mm/hmm: allow snapshot of the special zero page
authorRalph Campbell <rcampbell@nvidia.com>
Wed, 23 Oct 2019 19:55:14 +0000 (12:55 -0700)
committerJason Gunthorpe <jgg@mellanox.com>
Tue, 29 Oct 2019 17:26:28 +0000 (14:26 -0300)
commitac541f2503722943a9f13e0c92ed07632ba7fd38
tree66c9f31f41501650ebfc34378d0bed0addb111cd
parentd6d5df1db6e9d7f8f76d2911707f7d5877251b02
mm/hmm: allow snapshot of the special zero page

If a device driver like nouveau tries to use hmm_range_fault() to access
the special shared zero page in system memory, hmm_range_fault() will
return -EFAULT and kill the process.

Allow hmm_range_fault() to return success (0) when the CPU pagetable entry
points to the special shared zero page.

page_to_pfn() and pfn_to_page() are defined on the zero page so just
handle it like any other page.

Link: https://lore.kernel.org/r/20191023195515.13168-3-rcampbell@nvidia.com
Signed-off-by: Ralph Campbell <rcampbell@nvidia.com>
Reviewed-by: "Jérôme Glisse" <jglisse@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
mm/hmm.c