[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <85a7h1ixpe.fsf@gmail.com>
Date: Sun, 07 Apr 2019 19:31:25 +0300
From: Aleksey Cherepanov <aleksey.4erepanov@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: ldr_split_line() performance regression
Solar Designer <solar@...nwall.com> writes:
> On Wed, Apr 03, 2019 at 11:47:35PM +0300, Aleksey Cherepanov wrote:
>> Solar Designer <solar@...nwall.com> writes:
>> > On Wed, Sep 16, 2015 at 02:02:34PM -0500, jfoug wrote:
>> >>
>> >> On 9/16/2015 1:52 PM, Solar Designer wrote:
>> >> >strlen(*ciphertext) < 10 && strncmp(*ciphertext, "$dummy$", 7)) {
>> >>
>> >> These should be reversed, since strncmp should short circuit out much
>> >> earlier than the length check.
>
> I now think Jim was wrong about that: we're checking for _not_ matching
> "$dummy$", and that condition will usually be true, so with these checks
> reversed the strlen() will be reached almost all of the time anyway.
(*ciphertext)[0] != '$' check just passes everything with $ to full
handling of lines, so $dummy$ would be handled fully. So the check might
be suboptimal and against the intention, yet correct globally.
Further there is the following code:
if (((*ciphertext)[0] != '$' ||
(strncmp(*ciphertext, "$dummy$", 7) &&
strncmp(*ciphertext, "$0$", 3))) &&
p - *ciphertext != 10 /* not tripcode */) {
The code seems to do a similar check. Comparing shows that the NIS check
(written as intended) would skip short $0$cleartext hashes (that's for
--format=plaintext).
Also ch3root pointed out to me that gcc inlines strncmp. But clang still
calls function.
Thanks!
--
Regards,
Aleksey Cherepanov
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
