vfs: iov_iter: have iov_iter_advance decrement nr_segs appropriately
authorJeff Layton <jlayton@redhat.com>
Thu, 27 Oct 2011 21:53:08 +0000 (23:53 +0200)
committerChristoph Hellwig <hch@serles.lst.de>
Fri, 28 Oct 2011 11:55:08 +0000 (13:55 +0200)
Currently, when you call iov_iter_advance, then the pointer to the iovec
array can be incremented, but it does not decrement the nr_segs value in
the iov_iter struct. The result is a iov_iter struct with a nr_segs
value that goes beyond the end of the array.

While I'm not aware of anything that's specifically broken by this, it
seems odd and a bit dangerous not to decrement that value. If someone
were to trust the nr_segs value to be correct, then they could end up
walking off the end of the array.

Changing this might also provide some micro-optimization when dealing
with the last iovec in an array. Many of the other routines that deal
with iov_iter have optimized codepaths when nr_segs == 1.

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
mm/filemap.c

index 7771871fa3535d4895ba0d5e92b6a84870980dfb..5cf820a7c8ec2c1c0e8744eb03750e35b9255ef9 100644 (file)
@@ -2115,6 +2115,7 @@ void iov_iter_advance(struct iov_iter *i, size_t bytes)
        } else {
                const struct iovec *iov = i->iov;
                size_t base = i->iov_offset;
+               unsigned long nr_segs = i->nr_segs;
 
                /*
                 * The !iov->iov_len check ensures we skip over unlikely
@@ -2130,11 +2131,13 @@ void iov_iter_advance(struct iov_iter *i, size_t bytes)
                        base += copy;
                        if (iov->iov_len == base) {
                                iov++;
+                               nr_segs--;
                                base = 0;
                        }
                }
                i->iov = iov;
                i->iov_offset = base;
+               i->nr_segs = nr_segs;
        }
 }
 EXPORT_SYMBOL(iov_iter_advance);