On Sat, Oct 14, 2000 at 09:28:32AM +0000, Marc Mutz wrote:
> Alexander S A Kjeldaas wrote:
> >
> > 2.2.17.8 has now been released. This patch includes some
> > reorganization of the code, but also some important API changes. Or
> > rather, the interface is now defined :-). The encrypt/decrypt
> > functions are allowed to sleep, and they have to tackle all
> > input/output sizes. This is the new interface as defined in crypto.h:
> >
> <snip>
>
> This release breaks my blowfish-encrypted filesystem. I don't get any
> error messages from mount, like I did in pre2, besides "wrong
> superblock". There is not even a
>
> Q> VFS: Can't find an ext2 fs on loop...
>
> in /var/log/messages, as it usually shows up when you get the passphrase
> wrong. Strange.
>
It was an unused code-path in loop_gen.c that was triggered by the
changes in 2.2.17.8. A magic block-size of 0 no longer indicates that
a cipher supports encrypting variable-length blocks. I think the
following patch should fix the problem:
--- int.2.2.17.8/drivers/block/loop_gen.c Mon Oct 16 13:57:41 2000
+++ int.2.2.17.x/drivers/block/loop_gen.c Mon Oct 16 13:54:37 2000
@@ -17,7 +17,6 @@
{
struct cipher_context * cx = (struct cipher_context *) lo->key_data;
struct cipher_implementation *ci = cx->ci;
- int bsize = ci->blocksize;
int (*encdecfunc)(struct cipher_context *cx, const u8 *in, u8 *out, int
size);
char *in, *out;
@@ -35,20 +34,7 @@
return -EINVAL;
}
cx->iv[0] = real_block;
- if (bsize) {
- while (size >= bsize) {
- encdecfunc(cx, in, out, bsize);
- in += bsize;
- out += bsize;
- size -= bsize;
- }
- if (size != 0) {
- return -EINVAL;
- }
- } else { /* variable-size cipher */
- encdecfunc(cx, in, out, size);
- }
-
+ encdecfunc(cx, in, out, size);
return 0;
}
astor
--
Alexander Kjeldaas Mail: astor@xxxxxxx
finger astor@xxxxxxxxxxxxxxxxx for OpenPGP key.
Linux-crypto: cryptography in and on the Linux system
Archive: http://mail.nl.linux.org/linux-crypto/
