RE: [BugDB] Segfault on graceful restart (PR#434)

[Cliff Woolley]

>>> [EMAIL PROTECTED] 07/26/00 06:24PM >>> >Take a look at the patch I just posted a minute ago, it takes care of your concerns >(which were mine as well)   I have been looking at it.  It might very well be correct...   The problem appears to be that (at least in the case of static linking) SSL_CTX_new() is returning NULL in ssl_engine_init.c lines 530 or 532, and then a NULL ctx is being passed into SSL_CTX_set_options (aka SSL_CTX_ctrl via a macro) on line 533, and SSL_CTX_ctrl attempts to dereference the NULL context pointer; blam, segfault.   I'm trying to figure out how it could be that SSL_CTX_new() would return NULL in this case.  Maybe it's because too much of the library has been shut down, and for some reason in a static build it never gets reinitialized, while in a DSO build the whole library gets unloaded and restarted.  (Is that true?)  That would account for the fact that your #ifdef solves the problem.   Didn't you say, though, that even with the patch in DSO mode there's still memory that gets leaked?  Or did I misunderstand?   --Cliff