Tomáš Mráz
Tomáš Mráz
Still OK with the typo fixed.
Also the CI failure is relevant.
Just change the rule in https://github.com/openssl/openssl/blob/master/util/markdownlint.rb#L9 to exclude it. I did not like requiring the indented style anyway. And please unify the markup in the document in one way or...
But this is incorrect code! The libctx MUST NOT be used after it is freed! Why are you freeing it in the thread_func1() if you're using it in a simultaneous...
Hmm, could you somehow avoid freeing the context in the first thread and do it in the main one?
> @t8m I seem to remember there was some problem with atexit()? I thought some option went in to disable the atexit (which leaks but at least doesnt crash). Would...
> There are many components involved in our product. So its not always possible to control which thread would create and free the libctx. Any one of the thread can...
> @slontis @t8m Is there any option to disable the init_thread_destructor() calls for child threads ? Wouldn't that lead to leaks?
IMO in case of a connection with zero streams you just would not be able to SSL_write/SSL_read on such _connection_ SSL object.
If this cherry-picks cleanly, it should be merged to 3.0 as well.