>> I tried setting 9 usbredirection
>> channels in virt-manager but that configuration doesn't change and no
>> errors are reported, but I doubt that physical ports are being define
>> dynamically.
> QEMU creates hubs automatically I think in that case.
>
>> About spice vdagent at the moment I'm using this configuration that
>> seems to work:
>> -device virtio-serial -chardev spicevmc,id=vdagent,name=vdagent -device
>> virtserialport,chardev=vdagent,name=com.redhat.spice.0
>>
>> But I noticed that virt-manager use this slightly different one:
>> -chardev pty,id=charserial0 -device
>> isa-serial,chardev=charserial0,id=serial0 -chardev
>> spicevmc,id=charchannel0,name=vdagent -device
>> virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.spice.0
>>
>> Are there any problem with the more concise one I'm using?
> It is the same.
>
> Paolo
>
>
>
Thanks for reply.
I have another questions:
What exactly does the -usb parameter that seems to be necessary for usb 
passthrough?
I mean usb devices passed through host (with -usbdevice).
Is the number of such passthrough usb devices limiting the number of 
available channel for usb redirection devices?



[-- Attachment #2: Firma crittografica S/MIME --]
[-- Type: application/pkcs7-signature, Size: 4510 bytes --]

Re: [Qemu-devel] Advice on some configuration parameters

[Paolo Bonzini]

Il 01/03/2013 16:50, Fabio Fantoni ha scritto:
>>
>>
> Thanks for reply.
> I have another questions:
> What exactly does the -usb parameter that seems to be necessary for usb
> passthrough?
> I mean usb devices passed through host (with -usbdevice).
> Is the number of such passthrough usb devices limiting the number of
> available channel for usb redirection devices?

No idea. Hans?

Paolo

Re: [Qemu-devel] Advice on some configuration parameters

[Alon Levy]

> Il 01/03/2013 13:45, Paolo Bonzini ha scritto:
> > Il 12/02/2013 14:05, Fabio Fantoni ha scritto:
> >> I'm making patches to enable some qemu upstream features in xen
> >> that are
> >> missing in libxl.
> >> I'm trying to do it just by giving arguments to qemu, and I want
> >> them to
> >> be dynamic (e.g. without physical addresses if possible) and
> >> concise.
> > They should use physical addresses, otherwise changing the VM
> > hardware
> > in trivial ways may cause a waterfall effect and cause Windows to
> > reactivate.
> >
> >> If I try to set up virt-manager so that it doesn't point to a
> >> file, it
> >> uses this configuration:
> >>
> >> -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
> >> -device
> >> ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,addr=0x5
> >>
> >> -device
> >> ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
> >> -device
> >> ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2
> >>
> >> This seems to differ from the configuration file, e.g. you have
> >> multifunzion=on only on the first uhci
> > That doesn't matter.
> >
> >> , besides I can't understand how many physical ports are defined.
> > There are 6 ports, controlled by either the EHCI for USB 2.0
> > devices
> > (the EHCI controls 6 ports) or by one of the three UHCI for USB 1.1
> > ports (one UHCI has 2 ports).
> >
> >> I tried setting 9 usbredirection
> >> channels in virt-manager but that configuration doesn't change and
> >> no
> >> errors are reported, but I doubt that physical ports are being
> >> define
> >> dynamically.
> > QEMU creates hubs automatically I think in that case.
> >
> >> About spice vdagent at the moment I'm using this configuration
> >> that
> >> seems to work:
> >> -device virtio-serial -chardev spicevmc,id=vdagent,name=vdagent
> >> -device
> >> virtserialport,chardev=vdagent,name=com.redhat.spice.0
> >>
> >> But I noticed that virt-manager use this slightly different one:
> >> -chardev pty,id=charserial0 -device
> >> isa-serial,chardev=charserial0,id=serial0 -chardev
> >> spicevmc,id=charchannel0,name=vdagent -device
> >> virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.spice.0
> >>
> >> Are there any problem with the more concise one I'm using?
> > It is the same.
> >
> > Paolo
> >
> >
> >
> Thanks for reply.
> I have another questions:
> What exactly does the -usb parameter that seems to be necessary for
> usb
> passthrough?

It is shorthand for the appropriate "-device" to create the usb host (i.e. uhci etc.)

> I mean usb devices passed through host (with -usbdevice).
> Is the number of such passthrough usb devices limiting the number of
> available channel for usb redirection devices?

AFAIR yes.

> 
> 
> 

Re: [Qemu-devel] Advice on some configuration parameters

[Fabio Fantoni]

[-- Attachment #1: Type: text/plain, Size: 3048 bytes --]

Il 03/03/2013 11:27, Alon Levy ha scritto:
>> Il 01/03/2013 13:45, Paolo Bonzini ha scritto:
>>> Il 12/02/2013 14:05, Fabio Fantoni ha scritto:
>>>> I'm making patches to enable some qemu upstream features in xen
>>>> that are
>>>> missing in libxl.
>>>> I'm trying to do it just by giving arguments to qemu, and I want
>>>> them to
>>>> be dynamic (e.g. without physical addresses if possible) and
>>>> concise.
>>> They should use physical addresses, otherwise changing the VM
>>> hardware
>>> in trivial ways may cause a waterfall effect and cause Windows to
>>> reactivate.
>>>
>>>> If I try to set up virt-manager so that it doesn't point to a
>>>> file, it
>>>> uses this configuration:
>>>>
>>>> -device ich9-usb-ehci1,id=usb,bus=pci.0,addr=0x5.0x7
>>>> -device
>>>> ich9-usb-uhci1,masterbus=usb.0,firstport=0,bus=pci.0,multifunction=on,addr=0x5
>>>>
>>>> -device
>>>> ich9-usb-uhci2,masterbus=usb.0,firstport=2,bus=pci.0,addr=0x5.0x1
>>>> -device
>>>> ich9-usb-uhci3,masterbus=usb.0,firstport=4,bus=pci.0,addr=0x5.0x2
>>>>
>>>> This seems to differ from the configuration file, e.g. you have
>>>> multifunzion=on only on the first uhci
>>> That doesn't matter.
>>>
>>>> , besides I can't understand how many physical ports are defined.
>>> There are 6 ports, controlled by either the EHCI for USB 2.0
>>> devices
>>> (the EHCI controls 6 ports) or by one of the three UHCI for USB 1.1
>>> ports (one UHCI has 2 ports).
>>>
>>>> I tried setting 9 usbredirection
>>>> channels in virt-manager but that configuration doesn't change and
>>>> no
>>>> errors are reported, but I doubt that physical ports are being
>>>> define
>>>> dynamically.
>>> QEMU creates hubs automatically I think in that case.
>>>
>>>> About spice vdagent at the moment I'm using this configuration
>>>> that
>>>> seems to work:
>>>> -device virtio-serial -chardev spicevmc,id=vdagent,name=vdagent
>>>> -device
>>>> virtserialport,chardev=vdagent,name=com.redhat.spice.0
>>>>
>>>> But I noticed that virt-manager use this slightly different one:
>>>> -chardev pty,id=charserial0 -device
>>>> isa-serial,chardev=charserial0,id=serial0 -chardev
>>>> spicevmc,id=charchannel0,name=vdagent -device
>>>> virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.spice.0
>>>>
>>>> Are there any problem with the more concise one I'm using?
>>> It is the same.
>>>
>>> Paolo
>>>
>>>
>>>
>> Thanks for reply.
>> I have another questions:
>> What exactly does the -usb parameter that seems to be necessary for
>> usb
>> passthrough?
> It is shorthand for the appropriate "-device" to create the usb host (i.e. uhci etc.)

Thanks for the reply, I tried -usb and it seems to create an usb1 
controller with 2 ports and not an enhanced usb2 with 6 ports.

>> I mean usb devices passed through host (with -usbdevice).
>> Is the number of such passthrough usb devices limiting the number of
>> available channel for usb redirection devices?
> AFAIR yes.
>
>>
>>
>



[-- Attachment #2: Firma crittografica S/MIME --]
[-- Type: application/pkcs7-signature, Size: 4510 bytes --]