fscrypt fix for v5.6-rc6

Fix a bug where if userspace is writing to encrypted files while the
 FS_IOC_REMOVE_ENCRYPTION_KEY ioctl (introduced in v5.4) is running,
 dirty inodes could be evicted, causing writes could be lost or the
 filesystem to hang due to a use-after-free.  This was encountered during
 real-world use, not just theoretical.
 
 Tested with the existing fscrypt xfstests, and with a new xfstest I
 wrote to reproduce this bug.  This fix does expose an existing bug with
 '-o lazytime' that Ted is working on fixing, but this fix is more
 critical and needed anyway regardless of the lazytime fix.
 -----BEGIN PGP SIGNATURE-----
 
 iIoEABYIADIWIQSacvsUNc7UX4ntmEPzXCl4vpKOKwUCXmk8HxQcZWJpZ2dlcnNA
 Z29vZ2xlLmNvbQAKCRDzXCl4vpKOK4YiAQC1RZyH4/mZ890Or6s8SzCgJTVmiLk9
 ZTO/56XmLte6LAD+IBAExqDkkybmAF0rQ4kY1oL75f/e/nEs+50TXra9NQc=
 =s2KD
 -----END PGP SIGNATURE-----

Merge tag 'fscrypt-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt

Pull fscrypt fix from Eric Biggers:
 "Fix a bug where if userspace is writing to encrypted files while the
  FS_IOC_REMOVE_ENCRYPTION_KEY ioctl (introduced in v5.4) is running,
  dirty inodes could be evicted, causing writes could be lost or the
  filesystem to hang due to a use-after-free. This was encountered
  during real-world use, not just theoretical.

  Tested with the existing fscrypt xfstests, and with a new xfstest I
  wrote to reproduce this bug. This fix does expose an existing bug with
  '-o lazytime' that Ted is working on fixing, but this fix is more
  critical and needed anyway regardless of the lazytime fix"

* tag 'fscrypt-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt:
  fscrypt: don't evict dirty inodes after removing key
This commit is contained in:
Linus Torvalds 2020-03-11 13:35:34 -07:00
commit e6e6ec48dd

View File

@ -538,6 +538,15 @@ int fscrypt_drop_inode(struct inode *inode)
return 0;
mk = ci->ci_master_key->payload.data[0];
/*
* With proper, non-racy use of FS_IOC_REMOVE_ENCRYPTION_KEY, all inodes
* protected by the key were cleaned by sync_filesystem(). But if
* userspace is still using the files, inodes can be dirtied between
* then and now. We mustn't lose any writes, so skip dirty inodes here.
*/
if (inode->i_state & I_DIRTY_ALL)
return 0;
/*
* Note: since we aren't holding ->mk_secret_sem, the result here can
* immediately become outdated. But there's no correctness problem with