Re: [Xen-devel] [PATCH] acpi: do not flush cache if cx->type != ACPI_STATE_C3

[Steven <wangwangkang@xxxxxxxxx>]

Hi, Wei,
I have read you slides in xen summit 2007 about NPT in AMD, "AMD
Barcelona and Nested Paging Support in Xen".
However, I have some question regarding the nested page walk in your slide 8.

In your slide 8, the first step is to get gPA from get_PML4(gCR3,gVA).
I assume that it use the [47:39] bit of gVA.
However, in another AMD white paper, "AMD-VTM Nested Paging".
http://developer.amd.com/assets/NPT-WP-1%201-final-TM.pdf
In its figure 4, I saw that the first step is to translate gCR3 using
nested page walk and then combine with the gVA[47:39] to read the
table entry.

These two documents look having different order of reading the guest
page table. In the slides, it first get_PML4. But in the white paper,
it first does nested page walk.
I am wondering which one is true. Thanks.

- ha

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