[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <518D193E.3000609@mccme.ru>
Date: Fri, 10 May 2013 19:58:54 +0400
From: Alexander Cherepanov <cherepan@...me.ru>
To: john-dev@...ts.openwall.com
Subject: Re: Core: undefined behavior in DES_std.c and MD5_std.c
On 2013-05-08 07:41, Solar Designer wrote:
>> Is it really necessary to have ARCH_WORD signed?
>
> I think there are a few places that require a signed type like this,
> e.g. DES_bs_cmp_all().
Why is that? Because of unary minus? Then it's the other way around:
unary minus can overflow for signed integers but cannot for unsigned.
Ok, I've just tried it:
sed -i 's/long/unsigned long/' x86-64.h
sed -i 's/unsigned ARCH_WORD/ARCH_WORD/' *
make linux-x86-64
and john works fine -- at least selftests pass. No visible performance
gain though.
See also:
http://stackoverflow.com/questions/8026694/c-unary-minus-operator-behavior-with-unsigned-operands
http://embeddedgurus.com/stack-overflow/2009/07/efficient-c-tips-10-use-unsigned-integers/
> BTW, its "~(ARCH_WORD)0" should probably be
> "~(unsigned ARCH_WORD)0",
Or this: "(unsigned ARCH_WORD)-1".
See, for example,
http://stackoverflow.com/questions/809227/is-it-safe-to-use-1-to-set-all-bits-to-true
.
> its right shifts could be turned from signed
> (usually arithmetic) to unsigned (logical) ones and the value only then
> cast to a signed type for the negation.
Actually, C standard doesn't guarantee that's signed shift is either
arithmetic or logical. C11, 6.5.7p5 reads: "The result of E1 >> E2 is E1
right-shifted E2 bit positions. If E1 has an unsigned type or if E1 has
a signed type and a nonnegative value, the value of the result is the
integral part of the quotient of E1 / 2[to the power of]E2 . If E1 has a
signed type and a negative value, the resulting value is
implementation-defined."
This leads to a question about platforms supported by JtR. Is it
supposed to compile on a non-two's complement platform? On a platform
with CHAR_BIT != 8 ? Etc.
http://stackoverflow.com/questions/12276957/are-there-any-non-twos-complement-implementations-of-c
http://stackoverflow.com/questions/6971886/exotic-architectures-the-standard-committee-cares-about
http://stackoverflow.com/questions/2098149/what-platforms-have-something-other-than-8-bit-char
> All of this is just asking for a rewrite. Some years ago I actually
> thought of rewriting JtR from scratch, but I never found time.
I suppose that signed-ness of arithmetic is not the only problem?
--
Alexander 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.
