In message <1457543989853-64500.p...@n7.nabble.com> on Wed, 9 Mar 2016 10:19:49 -0700 (MST), danigrosu <dni.gr...@gmail.com> said:
dni.grosu> Richard Levitte - VMS wrote dni.grosu> > Exactly how did it fail? It's a bit hard to diagnose unless you show dni.grosu> > us what you were told... I assume there were some error messages? dni.grosu> dni.grosu> This is what I get if I use the the git version: dni.grosu> <http://openssl.6102.n7.nabble.com/file/n64500/19.png> dni.grosu> ... and if I use the blog code for the e_md5.c file (called md5-engine.c on dni.grosu> the blog) dni.grosu> it simply works with the same commands. As I said above, I had to make some dni.grosu> modifications in order to build the engine using autotools. and you discovered why on your own: In message <1457546255766-64501.p...@n7.nabble.com> on Wed, 9 Mar 2016 10:57:35 -0700 (MST), danigrosu <dni.gr...@gmail.com> said: dni.grosu> In git version, if I comment the block dni.grosu> dni.grosu> / if (id && strcmp(id, engine_id)) { dni.grosu> fprintf(stderr, "MD5 engine called with the unexpected id %s\n", id); dni.grosu> fprintf(stderr, "The expected id is %s\n", engine_id); dni.grosu> goto end; dni.grosu> }/ dni.grosu> dni.grosu> ... then I type dni.grosu> dni.grosu> /$ gcc -fPIC -o rfc1321/md5c.o -c rfc1321/md5c.c dni.grosu> $ gcc -fPIC -o md5-engine.o -c e_md5.c dni.grosu> $ gcc -shared -o md5-engine.so -lcrypto md5-engine.o rfc1321/md5c.o dni.grosu> dni.grosu> $ echo whatever | openssl dgst -engine `pwd`/md5-engine.so -md5 dni.grosu> engine "emd5" set. dni.grosu> (stdin)= d8d77109f4a24efc3bd53d7cabb7ee35/ dni.grosu> dni.grosu> ... everithing goes well Yes. The check that you commented away isn't strictly necessary, it's very much a paranoid check. Did you notice how, in the README, the example call is this? $ OPENSSL_ENGINES=.libs openssl engine -t -c emd5 The id that the engine's init function receives is exactly what the openssl app receives as an engine name on the command line, so if you give it the full path variant (like in my blog), that's what it gets, and if you do it with the OPENSSL_ENGINES env variable, it will get the name you gave ("emd5" in the example above). But yeah, strictly speaking, the id check in the engine's init function is not necessary. Cheers, Richard -- Richard Levitte levi...@openssl.org OpenSSL Project http://www.openssl.org/~levitte/ -- openssl-dev mailing list To unsubscribe: https://mta.openssl.org/mailman/listinfo/openssl-dev
