Comment 2 for bug 1011469

Revision history for this message
Yingying Zhao (yingying-zhao) wrote :

There are 3 commits for this feature:

1.
commit c892426ce32dc68487613a95c1bf6441e8b15e43
Author: Jiongxi Li <email address hidden>
Date: Mon Sep 17 21:06:02 2012 +0100

    xen: add virtual x2apic support for apicv

    basically to benefit from apicv, we need clear MSR bitmap for
    corresponding x2apic MSRs:
      0x800 - 0x8ff: no read intercept for apicv register virtualization
      TPR,EOI,SELF-IPI: no write intercept for virtual interrupt
        delivery

    Signed-off-by: Jiongxi Li <email address hidden>
    Committed-by: Keir Fraser <email address hidden>

2.
commit 25b81a7941d71fabd4c8c56e9d3d76801d061842
Author: Jiongxi Li <email address hidden>
Date: Mon Sep 17 21:05:11 2012 +0100

    xen: enable Virtual-interrupt delivery

    Virtual interrupt delivery avoids Xen to inject vAPIC interrupts
    manually, which is fully taken care of by the hardware. This needs
    some special awareness into existing interrupr injection path:
    For pending interrupt from vLAPIC, instead of direct injection, we may
    need update architecture specific indicators before resuming to guest.
    Before returning to guest, RVI should be updated if any pending IRRs
    EOI exit bitmap controls whether an EOI write should cause VM-Exit. If
    set, a trap-like induced EOI VM-Exit is triggered. The approach here
    is to manipulate EOI exit bitmap based on value of TMR. Level
    triggered irq requires a hook in vLAPIC EOI write, so that vIOAPIC EOI
    is triggered and emulated

    Signed-off-by: Gang Wei <email address hidden>
    Signed-off-by: Yang Zhang <email address hidden>
    Signed-off-by: Jiongxi Li <email address hidden>
    Committed-by: Keir Fraser <email address hidden>

3.
commit f62b02e177cd03da2b8397b9bdf1a53b541df3f3
Author: Jiongxi Li <email address hidden>
Date: Mon Sep 17 21:04:08 2012 +0100

    xen: enable APIC-Register Virtualization

    Add APIC register virtualization support
     - APIC read doesn't cause VM-Exit
     - APIC write becomes trap-like

    Signed-off-by: Gang Wei <email address hidden>
    Signed-off-by: Yang Zhang <email address hidden>
    Signed-off-by: Jiongxi Li <email address hidden>