wrong dict size?

[San]

Hi,

I found a possible bug with oclhashcat v0.25. I wanted to know if anyone else have the same issue...

When using the following syntax:
cudaHashcat32.exe -m 0 -o test.out -n 1 ../32_hex.txt -1 ?d ../dico/rockyou.txt -j $d$u ?1?1?1?10
i can read

...
Mode.Left.: Dict '../dico/rockyou.txt' (13690308)
Mode.Right: Mask '?1?1?1?10' (10000)
...
Running...: 6 secs
Estimated.: 9 mins, 58 secs

but when replacing "-m 0" by "-m 1000" :

...
Mode.Left.: Dict '../dico/rockyou.txt' (279909)
Mode.Right: Mask '?1?1?1?10' (10000)
...
Running...: 1 sec
Estimated.: 8 secs

see full log: pastebin.com/1qyZiDhF

Edit:
same syntax without rules, without "-n 1" and not using "-1":

...
Mode.Left.: Dict '../dico/rockyou.txt' (14100048)

...
Mode.Left.: Dict '../dico/rockyou.txt' (4733978)

[Xanadrel]

Also reproduced it : http://paste2.org/p/1356706

Looks like it comes from the -j switch.

[atom]

nope, its exactly doing what it is supposed to do. remember there is a length limit on each side (14). oclHashcat automatically skips words that are longer than 14 chars. it is -not- depeding on rules. but what you maybe did not take into calculation is that NTLM doubles the length of each word since NTLM plains converted to unicode before hashed. that means, including your rule from above, max length of the words in ../dico/rockyou.txt is 5! here is the proof:

root@sf:~# perl -ne 'chomp; next if length > 5; print $_, "\n"' < dict/rockyou.txt | wc -l
279910