Matt Johnston wrote: > On Fri, Aug 04, 2006 at 10:09:08AM +0200, Hans-Christian Egtvedt wrote: > >> So the error corrects it self when I ssh localhost, but external targets >> is a no-go. >> >> Could this be an alligment error? > > Possibly? To try and figure what's going on, I've put a > version with some copius debugging of the RSA bits at > http://matt.ucc.asn.au/dropbear/testing/db-unrandom1.tar.gz > The random number generator has been made totally > deterministic, so subsequent runs on different hosts can be > compared. > > Compile the server binary for your workstation and compile > the client for both the avr32 and the workstation. Run the > server on your workstation using the included "constant_rsa" > key - then I can compare what I'm seeing. You might want to > run it with "2> /dev/null" as it'll print a lot of junk.
Done > Then run the debugging client both on the avr32 and your > workstation, as "./dbclient -v workstation 2> logfile". > If it's working correctly, you should see mostly the same > output on each host (I've put a copy of what I see in that > website dir). If it seems to hang for ages before showing a > password prompt, you might need to press "y" for the hidden > key confirmation prompt. You were right about the confirmation prompt > I'm guessing that you'll probably see different big-number > output on the avr32, which might help narrow things down. > I'm not entirely sure what to be looking for, but with some > extra dropbear_trace() and print_mp_int() calls in > libtommath/bn_mp_exptmod_fast.c it should be possible to > narrow down the differences to a particular bit of code. I put the logs here: http://www.egtvedt.no/dropbear/ > Hopefully that might be of use. I'm not quite seeing what I should be looking for, if anybody have a pointer to what I should see (correct calculations etc) then I can start my debugging :) -- With kind regards, Med vennlig hilsen, Hans-Christian Egtvedt Applications Engineer - AVR Applications Lab Atmel Norway
