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

[Xen-users] VT-d CPU resources


  • To: <xen-users@xxxxxxxxxxxxx>
  • From: "amiard brice" <perso@xxxxxxxxxxxxxxx>
  • Date: Fri, 22 Jun 2012 15:58:30 +0200
  • Delivery-date: Fri, 22 Jun 2012 13:59:25 +0000
  • List-id: Xen user discussion <xen-users.lists.xen.org>
  • Thread-index: Ac1QfWvBOobu8r8gQ0SJhk8GYOufHw==

Hello

 

I'm working on XEN since one month now.

 

I want create four domU with 2 PCI device per dom.

I've bought a backplane with 3 bridges PCIe 1x to PCI and 4 PCI directly connected to the PCH (Q67).

(Moreover PCI card (ethernet card) are composed with a PCI bridge)

 

I've succefuly created 4 domU with bakend devices, with option "pci-backend-strict-check" disable.

I’ve try with it but even if all PCI ports behind PCIe to PCI bridge appears in  pci-list-assignable-devices, domU won’t start, with an error message "device XX:XX.X need to be co-assigned with XX:XX.X, but it's not owned by pci-backend

 

The probleme is :  even if my domU are configuer with 1 VCPUs for each of them, when domU acces to I/O devices, VCPUs aren't separated (I've notice this using virt-manager charts). During a basic CPU benchmark on a domU, VCPUs are isolated.

 

I Have attached "lspci-v", "xm dmesg", "xm pci-list-assignable-devices", "default-grub", "xm info" and a “domU cfg file”

 

A share core is very problematic my RT application

 

thank you for your help in advance

 

Attachment: default_grub.txt
Description: Text document

Attachment: lspci_v.txt
Description: Text document

Attachment: qnx_1.config
Description: Binary data

Attachment: xm_dmesg.txt
Description: Text document

Attachment: xm_info.txt
Description: Text document

Attachment: xm_pci_list_assi.txt
Description: Text document

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users

 


Rackspace

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