put_user() may fail.  In this case propagate error code from
privcmd_ioctl_mmap_batch().
Signed-off-by: NVasiliy Kulikov <segooon@gmail.com>
Signed-off-by: NJeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

This allows xenfs to be built as a module, previously it required flush_tlb_all
and arbitrary_virt_to_machine to be exported.
Signed-off-by: NIan Campbell <ian.campbell@citrix.com>
Signed-off-by: NJeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

On error IOCTL_PRIVCMD_MMAPBATCH is expected to set the top nibble of
the effected MFN and return 0. Currently it leaves the MFN unmodified
and returns the number of failures. Therefore:

- reimplement remap_domain_mfn_range() using direct
  HYPERVISOR_mmu_update() calls and small batches. The xen_set_domain_pte()
  interface does not report errors and since some failures are
  expected/normal using the multicall infrastructure is too noisy.
- return 0 as expected
- writeback the updated MFN list to mmapbatch->arr not over mmapbatch,
  smashing the caller's stack.
- remap_domain_mfn_range can be static.

With this change I am able to start an HVM domain.
Signed-off-by: NIan Campbell <ian.campbell@citrix.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: NJeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

Test vma->vm_ops is our operations to make sure we created it.
We don't want to stomp on other random vmas.

[ Impact: bugfix; prevent ioctl from affecting other mappings ]
Signed-off-by: NJeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

Print more detail about privcmd mapping faults for debugging.

[ Impact: debug ]
Signed-off-by: NJeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

The privcmd interface in xenfs allows the tool stack in the privileged
domain to get fairly direct access to the hypervisor in order to do
various management things such as domain construction.

[ Impact: new xenfs interface for privileged operations ]
Signed-off-by: NJeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>