troubles on Radeon RX Vega 8 in AMD Ryzen 3 2200G

[beroal]

My device:

Code:

$ hashcat -I
hashcat (v6.1.1) starting...

OpenCL Info:
============

OpenCL Platform ID #1
  Vendor..: Advanced Micro Devices, Inc.
  Name....: AMD Accelerated Parallel Processing
  Version.: OpenCL 2.0 AMD-APP.dbg (3212.0)

  Backend Device ID #1
    Type...........: GPU
    Vendor.ID......: 1
    Vendor.........: Advanced Micro Devices, Inc.
    Name...........: gfx902+xnack
    Version........: OpenCL 2.0
    Processor(s)...: 11
    Clock..........: 1100
    Memory.Total...: 7882 MB (limited to 6700 MB allocatable in one block)
    Memory.Free....: 7808 MB
    OpenCL.Version.: OpenCL C 2.0
    Driver.Version.: 3212.0 (HSA1.1,LC)

OpenCL Platform ID #2
  Vendor..: Advanced Micro Devices, Inc.
  Name....: AMD Accelerated Parallel Processing
  Version.: OpenCL 2.0 AMD-APP (1800.8)

  Backend Device ID #2
    Type...........: CPU
    Vendor.ID......: 1
    Vendor.........: AuthenticAMD
    Name...........: AMD Ryzen 3 2200G with Radeon Vega Graphics
    Version........: OpenCL 1.2 AMD-APP (1800.8)
    Processor(s)...: 4
    Clock..........: 3504
    Memory.Total...: 15765 MB (limited to 3941 MB allocatable in one block)
    Memory.Free....: 15701 MB
    OpenCL.Version.: OpenCL C 1.2
    Driver.Version.: 1800.8 (sse2,avx)

When I run a job the first time, it completes, albeit taking too much time for 10 passwords and the mouse pointer on the screen connected to RX Vega freezes sometimes:

Code:

$ hashcat -w 1 -d 1 -a 0 -m 14600 luks_header darkweb2017-top10.txt
hashcat (v6.1.1) starting...

/sys/bus/pci/devices/0000:38:00.0/hwmon/hwmon0/pwm1: No such file or directory

OpenCL API (OpenCL 2.0 AMD-APP.dbg (3212.0)) - Platform #1 [Advanced Micro Devices, Inc.]
=========================================================================================
* Device #1: gfx902+xnack, 7808/7882 MB (6700 MB allocatable), 11MCU

OpenCL API (OpenCL 2.0 AMD-APP (1800.8)) - Platform #2 [Advanced Micro Devices, Inc.]
=====================================================================================
* Device #2: AMD Ryzen 3 2200G with Radeon Vega Graphics, skipped

Minimum password length supported by kernel: 0
Maximum password length supported by kernel: 256

Hashes: 1 digests; 1 unique digests, 1 unique salts
Bitmaps: 16 bits, 65536 entries, 0x0000ffff mask, 262144 bytes, 5/13 rotates
Rules: 1

Applicable optimizers applied:
* Zero-Byte
* Single-Hash
* Single-Salt
* Slow-Hash-SIMD-LOOP

Watchdog: Temperature abort trigger set to 90c

Initializing backend runtime for device #1...Signal 0x7fd947dc0b80 time stamps may be invalid.Signal 0x7fd947dc0b00 time stampHost memory required for this attack: 257 MB

Dictionary cache built:
* Filename..: darkweb2017-top10.txt
* Passwords.: 10
* Bytes.....: 81
* Keyspace..: 10
* Runtime...: 0 secs

The wordlist or mask that you are using is too small.
This means that hashcat cannot use the full parallel power of your device(s).
Unless you supply more work, your cracking speed will drop.
For tips on supplying more work, see: https://hashcat.net/faq/morework

Approaching final keyspace - workload adjusted. 

Cracking performance lower than expected?       

* Append -w 3 to the commandline.
  This can cause your screen to lag.

* Update your backend API runtime / driver the right way:
  https://hashcat.net/faq/wrongdriver

* Create more work items to make use of your parallelization power:
  https://hashcat.net/faq/morework

[s]tatus [p]ause [b]ypass [c]heckpoint [q]uit => b

Next dictionary / mask in queue selected. Bypassing current one.

Session..........: hashcat                     
Status...........: Bypass
Hash.Name........: LUKS
Hash.Target......: luks_header
Time.Started.....: Mon Jan 25 12:59:25 2021 (10 secs)
Time.Estimated...: Mon Jan 25 12:59:35 2021 (0 secs)
Guess.Base.......: File (darkweb2017-top10.txt)
Guess.Queue......: 1/1 (100.00%)
Speed.#1.........:        0 H/s (0.64ms) @ Accel:2 Loops:8 Thr:64 Vec:1
Recovered........: 0/1 (0.00%) Digests
Progress.........: 0/10 (0.00%)
Rejected.........: 0/0 (0.00%)
Restore.Point....: 0/10 (0.00%)
Restore.Sub.#1...: Salt:0 Amplifier:0-1 Iteration:355480-355488
Candidates.#1....: 123456 -> 123123
Hardware.Mon.#1..: Temp: 44c Core:1100MHz Mem:1600MHz Bus:0

Started: Mon Jan 25 12:57:03 2021
Stopped: Mon Jan 25 12:59:40 2021

When I run the job the second time, `hashcat` flashes and turns off the screen and terminates with a segmentation fault:

Code:

$ hashcat -w 1 -d 1 -a 0 -m 14600 luks_header darkweb2017-top10.txt
hashcat (v6.1.1) starting...

/sys/bus/pci/devices/0000:38:00.0/hwmon/hwmon0/pwm1: No such file or directory

OpenCL API (OpenCL 2.0 AMD-APP.dbg (3212.0)) - Platform #1 [Advanced Micro Devices, Inc.]
=========================================================================================
* Device #1: gfx902+xnack, 7808/7882 MB (6700 MB allocatable), 11MCU

OpenCL API (OpenCL 2.0 AMD-APP (1800.8)) - Platform #2 [Advanced Micro Devices, Inc.]
=====================================================================================
* Device #2: AMD Ryzen 3 2200G with Radeon Vega Graphics, skipped

Minimum password length supported by kernel: 0
Maximum password length supported by kernel: 256

Hashes: 1 digests; 1 unique digests, 1 unique salts
Bitmaps: 16 bits, 65536 entries, 0x0000ffff mask, 262144 bytes, 5/13 rotates
Rules: 1

Applicable optimizers applied:
* Zero-Byte
* Single-Hash
* Single-Salt
* Slow-Hash-SIMD-LOOP

Watchdog: Temperature abort trigger set to 90c

Host memory required for this attack: 257 MB

Dictionary cache hit:
* Filename..: darkweb2017-top10.txt
* Passwords.: 10
* Bytes.....: 81
* Keyspace..: 10

The wordlist or mask that you are using is too small.
This means that hashcat cannot use the full parallel power of your device(s).
Unless you supply more work, your cracking speed will drop.
For tips on supplying more work, see: https://hashcat.net/faq/morework

Approaching final keyspace - workload adjusted. 

[s]tatus [p]ause [b]ypass [c]heckpoint [q]uit => b

Next dictionary / mask in queue selected. Bypassing current one.

Session..........: hashcat                     
Status...........: Bypass
Hash.Name........: LUKS
Hash.Target......: luks_header
Time.Started.....: Mon Jan 25 13:01:25 2021 (9 secs)
Time.Estimated...: Mon Jan 25 13:01:34 2021 (0 secs)
Guess.Base.......: File (darkweb2017-top10.txt)
Guess.Queue......: 1/1 (100.00%)
Speed.#1.........:        0 H/s (0.11ms) @ Accel:48 Loops:4 Thr:64 Vec:1
Recovered........: 0/1 (0.00%) Digests
Progress.........: 0/10 (0.00%)
Rejected.........: 0/0 (0.00%)
Restore.Point....: 0/10 (0.00%)
Restore.Sub.#1...: Salt:0 Amplifier:0-1 Iteration:234952-234956
Candidates.#1....: 123456 -> 123123
Hardware.Mon.#1..: Temp: 46c Core:1100MHz Mem:1600MHz Bus:0

hashcat: /var/abs/local/hsa-rocr/src/ROCR-Runtime-rocm-4.0.0/src/core/runtime/amd_aql_queue.cpp:721: virtual hsa_status_t rocr::AMD::AqlQueue::Inactivate(): Assertion `err == HSAKMT_STATUS_SUCCESS && "hsaKmtDestroyQueue failed."' failed.
Перервано (збережено знімок оперативної пам’яті)

“Перервано (збережено знімок оперативної пам’яті)” means “Segmentation fault”.

When I run the job the third time, `hashcat` just freezes:

Code:

$ env LANG=en_US.utf8 hashcat -w 1 -d 1 -a 0 -m 14600 luks_header darkweb2017-top10.txt

hashcat (v6.1.1) starting...

I run these commands from another seat connected to another videocard. This seat works as usual, no freezes. I already used `-w 1`, so the freezes are not caused by GPU overload. I do not know what else to do.

There is a similar thread.

I use the following “RadeonOpenCompute (ROCm)”.

Code:

$ pacman -Q hashcat
hashcat 1:6.1.1-1
$ pacman -Q rocm-opencl-runtime
rocm-opencl-runtime 4.0.0-1

[beroal]

I did not manage to replicate the ROCm “assertion failed” error. Freezes are easily replicated. `hashcat` freezes somewhere in the POSIX thread libraries:

Code:

^C
Thread 1 "hashcat" received signal SIGINT, Interrupt.
0x00007ffff7ddf248 in do_futex_wait.constprop () from /usr/lib/libpthread.so.0
(gdb) bt
#0  0x00007ffff7ddf248 in do_futex_wait.constprop () from /usr/lib/libpthread.so.0
#1  0x00007ffff7ddf373 in __new_sem_wait_slow.constprop.0 () from /usr/lib/libpthread.so.0
#2  0x00007fffe701f04c in ?? () from /opt/rocm/lib/libamdocl64.so
#3  0x00007fffe701ec4e in ?? () from /opt/rocm/lib/libamdocl64.so
#4  0x00007fffe702ccc1 in ?? () from /opt/rocm/lib/libamdocl64.so
#5  0x00007fffe703d649 in ?? () from /opt/rocm/lib/libamdocl64.so
#6  0x00007fffe700f184 in clFinish () from /opt/rocm/lib/libamdocl64.so
#7  0x00007fffeae6cfa5 in clFinish () from /opt/AMDAPP/SDK/lib/libOpenCL.so
#8  0x00007ffff7e143d5 in hc_clFinish () from /usr/lib/libhashcat.so.6.1.1
#9  0x00007ffff7e164c3 in run_opencl_kernel_memset () from /usr/lib/libhashcat.so.6.1.1
#10 0x00007ffff7e204e7 in backend_session_begin () from /usr/lib/libhashcat.so.6.1.1
#11 0x00007ffff7e31e65 in ?? () from /usr/lib/libhashcat.so.6.1.1
#12 0x00007ffff7e32622 in hashcat_session_execute () from /usr/lib/libhashcat.so.6.1.1
#13 0x00005555555561e0 in ?? ()
#14 0x00007ffff7c2c152 in __libc_start_main () from /usr/lib/libc.so.6
#15 0x000055555555622e in ?? ()

[undeath]

Crashes and freezes seen when using hashcat are pretty much always hardware errors. There's a slight chance it's a bug in the OpenCL runtime. If possible you can test on Windows, where a completely different OpenCL runtime is used.