bug in oclhashcat 1.36???

[tibit]

When using a combination attack with with left($x) or right(^x) rules in version 1.36 it applies the rule at the start or end of the combination, not in between the two lists.
The same setup works ok in version 1.31 (don't know about other versions)

ex:
oclHashcat64.exe -a 1 -m 0 --session=all --force -o "G:\Outfiles\list1_found.txt" --outfile-format=2 --remove -w 2 --gpu-temp-abort=80 -j "$X" "G:\Hashes\md5\list1" "G:\PW\abc.txt" "G:\PW\def.txt"

v.1.36 results in format "abcdefX"
v.1.31 results in format "abcXdef" --> this is what I want

any idea??
Thank you

[undeath]

You are appending "X" to the right-hand word. Maybe use -k?

[tibit]

Thanks for replying,

As far as I know '-j' applies to the left wordlist and '-k' to the right. (so I think the code is correct)
I did try to prepend a character to the right wordlist, but in that case the character was found at the beginning of the combination, not between the two wordlists.

I'm running oclhashcat for AMD, don't know if the same happens for nvidia

The same code works good in v.1.31

[undeath]

oh crap, my alphabet has a different order than everyone elses'. Testing.

works for me on cudahashcat 1.37 beta

Code:

./cudaHashcat64.bin -m0 -a1 $(echo -n axa | md5sum | cut -c-32) /tmp/a /tmp/a -j '$x'
[snip]
100ee4be8175bd62d4f954323cff4325:axa

[tibit]

Thank you for testing,
suppose I'll have to wait for 1.37 then.

tibit

[tibit]

Hi all,

Tested cudahashcat 1.36 also on different computer and notebook.

The test with a single hash and one word in left and right wordlist turned out ok,

Code:

oclHashcat64.exe -a 1 -m 0 --session=test -w 1 -j "$." -k "$@$m$a$i$l$.$c$o$m" "testhash.txt" "1.txt" "2.txt"

BUT! when running 2wordlist of 50m in the same combination ... I get no results anymore in version 1.36!! (in version 1.31 it's ok)

So, there is definitly something wrong with both oclhashcat and cudahashcat in the combinationmode.

Could someone please try this out in a windows system with v1.36. with a big hashlist and two 50mb wordlists??

The fact that v1.31 has no problems at all, makes me suspecting an error in version 1.36

Thank you

tibit

[atom]

please prepare a package with all files needed so we devs can reproduce it.

[tibit]

Wow,

Thank you so much. I'm uploading the package right now containing the hashfile and the two wordlists. Also added my test.bat files and the resulting pot-files (fraction of results).

Trying out some more in version 1.36, I found out that somehow the "j" and "k" rule are reversed, using big lists. So instead of -j applying to the left list, it then applies to the right list. The -k now applies to the left list.

Switching both in my test (calles reverse.bat), I got the results, I was looking for.

When using v1.31 it's the opposite as found on the wiki site ("j" -> left list) ("k" -> right list).

Thank you for taking a look at my issue,

tibit

[atom]

thanks! bug will be fixed with v1.37

[tibit]

Great news,

Thank you!!