[PATCH] Unbound alloca in crypt routines

[Carlos O'Donell]

>> What kind of testing did you do here?
>
> So far just verifying that with very long keys that we fall back to malloc
> rather than segfaulting.  And that the we eventually returned from the crypt
> routine.   It's quite painful simply due to the time it takes to run the
> test with very large keys.

Thanks, that sounds good.

>> Do you have a test case we could add to the testsuite that exercises
>> the new malloc case?
>
> I just cobbled the attached code based on the python test I was provided.
>  Nothing terribly special, just create a big key, ensure it's misaligned and
> send it to the various crypt routines.

OK, it's probably not possible to get this into the testsuite because
it would be horribly painful to test.

Thanks.

Cheers,
Carlos.