Revert "x86/mm: Introduce lookup_address_in_mm()"
authorSean Christopherson <seanjc@google.com>
Fri, 29 Apr 2022 14:57:53 +0000 (07:57 -0700)
committerPaolo Bonzini <pbonzini@redhat.com>
Fri, 29 Apr 2022 16:40:41 +0000 (12:40 -0400)
commit643d95aac59a060c2730975988aedc387f0f9f44
treefaabcbdbf2505a9dbfcd7a585f5a1890259483cf
parent73331c5d84cf87974dc3616ef706847ff187d590
Revert "x86/mm: Introduce lookup_address_in_mm()"

Drop lookup_address_in_mm() now that KVM is providing it's own variant
of lookup_address_in_pgd() that is safe for use with user addresses, e.g.
guards against page tables being torn down.  A variant that provides a
non-init mm is inherently dangerous and flawed, as the only reason to use
an mm other than init_mm is to walk a userspace mapping, and
lookup_address_in_pgd() does not play nice with userspace mappings, e.g.
doesn't disable IRQs to block TLB shootdowns and doesn't use READ_ONCE()
to ensure an upper level entry isn't converted to a huge page between
checking the PAGE_SIZE bit and grabbing the address of the next level
down.

This reverts commit 13c72c060f1ba6f4eddd7b1c4f52a8aded43d6d9.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <YmwIi3bXr/1yhYV/@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/include/asm/pgtable_types.h
arch/x86/mm/pat/set_memory.c