Bug with incremental and --restore?
#1
Question 
Hello everyone,
I noticed everytime i stop ( Q ) and restore ( R ) my cudaHashcat, the new restored process start from another point, basically it skip x% of the entire process.

So i tryed a new session called X with the incremental attack (26 hours), started to stop and restore this session and the process finished in a snap
Now I wonder is a known bug or is it normal to be so? and I also wonder if there is a way to restore precisely from the point of the stop (a flag or something similar).

Thanks everyone and sorry for my english
#2
Could you please give more details about the command you are trying to run (without posting hashes of course)?
#3
(04-03-2015, 03:49 PM)philsmd Wrote: Could you please give more details about the command you are trying to run (without posting hashes of course)?

Oh Sorry, is a normal Brute-force incremental from 1 to 7 or from 1 to 8 of des unix password

EDIT
Anyway here is an example:

cudaHashcat64.exe -a 3 -m 1500 --session=x --force -o "out.txt" --outfile-format=3 -w 2 --gpu-temp-abort=85 --gpu-temp-retain=80 -i --increment-min=1 --increment-max=7 "desunix.txt" -1 ?l?d ?1?1?1?1?1?1?1
#4
Most important question:
why do you use --force... the warning clearly says, if you use --force, you shouldn't report problems and/or ask for support

Is there any reason you use --force? Does it work without --force?

Please give also more details about your setup:
- operating system version
- how many gpus and which driver model(s)
- which driver version do you use
etc
#5
(04-03-2015, 05:27 PM)philsmd Wrote: Most important question:
why do you use --force... the warning clearly says, if you use --force, you shouldn't report problems and/or ask for support

Is there any reason you use --force? Does it work without --force?

Please give also more details about your setup:
- operating system version
- how many gpus and which driver model(s)
- which driver version do you use
etc

No there is no reason. I honestly don't know what --force do and I had never noticed the warning (--force is automatically generated from the hashcat GUI, after i just copy/paste the code on the console or I just run it). I'll remove --force on the next attempts.

Anyway the setup from the device where i noticed this problem and that i'm using right now is:
- Wind8
- gpu : geforce 740m
- 347.88
(yes, i'm on the notebook now)

I can't test on other PCs right now since they running a bigger brutalforce.
Can be a problem on the session_name.restore or session_name.log ?
#6
If you use this command to start the attack:
Code:
cudaHashcat64.exe -m 1500 -a 3 --session x -o out.txt --outfile-format 3 -w 2 -i --increment-min 1 --increment-max 7 desunix.txt -1 ?l?d ?1?1?1?1?1?1?1

You need to restore it, as such:
Code:
cudaHashcat64.exe --session x --restore
#7
(04-03-2015, 06:26 PM)philsmd Wrote: You need to restore it, as such:
Code:
cudaHashcat64.exe --session x --restore

Yes i know, is not this the problem. I said that the --restore just skip a % of the process.
I give you some image example so you can understand.
In the image below i just started a random brute force and started to End-restore the process around 10 times (Q-restore-Q-restore-Q-restore.....)
It took about 20-30 seconds to do 10 times this thing and as you can see from the image, --restore skipped some things.

If is not clear:
- left image=start of the process
- right image= after 30 seconds (each Q-R skipped 2% in this case, around 1 minute each Q-R)

Recap: 10min = 30 seconds with Q-R

This is an example , but in the case of this morning , multiple Q-R skipped several more things , about 10% each time.
it is as if hashcat resume the process from a wrong point (a point not yet reached , A future or next char maybe ).

[Image: ZwbnY9s.png]