Thanks Ken, I’ll keep that in mind. In this particular case (and with your earlier patch applied) it appears that http_auth() in cyrus-imap’s httpd.c returns SASL_CONTINUE. The comment around line 3272 says “Need another step to complete authentication”, but the caller (response_header(), line ~2270) appears not to invoke that other step. I tested this by calling http_auth() again if it returns SASL_CONTINUE, and that authenticated me. // Johan On Jul 23, 2014, at 13:30, Ken Murchison <murch@xxxxxxxxxxxxxx> wrote: > I had issues with the Apple clients and Digest. Unless you really need Digest, I'd recommend using TLS + Basic. > > > > On 07/23/2014 01:27 PM, Johan Hattne wrote: >> Hi Ken; >> >> That fixes the crash but results in a “401 Unauthorized”. I’ll look into that a bit more at the next opportunity. >> >> This is using Contacts (8.0 1371) on an up-to-date OS X 10.9.4. It also works on the iPhone (iOS 7.1.2). >> >> // Johan >> >> >> On Jul 23, 2014, at 10:55, Ken Murchison <murch@xxxxxxxxxxxxxx> wrote: >> >>> Hi Johan, >>> >>> I believe this issue is fixed by the following commit: http://git.cyrusimap.org/cyrus-sasl/commit/?id=76ce885a44e7cb511ba54ceae46349036abb9cc8 >>> >>> BTW, which CardDAV client is using Digest? >>> >>> >>> On 07/22/2014 01:48 PM, Johan Hattne wrote: >>>> While PLAIN authentication works fine, I had the https daemon crash during DIGEST-MD5 authentication. The crash turned out to be a divide error in libdigestmd5 from cyrus-sasl. In particular (in cyrus-sasl’s plugins/digestmd5.c): >>>> >>>> /* Create an initial cache entry for non-persistent HTTP connections */ >>>> unsigned val = hash((char *) nonce) % text->reauth->size; >>>> >>>> would fail due to text->reauth->size being zero. If I’m reading this correctly, this appears to be the effect of initializing the plugin (as done in digestmd5_server_plug_init(), defined in same file as the snippet above) with an undefined reauth_timeout. And indeed, adding "sasl_reauth_timeout: 10” to /etc/imapd.conf makes the crash go away. >>>> >>>> I didn’t expect a configuration without reauth_timeout to crash imapd, but I haven’t done enough research to be sure, nor to tell where the problem lies should this be a real issue. Any further insight is greatly appreciated! >>>> >>>> // Cheers; Johan >>>> >>>> ---- >>>> Cyrus Home Page: http://www.cyrusimap.org/ >>>> List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/ >>>> To Unsubscribe: >>>> https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus >>> >>> -- >>> Kenneth Murchison >>> Principal Systems Software Engineer >>> Carnegie Mellon University >>> > > > -- > Kenneth Murchison > Principal Systems Software Engineer > Carnegie Mellon University > ---- Cyrus Home Page: http://www.cyrusimap.org/ List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/ To Unsubscribe: https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus
