Crack WPA2 (.hc22000 file) with list not completing

[Joe_Baker]

I have a WPA2 hash file .hc22000 (so mode 22000) but when I try to find the password located in a small list of 5 words it just keeps running but doesn't complete it. I let the command run for an hour before closing it, it kept loading on "Initializing backend runtime for device #1. Please be patient...". I'm using the command:
"hashcat -a 0 -m 22000 hashfile.hc22000 wordlist.txt". Does someone have experience with these .hc22000 files or maybe something wrong with my command?

The hash looks like following:
"WPA*02*<bunch of letters and numbers with a * from time to time>*02"

Text file looks like following:
"
RandomWord
anotherRandomWord
password
notMyPassword
another
"

The command is running when I'm in the folder of hashcat (hashcat-6.2.5) and the files used are located in this folder as well. I get no error codes except  "nvmlDeviceGetFanSpeed(): Not Supported" but this shouldn't be an issue from what I've read.

I'm using a i7-9750h and RTX2060 so you would expect that it wouldn't take that long to get a hash from a 5 word long list (let alone a huge list like rockyou).

P.S. I'm new to hashcat so it's possible I'm missing some obvious steps.

[v71221]

Try to play with -D option.
At first, to show info about detected backend devices, run

Code:

hashcat.exe -I

Then choose your device.
In my case
-D 1  means use CPU, works!
-D 2  means use GPU, doesn't work, Device #2: Not enough allocatable device memory for this attack.

For simplicity, you can enter the hash and password directly into the command line.

Code:

hashcat.exe  -D 1  -a 3  -m 22000  "WPA*01*4d4fe7aac3a2cecab195321ceb99a7d0*fc690c158264*f4747f87f9f4*686173686361742d6573736964***"  "hashcat!"

It takes about 16 minutes in my case and it works. Status: Cracked
This is an example hash you can find here:
https://hashcat.net/wiki/doku.php?id=example_hashes
or just

Code:

hashcat.exe  -m 22000  --example-hashes

By the way, I'm also new to hashcat.
I'm using Windows and a 10-year-old laptop with an Intel Celeron CPU and an Intel GPU.
I was not able to use hashcat on Linux. Every time I got an "illegal hardware instruction" error.

Now the fun part.
pmkid-hash (format .hc22000) from real dump (captured by hcxdumptool) is not cracked. Status: Exhausted
eapol-hash (format .hc22000) from the same real dump is cracked. Status: Cracked

So far I have not been able to crack pmkid.
I tried wordlist attack, brute-force attack, different dumpfiles, however result is the same. Status: Exhausted
I can crack eapol-hash, but something wrong with pmkid-hash. May be the main reason is my weak hardware.
Please answer what status you saw when you ran the commands below on your hardware. Cracked or Exhausted ?

Code:

hashcat.exe  -D 1  -a 3  -m 22000  "WPA*01*f8dc238fb156874627b5ff251b8ab53c*020000000001*020000000020*61703031***"  "12345678"

hashcat.exe  -D 1  -a 3  -m 22000  "WPA*02*6ec572e97e2ede5a6099bf964fa880fd*020000000001*020000000020*61703031*013ebd2420f2dedcfb7ad5cf967c902c5f40031574352a492e809b58b0e74e4a*0103007502010a00000000000000000000f97e365fcdcfcf2ccb91fa35c25c345eaf34b638c15926eb43a1cc78876d7c86000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001630140100000fac040100000fac040100000fac023c00*02"  "12345678"

Explanation of the hc22000 hash line you can find here
https://hashcat.net/wiki/doku.php?id=cracking_wpawpa2

Please read this post as an example of troubleshooting of dictionary attack.
https://hashcat.net/forum/thread-8602.html

[ZerBea]

Now the fun part.
pmkid-hash (format .hc22000) from real dump (captured by hcxdumptool) is not cracked. Status: Exhausted
eapol-hash (format .hc22000) from the same real dump is cracked. Status: Cracked

Indeed funny, but related to 802.11 attack mode and conversion mode:
PMKID retrieved from ACCESS POINT.
EAPOL MESSAGE PAIR retrieved from CLIENT M2.
It the CLIENT is authorized, the PSK should be the same on both. If not, you'll get two different PSKs. The same will happen if the PSK is changed during capturing time.
(BTW: both MACs look very synthetic - which let me assume that you're running a test environment)
By default hcxdumptool/hcxlabtool attack both (AP and CLIENT) and hcxpcapngtool convert everything.
All tools are analysis tools and it is mandatory that you know what you are doing (choosing the attack vector, converting the hash, selecting the desired hash to feed hashcat). Otherwise the result will be completely unexpected.

[v71221]

@ZerBea
Thank you for your prompt reply. Yes, I am a newcomer, diligently studying hcxdumptool/hcxtools and using a test environment. Three notebooks with wifi-adapters, 1st with Linux and hcxdumptool/hcxtools, 2nd with Windows as wifi access point, and 3rd with Windows as client. For clarity and readability I changed MACs on AP and CLIENT.

AP is created by these commands on Windows 7

Code:

netsh wlan set hostednetwork mode=allow ssid=ap01 key=12345678 keyUsage=temporary
netsh wlan start hostednetwork

I ran this command to capture AP-CLIENT session.

Code:

$ sudo hcxdumptool  -i wlan0  -o dump.pcapng  --silent  --enable_status=127  -c 1

I used silent "passive" mode because client hung if I ran hcxdumptool in "active" mode.
Could you kindly provide me with "proper" syntax of hcxdumptool options if I'm targeting PMKID only.

By the way, I noticed that
hcxhash2cap with option "--pmkid=" gives an error "reading hash line 1 failed".
hcxhash2cap with option "--pmkid-eapol=" works fine.
Input file in both cases is the same one-line-file pmkid.22000

Code:

$ hcxhash2cap --pmkid=pmkid.22000 -c test.cap
reading hash line 1 failed: WPA*01*f8dc238fb156874627b5ff251b8ab53c*020000000001*020000000020*61703031***

$ hcxhash2cap --pmkid-eapol=pmkid.22000 -c test2.cap
PMKIDs/EAPOL messages written to capfile(s): 1 (0 skipped)

[ZerBea]

--pmkid option is for old 16800 hash lines. It will give an ERROR on hc22000 files.
By latest commit:
https://github.com/ZerBea/hcxtools/commi...342a6f71ad
I added an additional information to --help:

Code:

$ hcxhash2cap --help
hcxhash2cap 6.2.5-2-ge6c830a (C) 2021 ZeroBeat
usage:
hcxhash2cap <options>

options:
-c <file> : output cap file
            if no cap file is selected, output will be written to single cap files
            format: mac_sta.cap (mac_sta.cap_x)
-h        : show this help
-v        : show version

--pmkid-eapol=<file> : input PMKID EAPOL (22000) combi hash file
--pmkid=<file>       : input deprecated PMKID (16800) hash file
--hccapx=<file>      : input deprecated hccapx (2500) hash file
--hccap=<file>       : input ancient hccap (2500) file
--john=<file>        : input John the Ripper WPAPSK hash file
--help               : show this help
--version            : show version

If you use --silent, hcxdumptool will become a simple dump tool like tshark, Wireshark, tcpdump. PMKIDs are not requested and a possible packet loss has to be expected.
To request PMKIDs only:
$ sudo hcxdumptool -i INTERFACE -o dump.pcapng --disable_client_attacks --disable_deauthentication --enable_status=95

For sure, some attack modes are extreme aggressive (as hell). They prevent that a CLIENT is able to connect to a NETWORK or they will let a CLIENT crash completely.

BTW:
I'm interested in a dump file from netsh hostednetwork. Can you please add a pcapng file from:
netsh wlan set hostednetwork mode=allow ssid=ap01 key=12345678 keyUsage=temporary

Usually the PMKID and the MIC should be calculated using the same PMK. It looks like this is not the case on netsh, which could be a bug inside of this tool.

From what I read here:
https://stackoverflow.com/questions/2316...e-authenti
only this types are supported by netsh:
Radio types supported : 802.11n 802.11g 802.11b
By default, PMKID caching is not activated.

[ZerBea]

Great. The dump files are very appreciated.
I'll take a look at them.
Thanks.

[ZerBea]

I have finished the analysis.
The PMKID calculated by netsh is wrong!
Looks like Windows has a problem with PMKIDs (not only on WPA2 Enterprise) since Windows 7:
https://social.technet.microsoft.com/For...rect-pmkid

Calculated PMKID by netsh (in WPA KEY DATA field packet 29 file 1, packet 27 file 2):
f8dc238fb156874627b5ff251b8ab53c

Calculated PMKID by function:
ca5396d611cf330aebefd48ebbfb0e63

Code:

PMKID = HMAC-SHA1-128(PMK, "PMK Name" | MAC_AP | MAC_STA)

Corrected hash line to reproduce that hashcat will not fail:

Code:

WPA*01*ca5396d611cf330aebefd48ebbfb0e63*020000000001*020000000020*61703031***

To answer your questions:
1. It doesn't matter if you capture PMKIDROGUE or PMKID. Both are suitable for PMKID-attacks.
correct
PMKIDROGUE = PMKID requested by hcxdumptool
PMKID = PMKID captured after CLIENT request

2. In my case, pmkid-hash was not cracked (Status: Exhausted), probably due to a bug.
correct, because netsh calculated a wrong PMKID!!!


Now I have to find a way to detect this garbage.

[v71221]

@ZerBea
I think we should start another thread called "PMKID Attack, Best Practices, Miscellaneous".
In the meantime, could you advise something to the author of the current thread (Joe_Baker) based on your experience?

For educational purposes, it is desirable to calculate PMK and PMKID manually.
I found this link http://jorisvr.nl/wpapsk.html
Could you please share your method. Perhaps you have written your own utility.
Such a utility along with the source code would be a great help for newbies like me.

[ZerBea]

"In the meantime, could you advise something to the author of the current thread (Joe_Baker) based on your experience?"
To gain the necessary basic knowledge, hashcat FAQ are very helpful:
https://hashcat.net/wiki/doku.php?id=fre...s#overview
I couldn't explain it better than what is described in this general guide.
BTW:
It makes it very difficult to give an advice, because of missing information about the OS, version of NVIDA driver and version of CUDA SDK.

There is no need to open a new thread, because nearly everything is already explained.
Since Atom persuaded me to publish hcxtools (nearly the same time when hashcat went open source) I started a thread:
https://hashcat.net/forum/thread-6661.html
It describe how to use hcxtools and how to build a WiFi analysis environment.

Another thread followed after we (again thanks to Atom and RealEnder) discovered the PMKID attack:
https://hashcat.net/forum/thread-7717.html

A WPA1/2 basic tutorial is here:
https://hashcat.net/wiki/doku.php?id=cracking_wpawpa2

Inside this threads are several links to get more background information about the functions "behind the scenes".

My advice is to read this basics and to play around with the examples mentioned above and here:
https://hashcat.net/wiki/doku.php?id=example_hashes

My second advice is to learn and understand Linux step by step:
https://wiki.archlinux.org/title/Installation_guide
BTW:
A successful installation of K A L I by graphical installer is far away from learning and understanding Linux.

That include openssl crypto:
https://www.openssl.org/docs/man3.0/man7/crypto.html
because it provide all functions to calculate and verify PMKs and PMKIDs.

"Perhaps you have written your own utility."
To find out how a PMK is calculated, please take a look at the source code of wlangenpmk (CPU based):
https://github.com/ZerBea/hcxkeys

Code:

$ wlangenpmk -e ap01 -p 12345678

essid (networkname)....: ap01
password...............: 12345678
plainmasterkey (SHA1)..: 5577866bc5e9778a3ca3d8730e97f258e2a9ae2afd95bbd63c4f383275c8ba93

or wlangenpmkocl (OpenCL based):

Code:

$ wlangenpmkocl -e ap01 -p 12345678
using: NVIDIA GeForce GTX 1080 Ti

essid (networkname)....: ap01
password...............: 12345678
plainmasterkey (SHA1)..: 5577866bc5e9778a3ca3d8730e97f258e2a9ae2afd95bbd63c4f383275c8ba93

There are similar functions (CPU based) in hcxpcapngtool, hcxhashtool and hcxpmkidtool as well as in hcxdumptool.

[v71221]

@ZerBea
Great! Thanks!
In the meantime, I discovered that the freshly installed Windows 11 Enterprise no longer sends PMKID (in contrast to Windows 7 Enterprise). At least by default. Please see the attachment. If you need dumps, please let me know.

Could you please explain what "2412/1" means in the log of hcxdumptool (v6.2.5).
For example, line like this

Code:

22:09:57 2412/1  0015999e54c4 000bf4ad5332 TEST_AP [ROGUE PROBERESPONSE]

What's the point of specifying [ROGUE PROBERESPONSE] in the log if hcxdumptool works with the --silent option
From my newcomer point of view, it makes more sense to specify [PROBEREQUEST] instead.