Re: loop-aes mount problem

GNUtoo wrote:
> I've re(cross)compiled the custom 2.4.19-rmk7-omap1 linux kernel
> and removed the loop module that is only needed if you install
> a new system...

Can you post here those commands that you used to compile the kernel.
ARCH=/CROSS_COMPILE=/CC=/LD= and such make overrides being the interesting
parts.

Also, please include gcc parameters that the kernel build uses to compile
one .c source file, for example kernel/signal.c

> I've cross compiled the following programs or modules:
> -loop-aes v3.2c module(it lives at /lib/modules/2.4.19-rmk7-omap1/block/loop.o

Can you post here the command that you used to compile the module.
ARCH=/CROSS_COMPILE=/CC=/LD= and such make overrides being the interesting
parts.

Also, please include the output of the make command so that I can see that
gcc parameters match those used to build kernel sources.

> insmod works fine: dmesg shows: "loop: loaded(max 8 devices)"

OK. some code ran OK in the kernel module.

> but the mount command doesn't work:
> it asks for the password, then use the cpu at 100% and it is used by
> system not application according to a qopia application named system
> info...but didn't give me back my prompt
> 
> with the help of ##crypto in freenode i've tried the following command:
> losetup -e aes128 -K key.gpg /dev/loop/6 /dev/ahd/disc0/part3
> and again it asked for the password,used the cpu at 100%
> and didn't give me back my prompt...

How long did you wait?

The processor that you are using isn't a fast one. The slowest box that I
have access to, and that still boots, is 1991 vintage Intel 486 33 MHz.
Decrypting a gpg encrypted key file takes almost 3 minutes on that poor
machine. See:

# date; gpg --decrypt --passphrase-fd=3 3<foo <key.gpg >/dev/null; date
Wed Jul  9 08:13:04 EEST 2008
Reading passphrase from file descriptor 3
gpg: AES256 encrypted data
gpg: encrypted with 1 passphrase
Wed Jul  9 08:15:57 EEST 2008
# cat /proc/cpuinfo | grep bogomips
bogomips        : 16.53

The optional gpg patch that is included in loop-AES tarball changes default
passphrase iteration count so that dictionary attacks against a key file
will be much slower. It does have a side effect that decrypting a key file
on a slow machine can be very slow.

What happens if you create a symmetrically encrypted key file using
unpatched gpg, or using a patched one but with these gpg options:

 --s2k-count 65536

That option reverts the passphrase iteration count back to original default,
which is more suitable on slow computers.

-- 
Jari Ruusu  1024R/3A220F51 5B 4B F9 BB D3 3F 52 E9  DB 1D EB E3 24 0E A9 DD

-
Linux-crypto:  cryptography in and on the Linux system
Archive:       http://mail.nl.linux.org/linux-crypto/