Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.os.linux.advocacy > #350990 > unrolled thread

LOL underestimated

Started byowl <owl@rooftop.invalid>
First post2016-04-20 15:10 +0000
Last post2016-04-21 16:26 +0000
Articles 20 on this page of 36 — 6 participants

Back to article view | Back to comp.os.linux.advocacy


Contents

  LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 15:10 +0000
    Re: LOL underestimated vallor <vallor@cultnix.org> - 2016-04-20 16:55 +0000
      Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 17:25 +0000
        Re: LOL underestimated vallor <vallor@cultnix.org> - 2016-04-20 18:45 +0000
          Re: LOL underestimated vallor <vallor@cultnix.org> - 2016-04-20 18:47 +0000
            Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 21:05 +0000
              Re: LOL underestimated vallor <vallor@cultnix.org> - 2016-04-25 19:07 +0000
                Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-25 21:16 +0000
          Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 20:51 +0200
    Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 19:19 +0200
      Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 17:34 +0000
        Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 20:31 +0200
          Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 21:13 +0000
            Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 23:20 +0200
              Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 21:29 +0000
                Re: LOL underestimated Me Sham <osirus47@yahoo.com> - 2016-04-20 14:33 -0700
                  Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 23:58 +0200
                  Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 22:04 +0000
            Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 23:25 +0200
              Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 21:41 +0000
                Re: LOL underestimated Peter Köhlmann <peter-koehlmann@t-online.de> - 2016-04-20 23:50 +0200
                  Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 22:06 +0000
                    Re: LOL underestimated Peter Köhlmann <peter-koehlmann@t-online.de> - 2016-04-21 08:54 +0200
                      Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-21 17:12 +0000
                Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 23:50 +0200
                  Re: LOL underestimated Peter Köhlmann <peter-koehlmann@t-online.de> - 2016-04-20 23:51 +0200
                    Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-20 23:59 +0200
                      Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 22:10 +0000
                        Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-21 00:11 +0200
                          Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-20 22:19 +0000
                            Re: LOL underestimated Melzzzzz <mel@zzzzz.com> - 2016-04-21 00:57 +0200
                              Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-21 00:07 +0000
        Re: LOL underestimated DFS <nospam@dfs.com> - 2016-04-20 23:17 -0400
          Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-21 03:49 +0000
            Re: LOL underestimated DFS <nospam@dfs.com> - 2016-04-21 10:27 -0400
              Re: LOL underestimated owl <owl@rooftop.invalid> - 2016-04-21 16:26 +0000

Page 1 of 2  [1] 2  Next page →


#350990 — LOL underestimated

Fromowl <owl@rooftop.invalid>
Date2016-04-20 15:10 +0000
SubjectLOL underestimated
Message-ID<fjgu003af.p4e@rooftop.invalid>
Underestimated the time it would take.  Probably about 90 days at this
rate.  LOL.  Restarted the 200GB file write after starting a series of
attacks on the problem space for a particular match.

https://vid.me/9qwp

They've been running all night and only about half way to flipping the
first digit on the left part.  It needs to flip 32 times before it's done.
This is when you appreciate Linux uptime and X11 virtual desktops.

[toc] | [next] | [standalone]


#351014

Fromvallor <vallor@cultnix.org>
Date2016-04-20 16:55 +0000
Message-ID<dnpqjfFf3j2U3@mid.individual.net>
In reply to#350990
On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:

> Underestimated the time it would take.  Probably about 90 days at this
> rate.  LOL.  Restarted the 200GB file write after starting a series of
> attacks on the problem space for a particular match.
> 
> https://vid.me/9qwp
> 
> They've been running all night and only about half way to flipping the
> first digit on the left part.  It needs to flip 32 times before it's
> done.
> This is when you appreciate Linux uptime and X11 virtual desktops.

Got GPU?

   https://github.com/xpn/CUDA-MD5-Crack

-- 
 -v
"Desktops, workstations and servers are and Microsoft
is doing very well. AS well as Linux." -"Slimer"

[toc] | [prev] | [next] | [standalone]


#351021

Fromowl <owl@rooftop.invalid>
Date2016-04-20 17:25 +0000
Message-ID<ghjdk90aafe.kgi@rooftop.invalid>
In reply to#351014
vallor <vallor@cultnix.org> wrote:
> On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:
> 
>> Underestimated the time it would take.  Probably about 90 days at this
>> rate.  LOL.  Restarted the 200GB file write after starting a series of
>> attacks on the problem space for a particular match.
>> 
>> https://vid.me/9qwp
>> 
>> They've been running all night and only about half way to flipping the
>> first digit on the left part.  It needs to flip 32 times before it's
>> done.
>> This is when you appreciate Linux uptime and X11 virtual desktops.
> 
> Got GPU?
> 
>    https://github.com/xpn/CUDA-MD5-Crack
> 

No NVIDIA here.

[toc] | [prev] | [next] | [standalone]


#351035

Fromvallor <vallor@cultnix.org>
Date2016-04-20 18:45 +0000
Message-ID<dnq12vFf3j2U5@mid.individual.net>
In reply to#351021
On Wed, 20 Apr 2016 17:25:14 +0000, owl wrote:

> vallor <vallor@cultnix.org> wrote:
>> On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:
>> 
>>> Underestimated the time it would take.  Probably about 90 days at this
>>> rate.  LOL.  Restarted the 200GB file write after starting a series of
>>> attacks on the problem space for a particular match.
>>> 
>>> https://vid.me/9qwp
>>> 
>>> They've been running all night and only about half way to flipping the
>>> first digit on the left part.  It needs to flip 32 times before it's
>>> done.
>>> This is when you appreciate Linux uptime and X11 virtual desktops.
>> 
>> Got GPU?
>> 
>>    https://github.com/xpn/CUDA-MD5-Crack
>> 
>> 
> No NVIDIA here.

Well, you might save yourself some time by getting one. ;P

Or...

   http://www.nvidia.com/object/gpu-cloud-computing-services.html

-- 
 -v
"Desktops, workstations and servers are and Microsoft
is doing very well. AS well as Linux." -"Slimer"

[toc] | [prev] | [next] | [standalone]


#351037

Fromvallor <vallor@cultnix.org>
Date2016-04-20 18:47 +0000
Message-ID<dnq158Ff3j2U6@mid.individual.net>
In reply to#351035
On Wed, 20 Apr 2016 18:45:51 +0000, vallor wrote:

> On Wed, 20 Apr 2016 17:25:14 +0000, owl wrote:
> 
>> vallor <vallor@cultnix.org> wrote:
>>> On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:
>>> 
>>>> Underestimated the time it would take.  Probably about 90 days at
>>>> this rate.  LOL.  Restarted the 200GB file write after starting a
>>>> series of attacks on the problem space for a particular match.
>>>> 
>>>> https://vid.me/9qwp
>>>> 
>>>> They've been running all night and only about half way to flipping
>>>> the first digit on the left part.  It needs to flip 32 times before
>>>> it's done.
>>>> This is when you appreciate Linux uptime and X11 virtual desktops.
>>> 
>>> Got GPU?
>>> 
>>>    https://github.com/xpn/CUDA-MD5-Crack
>>> 
>>> 
>> No NVIDIA here.
> 
> Well, you might save yourself some time by getting one. ;P
> 
> Or...
> 
>    http://www.nvidia.com/object/gpu-cloud-computing-services.html

Well, assuming you don't have some other sort of GPU -- there's always 
OpenCL for ATI, I do believe.

-- 
 -v
"Desktops, workstations and servers are and Microsoft
is doing very well. AS well as Linux." -"Slimer"

[toc] | [prev] | [next] | [standalone]


#351053

Fromowl <owl@rooftop.invalid>
Date2016-04-20 21:05 +0000
Message-ID<hgji9a03f.a@rooftop.invalid>
In reply to#351037
vallor <vallor@cultnix.org> wrote:
> On Wed, 20 Apr 2016 18:45:51 +0000, vallor wrote:
> 
>> On Wed, 20 Apr 2016 17:25:14 +0000, owl wrote:
>> 
>>> vallor <vallor@cultnix.org> wrote:
>>>> On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:
>>>> 
>>>>> Underestimated the time it would take.  Probably about 90 days at
>>>>> this rate.  LOL.  Restarted the 200GB file write after starting a
>>>>> series of attacks on the problem space for a particular match.
>>>>> 
>>>>> https://vid.me/9qwp
>>>>> 
>>>>> They've been running all night and only about half way to flipping
>>>>> the first digit on the left part.  It needs to flip 32 times before
>>>>> it's done.
>>>>> This is when you appreciate Linux uptime and X11 virtual desktops.
>>>> 
>>>> Got GPU?
>>>> 
>>>>    https://github.com/xpn/CUDA-MD5-Crack
>>>> 
>>>> 
>>> No NVIDIA here.
>> 
>> Well, you might save yourself some time by getting one. ;P
>> 
>> Or...
>> 
>>    http://www.nvidia.com/object/gpu-cloud-computing-services.html
> 
> Well, assuming you don't have some other sort of GPU -- there's always 
> OpenCL for ATI, I do believe.
> 

That sounds cool, but I don't really want to spend any money. 

[toc] | [prev] | [next] | [standalone]


#351803

Fromvallor <vallor@cultnix.org>
Date2016-04-25 19:07 +0000
Message-ID<do7888FsmspU1@mid.individual.net>
In reply to#351053
On Wed, 20 Apr 2016 21:05:50 +0000, owl wrote:

> vallor <vallor@cultnix.org> wrote:
>> On Wed, 20 Apr 2016 18:45:51 +0000, vallor wrote:
>> 
>>> On Wed, 20 Apr 2016 17:25:14 +0000, owl wrote:
>>> 
>>>> vallor <vallor@cultnix.org> wrote:
>>>>> On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:
>>>>> 
>>>>>> Underestimated the time it would take.  Probably about 90 days at
>>>>>> this rate.  LOL.  Restarted the 200GB file write after starting a
>>>>>> series of attacks on the problem space for a particular match.
>>>>>> 
>>>>>> https://vid.me/9qwp
>>>>>> 
>>>>>> They've been running all night and only about half way to flipping
>>>>>> the first digit on the left part.  It needs to flip 32 times before
>>>>>> it's done.
>>>>>> This is when you appreciate Linux uptime and X11 virtual desktops.
>>>>> 
>>>>> Got GPU?
>>>>> 
>>>>>    https://github.com/xpn/CUDA-MD5-Crack
>>>>> 
>>>>> 
>>>> No NVIDIA here.
>>> 
>>> Well, you might save yourself some time by getting one. ;P
>>> 
>>> Or...
>>> 
>>>    http://www.nvidia.com/object/gpu-cloud-computing-services.html
>> 
>> Well, assuming you don't have some other sort of GPU -- there's always
>> OpenCL for ATI, I do believe.
>> 
>> 
> That sounds cool, but I don't really want to spend any money.

How goes the cracking?

Do you not have a GPU in your computer?  Even the SIMD stuff in Intel HD 
graphics could help, if I'm not mistaken.

-- 
 -v
"Desktops, workstations and servers are and Microsoft
is doing very well. AS well as Linux." -"Slimer"

[toc] | [prev] | [next] | [standalone]


#351826

Fromowl <owl@rooftop.invalid>
Date2016-04-25 21:16 +0000
Message-ID<shjg303.aifr@rooftop.invalid>
In reply to#351803
vallor <vallor@cultnix.org> wrote:
> On Wed, 20 Apr 2016 21:05:50 +0000, owl wrote:
> 
>> vallor <vallor@cultnix.org> wrote:
>>> On Wed, 20 Apr 2016 18:45:51 +0000, vallor wrote:
>>> 
>>>> On Wed, 20 Apr 2016 17:25:14 +0000, owl wrote:
>>>> 
>>>>> vallor <vallor@cultnix.org> wrote:
>>>>>> On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:
>>>>>> 
>>>>>>> Underestimated the time it would take.  Probably about 90 days at
>>>>>>> this rate.  LOL.  Restarted the 200GB file write after starting a
>>>>>>> series of attacks on the problem space for a particular match.
>>>>>>> 
>>>>>>> https://vid.me/9qwp
>>>>>>> 
>>>>>>> They've been running all night and only about half way to flipping
>>>>>>> the first digit on the left part.  It needs to flip 32 times before
>>>>>>> it's done.
>>>>>>> This is when you appreciate Linux uptime and X11 virtual desktops.
>>>>>> 
>>>>>> Got GPU?
>>>>>> 
>>>>>>    https://github.com/xpn/CUDA-MD5-Crack
>>>>>> 
>>>>>> 
>>>>> No NVIDIA here.
>>>> 
>>>> Well, you might save yourself some time by getting one. ;P
>>>> 
>>>> Or...
>>>> 
>>>>    http://www.nvidia.com/object/gpu-cloud-computing-services.html
>>> 
>>> Well, assuming you don't have some other sort of GPU -- there's always
>>> OpenCL for ATI, I do believe.
>>> 
>>> 
>> That sounds cool, but I don't really want to spend any money.
> 
> How goes the cracking?
> 
> Do you not have a GPU in your computer?  Even the SIMD stuff in Intel HD 
> graphics could help, if I'm not mistaken.
> 

Actually, I had turned it off after a couple days.  Your post made me
take a look at it again, and I modified the way I'm running it.  Now
getting 3500-3800 results per second, versus previous 500 per sec.
So it should be doable in a couple weeks vs the original 90 day
estimate.  I'll let it run for a bit and see if running count jibes
with estimate.

[toc] | [prev] | [next] | [standalone]


#351038

FromMelzzzzz <mel@zzzzz.com>
Date2016-04-20 20:51 +0200
Message-ID<20160420205154.123a3808@maxa-pc>
In reply to#351035
On 20 Apr 2016 18:45:51 GMT
vallor <vallor@cultnix.org> wrote:

> On Wed, 20 Apr 2016 17:25:14 +0000, owl wrote:
> 
> > vallor <vallor@cultnix.org> wrote:  
> >> On Wed, 20 Apr 2016 15:10:00 +0000, owl wrote:
> >>   
> >>> Underestimated the time it would take.  Probably about 90 days at
> >>> this rate.  LOL.  Restarted the 200GB file write after starting a
> >>> series of attacks on the problem space for a particular match.
> >>> 
> >>> https://vid.me/9qwp
> >>> 
> >>> They've been running all night and only about half way to
> >>> flipping the first digit on the left part.  It needs to flip 32
> >>> times before it's done.
> >>> This is when you appreciate Linux uptime and X11 virtual
> >>> desktops.  
> >> 
> >> Got GPU?
> >> 
> >>    https://github.com/xpn/CUDA-MD5-Crack
> >> 
> >>   
> > No NVIDIA here.  
> 
> Well, you might save yourself some time by getting one. ;P

Hm AMD was better for bitcoin mining, and that's about computing
hashes...

[toc] | [prev] | [next] | [standalone]


#351020

FromMelzzzzz <mel@zzzzz.com>
Date2016-04-20 19:19 +0200
Message-ID<20160420191915.3cf2c768@maxa-pc>
In reply to#350990
On Wed, 20 Apr 2016 15:10:00 +0000 (UTC)
owl <owl@rooftop.invalid> wrote:

> Underestimated the time it would take.  Probably about 90 days at this
> rate.  LOL.  Restarted the 200GB file write after starting a series of
> attacks on the problem space for a particular match.
> 
> https://vid.me/9qwp
> 
> They've been running all night and only about half way to flipping the
> first digit on the left part.  It needs to flip 32 times before it's
> done. This is when you appreciate Linux uptime and X11 virtual
> desktops.
> 

What are you trying to do?

[toc] | [prev] | [next] | [standalone]


#351022

Fromowl <owl@rooftop.invalid>
Date2016-04-20 17:34 +0000
Message-ID<fhjgda03.af3@rooftop.invalid>
In reply to#351020
Melzzzzz <mel@zzzzz.com> wrote:
> On Wed, 20 Apr 2016 15:10:00 +0000 (UTC)
> owl <owl@rooftop.invalid> wrote:
> 
>> Underestimated the time it would take.  Probably about 90 days at this
>> rate.  LOL.  Restarted the 200GB file write after starting a series of
>> attacks on the problem space for a particular match.
>> 
>> https://vid.me/9qwp
>> 
>> They've been running all night and only about half way to flipping the
>> first digit on the left part.  It needs to flip 32 times before it's
>> done. This is when you appreciate Linux uptime and X11 virtual
>> desktops.
>> 
> 
> What are you trying to do?

At least some eternal-september posts have an Injection-Info
header, part of which consists of a "posting-host" string that
appears to be an md5 hash.  I'm working on the assumption
that it is a direct hash of a dotted quad IP address (could
very well a bogus assumption).

So, I am creating an md5 lookup table for all IPv4 addresses.
That will take several months to create what will be about
200GB file in the format I'm using:

anon@lowtide:~$ head /media/anon/5331134e-f853-4f49-876b-863d26af6261/out
1.1.1.1
a56bdbb2cc577c3eb01b8575c9241d51  -
1.1.1.2
d79e1566ae16cc3ab5060b74df13c9a4  -
1.1.1.3
117786110e867993f7d584b6097f2203  -
1.1.1.4
254e7213914e76fa50811e5fd46e5dc5  -
1.1.1.5
954b258b29028c2c250ecd3dd7b0c5da  -

progress far (since last night):

anon@lowtide:~$ tail /media/anon/5331134e-f853-4f49-876b-863d26af6261/out
1.173.19.22
0986c814c63feea0933437ef5de7a6a8  -
1.173.19.23
fb59ae6da90786e4d0ae256bdf6703b0  -
1.173.19.24
216a58bb2825f4c4cdeaf4e3c6fddce7  -
1.173.19.25
87fe5a2b16d563265e9ec881994533c5  -
1.173.19.26
924e64c48cadbe4133c646f698eca4a2  -
anon@lowtide:~$

I'm simultaneously running an attack on a particular hash over the entire
address space (those eight small windows in the vid).

[toc] | [prev] | [next] | [standalone]


#351029

FromMelzzzzz <mel@zzzzz.com>
Date2016-04-20 20:31 +0200
Message-ID<20160420203110.1b2edf85@maxa-pc>
In reply to#351022
On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
owl <owl@rooftop.invalid> wrote:

> 
> I'm simultaneously running an attack on a particular hash over the
> entire address space (those eight small windows in the vid).
> 

https://eprint.iacr.org/2006/105
Tunnels in Hash Functions: MD5 Collisions Within a Minute

[toc] | [prev] | [next] | [standalone]


#351056

Fromowl <owl@rooftop.invalid>
Date2016-04-20 21:13 +0000
Message-ID<ghdmjbd.lo3@rooftop.invalid>
In reply to#351029
Melzzzzz <mel@zzzzz.com> wrote:
> On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
> owl <owl@rooftop.invalid> wrote:
> 
>> 
>> I'm simultaneously running an attack on a particular hash over the
>> entire address space (those eight small windows in the vid).
>> 
> 
> https://eprint.iacr.org/2006/105
> Tunnels in Hash Functions: MD5 Collisions Within a Minute

Unless I'm misunderstanding things, that's just generating collision
pairs.  Correct me if I'm wrong, but I assume that all attacks against a
hash still require a dictionary of hashes (which is what I'm generating).

[toc] | [prev] | [next] | [standalone]


#351059

FromMelzzzzz <mel@zzzzz.com>
Date2016-04-20 23:20 +0200
Message-ID<20160420232012.7027be06@maxa-pc>
In reply to#351056
On Wed, 20 Apr 2016 21:13:16 +0000 (UTC)
owl <owl@rooftop.invalid> wrote:

> Melzzzzz <mel@zzzzz.com> wrote:
> > On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
> > owl <owl@rooftop.invalid> wrote:
> >   
> >> 
> >> I'm simultaneously running an attack on a particular hash over the
> >> entire address space (those eight small windows in the vid).
> >>   
> > 
> > https://eprint.iacr.org/2006/105
> > Tunnels in Hash Functions: MD5 Collisions Within a Minute  
> 
> Unless I'm misunderstanding things, that's just generating collision
> pairs.  Correct me if I'm wrong, but I assume that all attacks
> against a hash still require a dictionary of hashes (which is what
> I'm generating).
> 

This paper I got earlier, it's about cracking md5 quickly. It should
help your cause ;)

[toc] | [prev] | [next] | [standalone]


#351061

Fromowl <owl@rooftop.invalid>
Date2016-04-20 21:29 +0000
Message-ID<ghjdie003.afp4@rooftop.invalid>
In reply to#351059
Melzzzzz <mel@zzzzz.com> wrote:
> On Wed, 20 Apr 2016 21:13:16 +0000 (UTC)
> owl <owl@rooftop.invalid> wrote:
> 
>> Melzzzzz <mel@zzzzz.com> wrote:
>> > On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
>> > owl <owl@rooftop.invalid> wrote:
>> >   
>> >> 
>> >> I'm simultaneously running an attack on a particular hash over the
>> >> entire address space (those eight small windows in the vid).
>> >>   
>> > 
>> > https://eprint.iacr.org/2006/105
>> > Tunnels in Hash Functions: MD5 Collisions Within a Minute  
>> 
>> Unless I'm misunderstanding things, that's just generating collision
>> pairs.  Correct me if I'm wrong, but I assume that all attacks
>> against a hash still require a dictionary of hashes (which is what
>> I'm generating).
>> 
> 
> This paper I got earlier, it's about cracking md5 quickly. It should
> help your cause ;)
> 

Are you sure?  I skimmed it, and it seems to be proof of concept of
quickly generating md5 collisions, which just means two data which
generate the same hash, which makes md5 unreliable for verifying
data integrity.  But here I need to find the unknown original input
that was hashed, given only the hash.  Is there another method other
than brute force dictionary comparison of hashes to find original
input to an md5 hash?  Does this paper describe such a method?

[toc] | [prev] | [next] | [standalone]


#351062

FromMe Sham <osirus47@yahoo.com>
Date2016-04-20 14:33 -0700
Message-ID<21340349-1215-4478-adb6-4950ad82b948@googlegroups.com>
In reply to#351061
On Wednesday, April 20, 2016 at 4:30:02 PM UTC-5, owl wrote:
> Melzzzzz <mel@zzzzz.com> wrote:
> > On Wed, 20 Apr 2016 21:13:16 +0000 (UTC)
> > owl <owl@rooftop.invalid> wrote:
> > 
> >> Melzzzzz <mel@zzzzz.com> wrote:
> >> > On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
> >> > owl <owl@rooftop.invalid> wrote:
> >> >   
> >> >> 
> >> >> I'm simultaneously running an attack on a particular hash over the
> >> >> entire address space (those eight small windows in the vid).
> >> >>   
> >> > 
> >> > https://eprint.iacr.org/2006/105
> >> > Tunnels in Hash Functions: MD5 Collisions Within a Minute  
> >> 
> >> Unless I'm misunderstanding things, that's just generating collision
> >> pairs.  Correct me if I'm wrong, but I assume that all attacks
> >> against a hash still require a dictionary of hashes (which is what
> >> I'm generating).
> >> 
> > 
> > This paper I got earlier, it's about cracking md5 quickly. It should
> > help your cause ;)
> > 
> 
> Are you sure?  I skimmed it, and it seems to be proof of concept of
> quickly generating md5 collisions, which just means two data which
> generate the same hash, which makes md5 unreliable for verifying
> data integrity.  But here I need to find the unknown original input
> that was hashed, given only the hash.  Is there another method other
> than brute force dictionary comparison of hashes to find original
> input to an md5 hash?  Does this paper describe such a method?

You can't reverse a hash. MD5 output is only 128-bit, but its inputs can be any length, which means for every hash output K there exists an infinite number of inputs P such that hash(P) = K.

[toc] | [prev] | [next] | [standalone]


#351070

FromMelzzzzz <mel@zzzzz.com>
Date2016-04-20 23:58 +0200
Message-ID<20160420235843.6a42f107@maxa-pc>
In reply to#351062
On Wed, 20 Apr 2016 14:33:45 -0700 (PDT)
Me Sham <osirus47@yahoo.com> wrote:

> On Wednesday, April 20, 2016 at 4:30:02 PM UTC-5, owl wrote:
> > Melzzzzz <mel@zzzzz.com> wrote:  
> > > On Wed, 20 Apr 2016 21:13:16 +0000 (UTC)
> > > owl <owl@rooftop.invalid> wrote:
> > >   
> > >> Melzzzzz <mel@zzzzz.com> wrote:  
> > >> > On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
> > >> > owl <owl@rooftop.invalid> wrote:
> > >> >     
> > >> >> 
> > >> >> I'm simultaneously running an attack on a particular hash
> > >> >> over the entire address space (those eight small windows in
> > >> >> the vid). 
> > >> > 
> > >> > https://eprint.iacr.org/2006/105
> > >> > Tunnels in Hash Functions: MD5 Collisions Within a Minute    
> > >> 
> > >> Unless I'm misunderstanding things, that's just generating
> > >> collision pairs.  Correct me if I'm wrong, but I assume that all
> > >> attacks against a hash still require a dictionary of hashes
> > >> (which is what I'm generating).
> > >>   
> > > 
> > > This paper I got earlier, it's about cracking md5 quickly. It
> > > should help your cause ;)
> > >   
> > 
> > Are you sure?  I skimmed it, and it seems to be proof of concept of
> > quickly generating md5 collisions, which just means two data which
> > generate the same hash, which makes md5 unreliable for verifying
> > data integrity.  But here I need to find the unknown original input
> > that was hashed, given only the hash.  Is there another method other
> > than brute force dictionary comparison of hashes to find original
> > input to an md5 hash?  Does this paper describe such a method?  
> 
> You can't reverse a hash. MD5 output is only 128-bit, but its inputs
> can be any length, which means for every hash output K there exists
> an infinite number of inputs P such that hash(P) = K.

While there is infinite number of variable length inputs there is
limited number of inputs of length n.

[toc] | [prev] | [next] | [standalone]


#351073

Fromowl <owl@rooftop.invalid>
Date2016-04-20 22:04 +0000
Message-ID<hgjdi03afa3.441@rooftop.invalid>
In reply to#351062
Me Sham <osirus47@yahoo.com> wrote:
> On Wednesday, April 20, 2016 at 4:30:02 PM UTC-5, owl wrote:
>> Melzzzzz <mel@zzzzz.com> wrote:
>> > On Wed, 20 Apr 2016 21:13:16 +0000 (UTC)
>> > owl <owl@rooftop.invalid> wrote:
>> > 
>> >> Melzzzzz <mel@zzzzz.com> wrote:
>> >> > On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
>> >> > owl <owl@rooftop.invalid> wrote:
>> >> >   
>> >> >> 
>> >> >> I'm simultaneously running an attack on a particular hash over the
>> >> >> entire address space (those eight small windows in the vid).
>> >> >>   
>> >> > 
>> >> > https://eprint.iacr.org/2006/105
>> >> > Tunnels in Hash Functions: MD5 Collisions Within a Minute  
>> >> 
>> >> Unless I'm misunderstanding things, that's just generating collision
>> >> pairs.  Correct me if I'm wrong, but I assume that all attacks
>> >> against a hash still require a dictionary of hashes (which is what
>> >> I'm generating).
>> >> 
>> > 
>> > This paper I got earlier, it's about cracking md5 quickly. It should
>> > help your cause ;)
>> > 
>> 
>> Are you sure?  I skimmed it, and it seems to be proof of concept of
>> quickly generating md5 collisions, which just means two data which
>> generate the same hash, which makes md5 unreliable for verifying
>> data integrity.  But here I need to find the unknown original input
>> that was hashed, given only the hash.  Is there another method other
>> than brute force dictionary comparison of hashes to find original
>> input to an md5 hash?  Does this paper describe such a method?
> 
> You can't reverse a hash. MD5 output is only 128-bit, but its inputs can be any length, which means for every hash output K there exists an infinite number of inputs P such that hash(P) = K.

I know it's one-way.  That's why I'm questioning this collision generator
as being applicable to my problem of having a known hash and wanting to 
find the original data that was hashed.  Here I'm making an educated guess
that the original data is an IP address, and I've decided to generate
a hash table of the IPv4 address space, which even though relatively large,
is doable in a couple of months on my machine.

[toc] | [prev] | [next] | [standalone]


#351060

FromMelzzzzz <mel@zzzzz.com>
Date2016-04-20 23:25 +0200
Message-ID<20160420232546.4804cf6f@maxa-pc>
In reply to#351056
On Wed, 20 Apr 2016 21:13:16 +0000 (UTC)
owl <owl@rooftop.invalid> wrote:

> Melzzzzz <mel@zzzzz.com> wrote:
> > On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
> > owl <owl@rooftop.invalid> wrote:
> >   
> >> 
> >> I'm simultaneously running an attack on a particular hash over the
> >> entire address space (those eight small windows in the vid).
> >>   
> > 
> > https://eprint.iacr.org/2006/105
> > Tunnels in Hash Functions: MD5 Collisions Within a Minute  
> 
> Unless I'm misunderstanding things, that's just generating collision
> pairs.  Correct me if I'm wrong, but I assume that all attacks
> against a hash still require a dictionary of hashes (which is what
> I'm generating).
> 

If you find collision of particular length, you have probable
solution...

[toc] | [prev] | [next] | [standalone]


#351063

Fromowl <owl@rooftop.invalid>
Date2016-04-20 21:41 +0000
Message-ID<ghjmbnjc03.are@rooftop.invalid>
In reply to#351060
Melzzzzz <mel@zzzzz.com> wrote:
> On Wed, 20 Apr 2016 21:13:16 +0000 (UTC)
> owl <owl@rooftop.invalid> wrote:
> 
>> Melzzzzz <mel@zzzzz.com> wrote:
>> > On Wed, 20 Apr 2016 17:34:37 +0000 (UTC)
>> > owl <owl@rooftop.invalid> wrote:
>> >   
>> >> 
>> >> I'm simultaneously running an attack on a particular hash over the
>> >> entire address space (those eight small windows in the vid).
>> >>   
>> > 
>> > https://eprint.iacr.org/2006/105
>> > Tunnels in Hash Functions: MD5 Collisions Within a Minute  
>> 
>> Unless I'm misunderstanding things, that's just generating collision
>> pairs.  Correct me if I'm wrong, but I assume that all attacks
>> against a hash still require a dictionary of hashes (which is what
>> I'm generating).
>> 
> 
> If you find collision of particular length, you have probable
> solution...

Length of what?
Can you give me a layman's executive summary of how the described
approach can reveal the original input given only the known hash? 

[toc] | [prev] | [next] | [standalone]


Page 1 of 2  [1] 2  Next page →

Back to top | Article view | comp.os.linux.advocacy


csiph-web