ntfs: aops: remove VLA usage
authorKees Cook <keescook@chromium.org>
Fri, 17 Aug 2018 22:44:07 +0000 (15:44 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 17 Aug 2018 23:20:27 +0000 (16:20 -0700)
In the quest to remove all stack VLA usage from the kernel[1], this uses
the maximum size needed on the stack and adds a sanity check for
robustness: index.block_size cannot be larger than PAGE_SIZE nor less
than NTFS_BLOCK_SIZE.

[1] https://lkml.kernel.org/r/CA+55aFzCG-zNmZwX4A2FQpadafLfEzK6CC=qPXydAacU1RqZWA@mail.gmail.com

Link: http://lkml.kernel.org/r/20180626172909.41453-2-keescook@chromium.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: Anton Altaparmakov <anton@tuxera.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ntfs/aops.c

index 01c770979921dfdc379f9b3567bbb3ee375dfa9d..8946130c87ad7d10b3057ba826c4d5141dca7e91 100644 (file)
@@ -922,7 +922,7 @@ static int ntfs_write_mst_block(struct page *page,
        ntfs_volume *vol = ni->vol;
        u8 *kaddr;
        unsigned int rec_size = ni->itype.index.block_size;
-       ntfs_inode *locked_nis[PAGE_SIZE / rec_size];
+       ntfs_inode *locked_nis[PAGE_SIZE / NTFS_BLOCK_SIZE];
        struct buffer_head *bh, *head, *tbh, *rec_start_bh;
        struct buffer_head *bhs[MAX_BUF_PER_PAGE];
        runlist_element *rl;
@@ -931,6 +931,9 @@ static int ntfs_write_mst_block(struct page *page,
        bool sync, is_mft, page_is_dirty, rec_is_dirty;
        unsigned char bh_size_bits;
 
+       if (WARN_ON(rec_size < NTFS_BLOCK_SIZE))
+               return -EINVAL;
+
        ntfs_debug("Entering for inode 0x%lx, attribute type 0x%x, page index "
                        "0x%lx.", vi->i_ino, ni->type, page->index);
        BUG_ON(!NInoNonResident(ni));