nommu: fix race between ramfs truncation and shared mmap
authorDavid Howells <dhowells@redhat.com>
Sat, 16 Jan 2010 01:01:36 +0000 (17:01 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 16 Jan 2010 20:15:40 +0000 (12:15 -0800)
commit81759b5b221107488bda99fe7beeb7b734f61133
tree2ca1207d4ae5d6bbdc49db727a35b22a502fc76f
parentefc1a3b16930c41d64ffefde16b87d82f603a8a0
nommu: fix race between ramfs truncation and shared mmap

Fix the race between the truncation of a ramfs file and an attempt to make
a shared mmap of region of that file.

The problem is that do_mmap_pgoff() calls f_op->get_unmapped_area() to
verify that the file region is made of contiguous pages and to find its
base address - but there isn't any locking to guarantee this region until
vma_prio_tree_insert() is called by add_vma_to_mm().

Note that moving the functionality into f_op->mmap() doesn't help as that
is also called before vma_prio_tree_insert().

Instead make ramfs_nommu_check_mappings() grab nommu_region_sem whilst it
does its checks.  This means that this function will wait whilst mmaps
take place.

Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ramfs/file-nommu.c