[erlang-bugs] Weird crypto:start() bug with 717cf073 / OTP-11408

[Sverker Eriksson]

On 12/10/2013 12:51 AM, Geoff Cant wrote:
> Hi all, I just updated my OTP install to today's master and ran into a really weird bug - running 'crypto:start()' in the first ~5s after starting erl would fail with the error:
>
>> Erlang R17A (erts-5.11) [source-236b48c] [64-bit] [smp:4:4] [async-threads:10] [hipe] [kernel-poll:false]
>>
>> Eshell V5.11  (abort with ^G)
>> 1> crypto:start().
>> ** exception error: undefined function crypto:start/0
>> 2>
>> =ERROR REPORT==== 9-Dec-2013::13:55:59 ===
>> Unable to load crypto library. Failed with error:
>> "load, Library load-call unsuccessful."
>> OpenSSL might not be installed on this system.
>>
>> =ERROR REPORT==== 9-Dec-2013::13:55:59 ===
>> The on_load function for module crypto returned {error,
>>                                                  {load,
>>                                                   "Library load-call unsuccessful."}}
>> 2> crypto:start().
>> ok
>> 3> crypto:md5(term_to_binary(1)).
>> <<67,214,80,204,205,119,185,209,192,165,163,106,79,236,31,
>>   196>>
>
> If I did any of: calling crypto:start() a second time, or waiting a few seconds before calling it, or evaluating another shell command before crypto:start() , it would work just fine.
>
> This was particularly annoying with rebar, as rebar wants to start crypto at startup and this would continually fail.
>
> Backing out the commit '717cf073' fixed my problem. I'm running OS X 10.9, 'Darwin Kernel Version 13.0.0: Thu Sep 19 22:22:27 PDT 2013; root:xnu-2422.1.72~6/RELEASE_X86_64' etc.
>
> Cheers,
> --
> Geoff Cant
>
>
>
I think this is fixed by 94576fd805c9a8 which has been merged to master 
now accessible at github.

/Sverker, Erlang/OTP