Subject: Re: DIAGASSERT in cryptographic functions.
To: Luke Mewburn <firstname.lastname@example.org>
From: Bill Sommerfeld <email@example.com>
Date: 09/17/1999 08:40:54
Depending on the routine, on[e] of the following should be done:
* replace the return with abort().
The hash/md functions should probably have this done.
* remove the /* XXXLUKEM */ comment; it's ok to return in this
case. I effectively did this for functions which do stuff
like free a list; it really doesn't matter if the top-level
pointer is checked against NULL because the invoker doesn't
I think that depending on context, any of the following may also be
* For functions which take pointer/length pairs, no longer flag
passing a NULL pointer with a length of zero as an error.
* Remove the #ifdef _DIAGNOSTIC code entirely, since the hardware
will do adequate NULL pointer checks if the pointer is referenced.