On Thu, Sep 15, 2016 at 04:16:56PM -0400, Theodore Ts'o wrote:
On Mon, Sep 12, 2016 at 12:24:00PM -0700, Eric Biggers wrote:
quoted
- Validate fscrypt_context.format and fscrypt_context.flags. If
unrecognized values are set, then the kernel may not know how to
interpret the encrypted file, so it should fail the operation.
- Validate that AES_256_XTS is used for contents and that AES_256_CTS is
used for filenames. It was previously possible for the kernel to
accept these reversed, though it would have taken manual editing of
the block device. This was not intended.
- Fail cleanly rather than BUG()-ing if a file has an unexpected type.
Signed-off-by: Eric Biggers <redacted>
Reviewed-by: Jaegeuk Kim <jaegeuk@kernel.org>
Thanks, applied. (I plan to carry Eric's fscrypto changes ext4 git
tree; Jaeguk, I assume you have no objections?)
No objection.
Thanks,
- Ted