>>>> This looks simular to the AES problem with had, with a length of 0?
For reference. What was going on was the RC4_set_key was generating compact key schedule on Intel legacy CPU, while rc4_md5_enc was treating as non-compact with 32-bit elements. As results it was messed up in such way that induced crash in *subsequent* call to RC4. >>>> More details are at: >>>> http://bugs.debian.org/666405 >>> http://cvs.openssl.org/chngview?cn=22424 is combined patch. >> This fails like this on non-i386/x86_64 arches: >> ../libcrypto.a(e_rc4_hmac_md5.o): In function `EVP_rc4_hmac_md5': >> /build/buildd-openssl_1.0.1a-1-ia64-b35uCp/openssl-1.0.1a/crypto/evp/e_rc4_hmac_md5.c:294: >> undefined reference to `OPENSSL_ia32cap_P' >> /build/buildd-openssl_1.0.1a-1-ia64-b35uCp/openssl-1.0.1a/crypto/evp/e_rc4_hmac_md5.c:294: >> undefined reference to `OPENSSL_ia32cap_P' Once in a while you get too "comfy" and make such blunder. Apologies! > I've used this patch to fix it: I've settled for http://cvs.openssl.org/chngview?cn=22455. ______________________________________________________________________ OpenSSL Project http://www.openssl.org Development Mailing List openssl-dev@openssl.org Automated List Manager majord...@openssl.org
