Still seems wrong: Caller passes len=500, generic_sb_sasl_write() returns 0 or -1 with LDAP_PVT_SASL_PARTIAL_WRITE, caller collects 300 more bytes and passes len=800, then generic_sb_sasl_write() ignores not 500 but 800 bytes and tells caller that they were all written.
Either make p->flags a ber_slen_t with the number of incoming bytes to ignore, or after p->ops->encode() = success, write() = <partial write or non-fatal error> return len-1 with with LDAP_PVT_SASL_PARTIAL_WRITE.
Hallvard B Furuseth wrote:
Still seems wrong: Caller passes len=500, generic_sb_sasl_write() returns 0 or -1 with LDAP_PVT_SASL_PARTIAL_WRITE, caller collects 300 more bytes and passes len=800, then generic_sb_sasl_write() ignores not 500 but 800 bytes and tells caller that they were all written.
Technically this is true, if you look at it in isolation, but overall the code doesn't work that way. Only ber_flush2() is ever used for actual I/O, and that can only flush one BerElement at a time, and the BerElement doesn't change. I.e., you can't "collect 300 more bytes" and pass it in on the retry, it's simply not possible in the API.
Either make p->flags a ber_slen_t with the number of incoming bytes to ignore, or after p->ops->encode() = success, write() =<partial write or non-fatal error> return len-1 with with LDAP_PVT_SASL_PARTIAL_WRITE.
-
Hallvard B Furuseth -
Howard Chu