gasp_map_vdma()

Map a guest physical address range into the qvm process address space for virtual hardware access

Synopsis:

#include <qvm/gasp.h>

void* gasp_map_vdma(const vdev_t *vdev,
                    int prot,
                    uint64_t guest_paddr,
                    size_t length)

Arguments:

vdev: A pointer to a vdev structure for a vdev that will make DMA-like transactions.
prot: PROT_* protection bits for the mappings, optionally GMF_MAP_ANY.
guest_paddr: The starting guest physical address to be mapped in.
length: The number of bytes to be mapped in.

Library:

Provided by qvm; no external library is required.

Description:

Use this function to map a guest physical address range into the qvm process (guest-physical) address space to provide virtual hardware access for vdevs that will make DMA-like transactions.

If you need to obtain information about the host-physical addresses, use gasp_map_contig().

Use this function if your vdev argument is not NULL. If your vdev argument is NULL, use gasp_map_sys().

When deciding whether to use gasp_map_sys() or gasp_map_vdma(), consider the following:

If you have allocated the guest-physical address, you should use gasp_map_sys(), unless you're in a context where it makes sense to allow the guest to decide if mapping is permitted.
If the guest has provided the guest-physical address you want to map, you should use gasp_map_vdma(), because with this function the mapping goes through the guest-controlled simulated IOMMU.

The above are guidelines, and your design may require that you use the function not recommended in the guidelines. If you don't follow the guidelines, though, be sure that you can justify your decision, and understand the implications of not following them.

Call gasp_unmap() to unmap the address range.

Returns:

A qvm process pointer that maps the guest physical address, or NULL if unsuccessful.