hugetlb, userfaultfd: fix reservation restore on userfaultfd error
authorMina Almasry <almasrymina@google.com>
Sat, 20 Nov 2021 00:43:43 +0000 (16:43 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 20 Nov 2021 18:35:54 +0000 (10:35 -0800)
commitcc30042df6fcc82ea18acf0dace831503e60a0b7
tree5b7527908eb8b3037d97befab83b1f5e7ec60a23
parentafe041c2d0febd83698b8b0164e6b3b1dfae0b66
hugetlb, userfaultfd: fix reservation restore on userfaultfd error

Currently in the is_continue case in hugetlb_mcopy_atomic_pte(), if we
bail out using "goto out_release_unlock;" in the cases where idx >=
size, or !huge_pte_none(), the code will detect that new_pagecache_page
== false, and so call restore_reserve_on_error().  In this case I see
restore_reserve_on_error() delete the reservation, and the following
call to remove_inode_hugepages() will increment h->resv_hugepages
causing a 100% reproducible leak.

We should treat the is_continue case similar to adding a page into the
pagecache and set new_pagecache_page to true, to indicate that there is
no reservation to restore on the error path, and we need not call
restore_reserve_on_error().  Rename new_pagecache_page to
page_in_pagecache to make that clear.

Link: https://lkml.kernel.org/r/20211117193825.378528-1-almasrymina@google.com
Fixes: c7b1850dfb41 ("hugetlb: don't pass page cache pages to restore_reserve_on_error")
Signed-off-by: Mina Almasry <almasrymina@google.com>
Reported-by: James Houghton <jthoughton@google.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Wei Xu <weixugc@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/hugetlb.c