Re: signify implementation details and memory usage

2017-07-25 Thread Ted Unangst
Fabio Scotoni wrote: > Is there any particular reason why things are being done this way? I > could imagine that it's to stay compatible with upstream SUPERCOP, but > mod_ed25519.c does not seem to have changed in CVS for over three years. Yes, the idea is you should be able to diff the files with

signify implementation details and memory usage

2017-07-25 Thread Fabio Scotoni
Hi misc@, I've been looking into signify(1) recently. I noticed that signmsg() in signify.c allocates a buffer sigbuf that is msglen + SIGBYTES in length. It then calls crypto_sign_ed25519(sigbuf, &siglen, msg, msglen, seckey). crypto_sign_ed25519() in mod_ed25519.c copies the entirety of msg, whi