[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 0/4] dom0 linux: Reserve device for guest with device path



On Fri, 26 Dec 2008 17:33:20 +0000
"Andrew Lyon" <andrew.lyon@xxxxxxxxx> wrote:

> On Fri, Dec 26, 2008 at 4:28 AM, Yuji Shimada
> <shimada-yxb@xxxxxxxxxxxxxxx> wrote:
> > Please find attached a example script to convert SBDF to device path.
> >
> > The script can be executed as follows.
> >
> > # chmod 755 device_path.py
> > # device_path.py 07:00.0
> > PNP0A03:100-1c.1-00.0
> 
> When I run this script it fails because /sys/firmware/acpi/namespace/
> is missing, I have checked kernel options and I cannot see why it is
> not here, do you have any suggestion as to what I need to enable?

device_path.py depends on the following patch, and it is for
linux-2.6.18-xen.hg.

    [PATCH 1/4] dom0 linux: Expose HID, UID, SEG, BBN of PCI root bridge
    via sysfs.

Please download the source code from linux-2.6.18-xen.hg, and apply
the patch to the code. Then build and install normally, and execute
device_path.py on dom0 linux.


> I am using 2.6.27.

ACPI driver is changed from 2.6.18, and the directory where PCI root
bridge resides is changed.

    2.6.18:
        /sys/firmware/acpi/namespace/ACPI/_SB

    2.6.27:
        /sys/bus/acpi/devices

I don't think it is difficult to create the patches for
linux-2.6.27-xen.hg. It is easy to make device_path.py support two
versions of dom0 linux.

Thanks,
--
Yuji Shimada

> andy
> 
> >
> > I will take a vacation from 27 Dec. to 4 Jan. I will not be able to
> > respond to a mail.
> >
> > Thanks,
> > --
> > Yuji Shimada
> >
> > On Thu, 25 Dec 2008 16:08:01 +0900
> > Yuji Shimada <shimada-yxb@xxxxxxxxxxxxxxx> wrote:
> >
> >> This series of patches add the function to dom0 linux to reserve PCI
> >> device for guest domain with device path. Following is example of device
> >> path.
> >>
> >>         PNP0A03:100-1c.1-00.0
> >>
> >>
> >> Currently if we assign PCI device to guest domain, we have to specify
> >> PCI device with "pciback.hide=" boot parameter. If PCI device has small
> >> memory resource and BIOS does not assign page-aligned memory resource,
> >> we have to specify PCI device with "reassigndev=" boot parameter also.
> >>
> >> "pciback.hide=" and "reassigndev=" use SBDF(Segment#, Bus#, Device#,
> >> Function#). Segment# and Bus# can be changed at boot time. So they are
> >> not suitable for specifying PCI device at boot time.
> >>
> >>
> >> My patches integrate "pciback.hide=" and "reassigndev=" into one
> >> "guestdev=" boot parameter. So we have to specify only one boot
> >> parameter. The format of device path is following.
> >>
> >>         <hid>[:<uid>]-<dev>.<func>[-<dev>.<func>[...]]
> >>
> >> Device path is not changed at boot time. So it is suitable for
> >> specifying PCI device at boot time.
> >>
> >>
> >> My patches add "reassign_resources" boot parameter too. If reassignment
> >> is needed, please set it.
> >>
> >>
> >> Following is example of dom0 linux's boot parameter.
> >>
> >>         reassign_resources guestdev=PNP0A03:100-1c.1-00.0,PNP0A03:100-1d.7
> >>
> >>
> >> The series consists of following patches.
> >>
> >> 1. Expose HID,UID,SEG,BBN of PCI root bridge via sysfs.
> >> 2. Add "guestdev=" boot parameter.
> >> 3. Bind PCI back driver to PCI device specified by "guestdev=".
> >> 4. Cleanup function name relating resource reassignment.
> >>
> >>
> >> I will take a vacation from 27 Dec. to 4 Jan. I will not be able to
> >> respond to a mail.
> >>
> >> Thanks,
> >> --
> >> Yuji Shimada.
> >>
> >>
> >> _______________________________________________
> >> Xen-devel mailing list
> >> Xen-devel@xxxxxxxxxxxxxxxxxxx
> >> http://lists.xensource.com/xen-devel
> > _______________________________________________
> > Xen-devel mailing list
> > Xen-devel@xxxxxxxxxxxxxxxxxxx
> > http://lists.xensource.com/xen-devel
> >
> >
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-devel

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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.