mm/mremap: for MREMAP_DONTUNMAP check security_vm_enough_memory_mm()

Currently memory is accounted post-mremap() with MREMAP_DONTUNMAP, which
may break overcommit policy.  So, check if there's enough memory before
doing actual VMA copy.

Don't unset VM_ACCOUNT on MREMAP_DONTUNMAP.  By semantics, such mremap()
is actually a memory allocation.  That also simplifies the error-path a
little.

Also, as it's memory allocation on success don't reset hiwater_vm value.

Link: https://lkml.kernel.org/r/20201013013416.390574-3-dima@arista.com
Fixes: commit e346b3813067 ("mm/mremap: add MREMAP_DONTUNMAP to mremap()")
Signed-off-by: Dmitry Safonov <dima@arista.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/mm/mremap.c b/mm/mremap.c
index 975a14c..90ff09c 100644 (file)
--- a/mm/mremap.c
+++ b/mm/mremap.c
@@ -515,11 +515,19 @@ static unsigned long move_vma(struct vm_area_struct *vma,
        if (err)
                return err;
 
+       if (unlikely(flags & MREMAP_DONTUNMAP && vm_flags & VM_ACCOUNT)) {
+               if (security_vm_enough_memory_mm(mm, new_len >> PAGE_SHIFT))
+                       return -ENOMEM;
+       }
+
        new_pgoff = vma->vm_pgoff + ((old_addr - vma->vm_start) >> PAGE_SHIFT);
        new_vma = copy_vma(&vma, new_addr, new_len, new_pgoff,
                           &need_rmap_locks);
-       if (!new_vma)
+       if (!new_vma) {
+               if (unlikely(flags & MREMAP_DONTUNMAP && vm_flags & VM_ACCOUNT))
+                       vm_unacct_memory(new_len >> PAGE_SHIFT);
                return -ENOMEM;
+       }
 
        moved_len = move_page_tables(vma, old_addr, new_vma, new_addr, old_len,
                                     need_rmap_locks);
@@ -548,7 +556,7 @@ static unsigned long move_vma(struct vm_area_struct *vma,
        }
 
        /* Conceal VM_ACCOUNT so old reservation is not undone */
-       if (vm_flags & VM_ACCOUNT) {
+       if (vm_flags & VM_ACCOUNT && !(flags & MREMAP_DONTUNMAP)) {
                vma->vm_flags &= ~VM_ACCOUNT;
                excess = vma->vm_end - vma->vm_start - old_len;
                if (old_addr > vma->vm_start &&
@@ -573,34 +581,16 @@ static unsigned long move_vma(struct vm_area_struct *vma,
                untrack_pfn_moved(vma);
 
        if (unlikely(!err && (flags & MREMAP_DONTUNMAP))) {
-               if (vm_flags & VM_ACCOUNT) {
-                       /* Always put back VM_ACCOUNT since we won't unmap */
-                       vma->vm_flags |= VM_ACCOUNT;
-
-                       vm_acct_memory(new_len >> PAGE_SHIFT);
-               }
-
-               /*
-                * VMAs can actually be merged back together in copy_vma
-                * calling merge_vma. This can happen with anonymous vmas
-                * which have not yet been faulted, so if we were to consider
-                * this VMA split we'll end up adding VM_ACCOUNT on the
-                * next VMA, which is completely unrelated if this VMA
-                * was re-merged.
-                */
-               if (split && new_vma == vma)
-                       split = 0;
-
                /* We always clear VM_LOCKED[ONFAULT] on the old vma */
                vma->vm_flags &= VM_LOCKED_CLEAR_MASK;
 
                /* Because we won't unmap we don't need to touch locked_vm */
-               goto out;
+               return new_addr;
        }
 
        if (do_munmap(mm, old_addr, old_len, uf_unmap) < 0) {
                /* OOM: unable to split vma, just get accounts right */
-               if (vm_flags & VM_ACCOUNT)
+               if (vm_flags & VM_ACCOUNT && !(flags & MREMAP_DONTUNMAP))
                        vm_acct_memory(new_len >> PAGE_SHIFT);
                excess = 0;
        }
@@ -609,7 +599,7 @@ static unsigned long move_vma(struct vm_area_struct *vma,
                mm->locked_vm += new_len >> PAGE_SHIFT;
                *locked = true;
        }
-out:
+
        mm->hiwater_vm = hiwater_vm;
 
        /* Restore VM_ACCOUNT if one or two pieces of vma left */