hi, Anybody can help me understand those issues?
And here is one more question, I want to implement this inside GuestControl/service.cpp as this is the place normally we are handling guest requst, but I just found out we are already at the end of the foodchain, so I don't think we can call into kernel to run functions like "PGMHandlerPhysicalPageAliasHC". So where would be the best place to do this? 1) I think I can use a PV guest driver to communicate between guest and host as long as the PV guest can do a hypercall into host. I don't see there is clear definintion of vbox hypercall interface, do we have one? 2) Still usiing the HGCM interface, but I need to do an ioctl manually into /dev/vboxdrv device and implemnt something there. Really appreciate all your comments and suggestions. Thanks, CJ On Tue, Feb 7, 2012 at 2:22 AM, Cyclonus J <[email protected]> wrote: > hi, > > I am new to vbox and try to do some experiments about its memory > mappings. Currently, I am trying to remap a given guest virtual > address to a specific host physical location. I found that > IOMMMIOMapMMIOHCPage is doing a similar work although it only maps > MMIO range, which will eventually call PGMHandlerPhysicalPageAliasHC > function. Is this the closest one we have in vbox? > > For MMIO aliased pages, we still have to pay vm_exit every time when > guest accesses those pages even the shadow page table has correctly > populated, right? We don't have to pay such price when using EPT > (probably just one EPT violation), right? > > One more question is about those pdmdev helper functions, can I use > those function on the host side by creating a HGCM service provider? > > Thanks, > CJ _______________________________________________ vbox-dev mailing list [email protected] https://www.virtualbox.org/mailman/listinfo/vbox-dev
