On 28/10/13 14:35, Roger Pau Monné wrote: > Hello, > > The Xen community is working on a new virtualization mode (or maybe I > should say an extension of HVM) to be able to run PV guests inside HVM > containers without requiring a device-model (Qemu). One of the > advantages of this new virtualization mode is that now it is much more > easier to port guests to run under it (as compared to pure PV guests). > > Given that FreeBSD already supports PVHVM, adding PVH support is quite > easy, we only need some glue for the PV entry point and then support > for diverging some early init functions (like fetching the e820 map or > starting the APs). > > The attached patch contains all this changes, and allows a SMP FreeBSD > guest to fully boot (and AFAIK work) under this new PVH mode. The patch > can also be found on my git repo: > > git://xenbits.xen.org/people/royger/freebsd.git pvh_v2 > > The patch touches quite a lot of the early init, so I've Cced the > persons that maintain those areas, so they can review it. > > In order to test it, and since the PVH changes are not yet merged into > upstream Xen, the use of a patched Xen is necessary. I've collected the > patches for PVH guest support from George Dunlap (v13) and fixed some > bugs on top of them, the tree can be found at: > > git://xenbits.xen.org/people/royger/xen.git fix_pvh I've updated the patch (as suggested by John Baldwin) and added a Xen Nexus, that attaches all the Xen top-level devices, this gets rid of the legacy bus. The new patch can be found at: git://xenbits.xen.org/people/royger/freebsd.git pvh_v2 And also attached on this email. Thanks for the review, Roger.
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:44 UTC