[openssl-dev] [openssl-commits] [openssl] master update

[Salz, Rich]

TL;DR -- Don't read the diff, look at the revised code.

> The comment and code here are a slight mismatch, since qsize is dynamically
> computed (but limited to three values, the largest of which is used to size the
> local variable).  It's not clear that using SHA256_DIGEST_LENGTH for the
> check would actually be better, though.

If you can think of a more-c lear comment, let me know.  But checking against qsize is the right thing to do.

> >+            int seed_is_random = seed_in == NULL;
> 
> This part seems really bogus; seed_is_random is an int, but seed_in is const
> unsigned char *; the assignment makes no sense.

No, it's like "seed_in == NULL ? 1 : 0"

> I guess the intent is to declare the variable in the outer loop?

Nope.  

> and seed_in is never read after this point.

It was set up before the loop. 

> The grammar here is slightly unusual; "for lengths of at least 2048 bits"
> or "for lengths 2048 bits and larger" would feel more natural to me.

Open a ticket :)