Hi,

Il 26/11/2014 15:15, Giuseppe D'Angelo ha scritto:
Heh, it's a bit complex as of now (you can find it in the patch linked
to my first email, [1]), I'll try tearing down into a pure C/C++ example
(also to exclude a silly mistake on my side)...

after further investigation, OpenSSL is doing the right thing. Apparently the code path simply exposes a serious bug in Qt: when the SSL context initialization fails (in this case, because of an invalid curve specified), Qt does not recover properly from the error and the gets stuck reading from data from the underlying fd.

Sorry for the noise...

(On a related note, I wonder if for 1.0.2 it would be accepted a patch that exposes the logic implemented by tls1_ec_nid2curve_id, namely, check if a given curve NID is usable with TLS as a named curve.)

Thanks,
--
Giuseppe D'Angelo | giuseppe.dang...@kdab.com | Software Engineer
KDAB (UK) Ltd., a KDAB Group company
Tel. UK +44-1738-450410, Sweden (HQ) +46-563-540090
KDAB - Qt Experts - Platform-independent software solutions

Attachment: smime.p7s
Description: Firma crittografica S/MIME

Reply via email to