john
john copied to clipboard
openwall
Formats failing on the new Intel OpenCL CPU intel-oneapi (2024.0)
FTR:
- I'm documenting here formats that fail on the new Intel OpenCL CPU driver/runtime.
- BTW: they were working on previous driver version 18.1.0.013 (for Ubuntu 16).
"$JTR_BIN" -test-full=0 --format=opencl
2023-11-26 16:39:13 (13.4 MB/s) - 'run_tests.sh' saved [14372/14372]
---------------------------- TESTING -----------------------------
Binary: ../run/john
=> ../run/john --list=build-info
Version: 1.9.0-jumbo-1+bleeding-09f03ea13e 2023-11-25 10:47:56 -0300
[...]
Testing: krb5tgs-opencl, Kerberos 5 TGS-REP etype 23 [MD4 HMAC-MD5 RC4 OpenCL]... Build log: Compilation started
Compilation done
Linking started
Linking done
Device build started
Options used by backend compiler: -I opencl -cl-mad-enable -DDEVICE_INFO=36 -D__SIZEOF_HOST_SIZE_T__=8 -DDEV_VER_MAJOR=2023 -DDEV_VER_MINOR=16 -D_OPENCL_COMPILER -DPLAINTEXT_LENGTH=27 -DDATA_LEN=1145 -DUCS_2 -DCONST_CACHE_SIZE=131072 -DENC_RAW -DENC_RAW -DLOC_0=-1 -DLOC_1=-1 -DLOC_2=-1 -DLOC_3=-1 -DNUM_INT_KEYS=1 -DIS_STATIC_GPU_MASK=0
Device build done
Kernel "krb5tgs_init" was successfully vectorized (16)
Kernel "krb5tgs_crypt" was successfully vectorized (16)
Done.
FAILED (crypt_all(49) zero return)
[...]
Testing: pfx-opencl, (.pfx, .p12) [PKCS#12 PBE (SHA1/SHA-256/512) OpenCL]... Build log: Compilation started
Compilation done
Linking started
Linking done
Device build started
Options used by backend compiler: -I opencl -cl-mad-enable -DDEVICE_INFO=36 -D__SIZEOF_HOST_SIZE_T__=8 -DDEV_VER_MAJOR=2023 -DDEV_VER_MINOR=16 -D_OPENCL_COMPILER -DPLAINTEXT_LENGTH=48 -DMAX_DATA_LENGTH=8192 -DENC_RAW -DENC_RAW
Device build done
Kernel "pfx" was successfully vectorized (16)
Done.
FAILED (cmp_all(49))
[...]
2 out of 73 tests have FAILED
Is the OpenCL "driver" upgrade the only change? I made relevant changes in July, so if your prior testing was before July, then those changes I made could also be suspect.
It is CI.
[Edited]
It was tested and passed yesterday. See the download date of run_tests.sh and --list=build-info output
Rebased yesterday (it includes NEWS: Add Keplr wallet support):
2023-11-25 16:20:07 (40.5 MB/s) - 'run_tests.sh' saved [14251/14251]
---------------------------- TESTING -----------------------------
Binary: ../run/john
=> ../run/john --list=build-info
Version: 1.9.0-jumbo-1+bleeding-09f03ea13e 2023-11-25 10:47:56 -0300
[...]
Testing: krb5tgs-opencl, Kerberos 5 TGS-REP etype 23 [MD4 HMAC-MD5 RC4 OpenCL]... Build log: Compilation started
Compilation done
Linking started
Linking done
Device build started
Device build done
Kernel <krb5tgs_init> was successfully vectorized (16)
Kernel <krb5tgs_crypt> was not vectorized
Done.
PASS
[...]
Testing: pfx-opencl, (.pfx, .p12) [PKCS#12 PBE (SHA1/SHA-256/512) OpenCL]... Build log: Compilation started
Compilation done
Linking started
Linking done
Device build started
Device build done
Kernel <pfx> was successfully vectorized (16)
Done.
PASS
In other words, it worked yesterday and stopped working today with only the OpenCL "driver" having changed, correct?
In other words, it worked yesterday and stopped working today with only the OpenCL "driver" having changed, correct?
The exact same (last) commit was tested yesterday and today.
Somehow we saw even more/different formats fail in: https://github.com/openwall/john/pull/5382#issuecomment-1838999062
mscash (OpenCL) formats are disabled during my tests. They were problematic somewhere and I keep them disabled (only inside CI).
It's good to have this recorded, but I see no need to track it as an issue.