[Xen-API] [PATCH 0 of 2] More experimental PCI passthrough fixes

more experimental PCI passthrough fixes

1. hotunplug the PCI devices in suspend; hotplug back in in resume
2. do the same in migrate (synchronously, no fancy early unplug, later sync)
3. make the low-level Device.PCI.* functions more robust by always checking for 
the xenstore response ("pci-inserted", "pci-removed")

Note that testing with an HVM SLES11 guest suggests that there's still a race 
in the hotplug code where, even though qemu acks 2 PCI device hotplugs, the 
guest only sees 1. The problem goes away if a sleep is added between the two 
hotplug requests.

Signed-off-by: David Scott <dave.scott@xxxxxxxxxxxxx>
6 files changed, 69 insertions(+), 64 deletions(-)
ocaml/xapi/vmops.ml           |   24 ++++++++++++++++++++--
ocaml/xapi/xapi_vm.ml         |   16 ++++++++++----
ocaml/xapi/xapi_vm_migrate.ml |   41 +++-----------------------------------
ocaml/xenops/device.ml        |   44 +++++++++++++++++++++++++----------------
ocaml/xenops/device.mli       |    6 +++--
ocaml/xenops/xenops.ml        |    2 -

_______________________________________________
xen-api mailing list
xen-api@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/mailman/listinfo/xen-api

<Prev in Thread]

Current Thread

[Next in Thread>

Previous by Date:

RE: [Xen-API]do NOT support vdi_introduce and vdi_import, Anthony Xu

Next by Date:

[Xen-API] [PATCH 1 of 2] CA-36384: [experimental PCI passthrough]: unplug/plug devices around suspend/resume, David Scott

Previous by Thread:

[Xen-API] [PATCH] [CA-35907] [DMC] Adjusts the HVM memory ratio to 0.25, for consistency with version 0.5 of the support requirements for dynamic memory control, Jonathan Knowles

Next by Thread:

[Xen-API] [PATCH 1 of 2] CA-36384: [experimental PCI passthrough]: unplug/plug devices around suspend/resume, David Scott

Indexes:

[Date] [Thread] [Top] [All Lists]