1
0
mirror of https://github.com/hashcat/hashcat.git synced 2024-12-24 07:28:10 +00:00
Commit Graph

2450 Commits

Author SHA1 Message Date
Gabriele Gristina
d711c593de fix to crack real hashes 2021-07-26 23:00:33 +02:00
Gabriele Gristina
d62fa617fb fix to crack real hashes 2021-07-26 22:45:02 +02:00
Gabriele Gristina
c20ff01c39 using shared buffer between md5 and sha1 SNMPV3_TMP_ELEMS_OPT, fix to crack real hashes 2021-07-26 22:25:15 +02:00
Gabriele Gristina
45fce5d3a3 fix snmpv3 md5/sha1, tested with real hashes 2021-07-26 19:04:30 +02:00
Jens Steube
7f419c68af Blowfish Kernels: Backport optimizations reducing bank conflicts from bcrypt to Password Safe v2 and Open Document Format (ODF) 1.1 2021-07-26 13:38:39 +02:00
Jens Steube
f3f6cfadb7 Password Safe v2: Backport optimizations reducing bank conflicts in bcrypt 2021-07-26 10:17:21 +02:00
Jens Steube
b66c093c17
Merge pull request #2902 from hashcat/master
Backports
2021-07-26 08:18:35 +02:00
Jens Steube
1e3bd2c8a0 AMD GPUs: Add inline assembly code for md5crypt/sha256crypt, PDF 1.7, 7-Zip, RAR3, Samsung Android and Windows Phone 8+ 2021-07-26 07:59:12 +02:00
Jens Steube
84a4058edf
Merge pull request #2900 from hashcat/master
Backport changes
2021-07-25 10:34:05 +02:00
Gabriele Gristina
58bb2b69b4 Added hash-mode: SNMPv3 HMAC-SHA256-192 2021-07-24 18:57:25 +02:00
Gabriele Gristina
7683ca19a3 Added hash-mode: SNMPv3 HMAC-SHA224-128 2021-07-24 18:36:10 +02:00
Gabriele Gristina
7c8d7f7831 Added hash-mode: SNMPv3 HMAC-MD5-96/HMAC-SHA1-96 2021-07-24 17:03:50 +02:00
Gabriele Gristina
4f10377703 Added hash-mode: SNMPv3 HMAC-MD5-96 2021-07-24 13:56:49 +02:00
Jens Steube
3a31b669b5 Optimize and clean up -m 25200 code.
Speed on CPU: 1521 -> 2066 (35% faster)
Speed on GPU: 16610 -> 157754 (9.4 times faster)
Works for all password lengths
2021-07-24 11:58:36 +02:00
Jens Steube
3f6c5a0042 Update module_unstable_warning() for -m 172xx on HIP 2021-07-23 21:09:55 +02:00
Jens Steube
5ffcaa980d HIP Backend: Added support to support HIP 4.4 and later, but added check to rule out older versions because they are incompatible 2021-07-23 16:04:34 +02:00
Gabriele Gristina
6d83a69f7d using hex_encode/hex_decode + small fixes 2021-07-22 19:10:34 +02:00
Gabriele Gristina
db7cb16a3b using sha1_hmac_update_global 2021-07-21 10:19:05 +02:00
Gabriele Gristina
7ea5da51a8 Added hash-mode: SNMPv3 HMAC-SHA1-96 2021-07-20 22:24:35 +02:00
Jens Steube
41156bb0a5 Improve PIM fix for use on macOS 2021-07-20 15:10:49 +02:00
Jens Steube
a026171fb3
Merge pull request #2879 from jtojanen/master
Implement gpu_bzero
2021-07-20 13:00:37 +02:00
Jens Steube
777512e862 Fixed false negative in all VeraCrypt hash-modes if both conditions are met: 1. use CPU for cracking and 2. PIM range was used 2021-07-20 11:31:31 +02:00
Jukka Ojanen
4c0f6dd263
Merge branch 'hashcat:master' into master 2021-07-20 12:00:41 +03:00
Jens Steube
92dc57df28 Revert adding make_u32x() for constants to crypto primitives, fails on OpenCL 2021-07-20 10:34:34 +02:00
Jens Steube
bdb7999f07 Switch HIP vector datatypes to OpenCL like ext_vector_type() 2021-07-19 20:24:30 +02:00
Jens Steube
9421e1f89d Fixed datatype in function sha384_hmac_init_vector_128() that could come into effect if vector datatype was manually set 2021-07-19 15:58:38 +02:00
Jens Steube
0d8b4b74ad More CUDA special backports to HIP 2021-07-18 22:56:22 +02:00
Jens Steube
257098a301 Get rid of hip/hip_runtime.h dependancy 2021-07-18 21:14:45 +02:00
Gabriele Gristina
977ceba046 Added hash-modes: VirtualBox (PBKDF2-HMAC-SHA256 & AES-128-XTS) and VirtualBox (PBKDF2-HMAC-SHA256 & AES-256-XTS) 2021-07-17 18:38:04 +02:00
Jens Steube
bd92589af1 Optimize BLAKE2B_ROUND() 64 bit rotates 2021-07-17 18:18:22 +02:00
Jens Steube
3becb253d3 Enable vector datatypes for HIP 2021-07-17 18:01:01 +02:00
Jukka Ojanen
a2a1d04bcf Implement gpu_bzero 2021-07-17 19:00:10 +03:00
Gabriele Gristina
0b19c4a688 fix m26700_comp, cleanup changes.txt 2021-07-17 14:44:36 +02:00
Jens Steube
141b59b80e
Merge pull request #2883 from hashcat/HIP
Merge HIP branch into master
2021-07-16 19:54:00 +02:00
Michael Kruger
39e78702e0 Move gid check below SHM stuff to imitate m06232 2021-07-16 16:01:57 +02:00
Jens Steube
45e65dd05a Backport more ROCm based optimizations to HIP 2021-07-15 23:34:27 +02:00
Gabriele Gristina
0f055d93ba fix 26700 struct name 2021-07-15 20:59:54 +02:00
Cablethief
431f6ec80e remove fast hash 27000 2021-07-15 16:43:48 +02:00
Jens Steube
4730cf6e79 WinZip Kernel: Increase supported data length from 8k to 16mb 2021-07-15 16:42:29 +02:00
Cablethief
02a9f3471b 27000 slow hash, same doubts as the first, but now with the shared mem concerns 2021-07-15 16:42:18 +02:00
Jens Steube
d130cc66b3 Optimize ISA code on HIP for V_ALIGNBIT_B32 using a different template for inline assembly 2021-07-15 09:57:41 +02:00
Gabriele Gristina
50aeaa299d Added hash-mode: VMware VMX (PBKDF2-HMAC-SHA1 + AES-256-CBC) 2021-07-14 19:31:21 +02:00
Jens Steube
cf512faa53 Update large switch() cases in inc_common.cl and some inline assembly common functions for devices managed with HIP backend 2021-07-14 17:06:20 +02:00
Cablethief
a38b5b3e25 27100 Slow hash, have doubts on how to correctly set iterations 2021-07-14 14:58:26 +02:00
Jens Steube
4b68a8fb4f
Merge pull request #2878 from hashcat/master
Backports
2021-07-13 22:26:28 +02:00
Jens Steube
4e773f32b2 Fix variable declaration block level in -m 19600 and -m 19700 2021-07-13 22:22:53 +02:00
Jens Steube
7faf6859d6 Backport hand-optimized compiler settings in modules from ROCM to HIP
Backport DECLSPEC settings from ROCM to HIP
2021-07-13 20:45:01 +02:00
Jens Steube
674ca7d88f Add GPU threads to kernel cache checksum because it has an influence on HIP offline compile options
Add V_ALIGNBIT_B32 inline assembly wrapper because HIP does not provide amd_bitalign()
2021-07-12 11:27:05 +02:00
Jens Steube
23c3c178bf Limit max threads per block to 64 to enable offline compiler to make better use if available registers
Fix double free() for hip_event1/hip_event2 and hip_stream causes segfault
Replace hc_cuCtxSetCurrent() with hc_cuCtxPushCurrent() in order to align changes with HIP
Add vector datatype operators (if we decide to use them - currently unused)
2021-07-12 09:28:26 +02:00
Jens Steube
1ab56afb9e
Merge pull request #2874 from hashcat/master
Backports from master
2021-07-12 08:57:50 +02:00
Jens Steube
5d05c95a2f Fix address space of crc32tab[] in -m 172xx 2021-07-12 08:56:10 +02:00
Jens Steube
20f7febd4c Workaround too intensive optimization in -m 2000 using HIPRTC 2021-07-11 15:54:13 +02:00
Jens Steube
1b84a9e53b Add missing backports from code base v6.2.2
Fix context to thread management
Fix missing code in selftest.c, autotune.c, hashes.c, dispatch.c and backend.c
Use IS_HIP depending code makes it easier for future optimization related to inline assembly calls - instead of using IS_CUDA || IS_HIP
See TODO markers for more optimizations / next steps
2021-07-11 12:38:59 +02:00
Jens Steube
a22f8149fc
Merge branch 'HIP' into hip 2021-07-10 21:34:09 +02:00
reger-men
ea7b74389f First draft HIP Version 2021-07-09 03:50:40 +00:00
Jens Steube
95861aa5d7 Fix password address space in m25600_init kernel 2021-07-07 10:35:41 +02:00
Jens Steube
17a47e390c Merge branch 'master' of https://github.com/hashcat/hashcat 2021-07-06 13:06:28 +02:00
Jens Steube
f602dd4e4f Bcrypt-SHA1/MD5 Kernels: Get rid of local memory use for binary to hex conversion to avoid false negatives on several OpenCL runtimes 2021-07-06 12:53:38 +02:00
Jens Steube
29c971187a
Merge pull request #2863 from matrix/metamask
Added hash-mode: MetaMask Wallet
2021-07-06 09:49:19 +02:00
Jens Steube
4934e92cda OpenCL Runtime: Workaround a problem of the AMD OpenCL GPU driver under macOS which results in false negatives due to changes in the kernel crypto library 2021-07-05 11:33:12 +02:00
Gabriele Gristina
7371cbebe7 add min/max data length limits to MetaMask Wallet 2021-07-04 20:14:25 +02:00
Gabriele Gristina
f571329c4f Added hash-mode: MetaMask Wallet 2021-07-04 15:47:52 +02:00
Jens Steube
badc51f3e7 WPA Kernels: Increased performance by 3.5% for backend devices controlled by CUDA backend 2021-07-04 12:24:55 +02:00
Jens Steube
7366c47447 Added hash-mode: iPhone passcode (UID key + System Keybag) 2021-07-03 19:03:06 +02:00
Jens Steube
56c2243dfb KeePass: Increase supported size for KeePass 1 databases from 300kB to 16MB 2021-06-30 13:16:03 +02:00
Jens Steube
909d5e64a5 Added hash-mode: AES-128/192/256-ECB NOKDF
This mode is probably very rare in real-life scenarios,
but it is a nice template for kernels which do
not use a KDF,
or use AES,
or simple fast hashes with lookup tables
or simple optimized kernels in general
2021-06-26 17:12:10 +02:00
Jens Steube
b2193e1af4 Add rc4_next_16_global() and fix address space of edata buffer in -m 13100 and -m18200 2021-06-23 08:36:17 +02:00
Jens Steube
f43ba9c8d6 minor whitespace fixes 2021-06-22 08:55:44 +02:00
Royce Williams
e4626dc471 Merge branch 'master' of https://github.com/roycewilliams/hashcat 2021-06-21 07:48:13 -08:00
Royce Williams
1e88990b46 minor whitespace fixes, per #2841 2021-06-21 07:47:22 -08:00
Jens Steube
05125eb67c LM: Workaround JiT compiler bug in -m 3000 on NV leading to false negatives with large amount of hashes 2021-06-18 23:48:39 +02:00
Jens Steube
62397283c1 VeraCrypt: Increase password length support for non-boot volumes from 64 to 128.
See https://github.com/hashcat/hashcat/issues/2616 for details.
2021-06-17 09:45:30 +02:00
Jens Steube
8548af5e79 Fixed false negative in hash-mode 15900 (DPAPI masterkey file v2) if password was longer than 64 characters.
Bug was also present in SHA512 based TC and VC kernels, but had no effect because maximum password length is 64.
2021-06-16 12:13:04 +02:00
Jens Steube
8ebf4b9858 Fixed 'E' rule in pure kernel mode which was ignoring letters that are in positions that are multiples of 4 2021-06-13 15:49:11 +02:00
Jens Steube
eddd11441f
Merge pull request #2664 from matrix/hash-mode-26000
Added hash-mode 26000 - OpenEdge Progress Encode
2021-06-12 18:22:09 +02:00
Jens Steube
ebeb003552 Unit-Tests: Added missing unit-test for Stargazer Stellar Wallet XLM 2021-06-10 09:02:26 +02:00
Jens Steube
796beba2bf
Merge branch 'master' into hash-mode-26000 2021-06-09 13:20:35 +02:00
Jens Steube
733f5e82ec Renamed -m 7010 to -m 26300 2021-06-08 11:12:54 +02:00
Jens Steube
9df4b7df92
Merge pull request #2810 from YSaxon/fortigate256
Added -m 07010: Fortigate256 (prefixed with SH2 instead of AK1)
2021-06-08 11:02:36 +02:00
Yaakov Saxon
db6a079bca (07000->07010 in code) 2021-06-07 12:34:30 -04:00
Yaakov Saxon
7a15466442 Update module_07000.c
better?
sensible change
revert dgst_pos


update kernel file


update other 2 pure kernels


updating DGST_POS* to match other SHA256 hash formats


comments


removing optimized that I haven't updated


updated testfile


renaming to 07010


restoring original 07000 files


comments
2021-06-07 12:07:29 -04:00
Gabriele Gristina
9b529dc34c update hash-mode to 26200 2021-06-05 00:10:17 +02:00
Jens Steube
978f4c7a85
Merge pull request #2591 from trounce1/20720
Added -m 20720 sha256($salt.sha256($pass))
2021-06-04 08:10:49 +02:00
Jens Steube
074026a1df Update -m 3500 and -m 5000 -a 3 function calls in optimized mode 2021-06-03 07:57:59 +02:00
Jens Steube
f016deaa76
Merge branch 'master' into new-3500-mode 2021-06-03 07:52:17 +02:00
Jens Steube
38beceb170
Merge pull request #2594 from trounce1/5000
Added -m 5000 = sha1(sha1($salt.$pass.$salt))
2021-06-03 07:51:48 +02:00
Jens Steube
ce8c121b50 BCRYPT Kernels: Improved bcrypt performance by 6.5% for high-end NVIDIA GPU devices using CUDA backend 2021-06-01 22:52:07 +02:00
Jens Steube
8901e657a5 RC4 Kernels: Improved performance by 20%+ for hash-modes Kerberos 5 (etype 23), MS Office (<= 2003) and PDF (<= 1.6) by using new RC4 code 2021-05-31 08:28:13 +02:00
Jens Steube
4e565efcf9 RC4: Updated hash-mode 7500, 9710, 9720, 10400 and 10410 to new RC4 crypto library code, improving performance by 20% or more 2021-05-30 19:53:28 +02:00
Jens Steube
2a55fd7f33 RC4: Update -m 97x0 and -m 104x0 to new RC4 crypto library code, improving performance by 20% or more 2021-05-30 15:40:33 +02:00
Jens Steube
58467e0869 PDF: Update -m 10400 and -m 10410 to new RC4 crypto library code, improving performance by 20% or more 2021-05-30 14:35:28 +02:00
Jens Steube
3192f55707 PDF: Update -m 10410 to new RC4 crypto library code, improving performance by 22% - 32%. 2021-05-30 12:36:17 +02:00
Jens Steube
6991c3bb65 Kernels: Add RC4 cipher to crypto library with shared memory access pattern which is not causing any bank conflicts 2021-05-30 12:15:03 +02:00
Jens Steube
e79a2aa90c Update -m 10870 -a 3 function calls in optimized mode 2021-05-29 10:49:06 +02:00
Jens Steube
99faf3e6e1
Merge branch 'master' into hash-mode-10870 2021-05-29 10:46:28 +02:00
Jens Steube
0b52409590 Modules: Recategorized HASH_CATEGORY option in various modules 2021-05-28 20:06:01 +02:00
Jens Steube
71969a33cd
Merge branch 'master' into hash-mode-1770 2021-05-28 18:58:19 +02:00
Jens Steube
7078ee4fe9
Merge branch 'master' into hash-mode-10870 2021-05-28 18:57:24 +02:00
Jens Steube
fbc390788c
Merge branch 'master' into hash-mode-1470 2021-05-28 18:56:39 +02:00
Cablethief
e818b56cf3 Change module number to match future grouping 2021-05-28 14:45:41 +02:00
Jens Steube
1785dfe089 Update -m 70 and -m 170 -a 3 function calls in optimized mode 2021-05-27 19:34:32 +02:00
Jens Steube
ef3b2fa61c
Merge branch 'master' into hash-mode-170 2021-05-27 19:31:44 +02:00
Jens Steube
6314c24f16
Merge pull request #2648 from matrix/hash-mode-70
Added hash-mode 70 - md5(utf16le(pass))
2021-05-27 19:30:54 +02:00
Jens Steube
66ba860c8a Update -m 108x0 -a 3 function calls in optimized mode 2021-05-26 19:34:59 +02:00
Jens Steube
6ce0b568b7
Merge branch 'master' into hash-mode-10840 2021-05-26 19:18:14 +02:00
Jens Steube
7378159487
Merge branch 'master' into hash-mode-10830 2021-05-26 19:16:22 +02:00
Jens Steube
6682574bd2 Update -m 108x0 -a 3 function calls and update some GitHub issue templates 2021-05-25 09:26:12 +02:00
Jens Steube
94df2349cb
Merge branch 'master' into hash-mode-10820 2021-05-25 08:52:57 +02:00
Jens Steube
0aba4cab6b
Merge pull request #2656 from matrix/hash-mode-10810
Added hash-mode 10810 - sha384(pass.salt)
2021-05-25 08:51:44 +02:00
Jens Steube
3faad47b48 Revert "Merge branch 'matrix-hash-mode-25100'"
This reverts commit a236b55cae, reversing
changes made to 1d965c3e78.
2021-05-23 10:18:51 +02:00
Jens Steube
87f3b2caaf Merge branch 'hash-mode-25100' of https://github.com/matrix/hashcat into matrix-hash-mode-25100 2021-05-23 09:58:42 +02:00
Jens Steube
19d3696bd5 UTF8-to-UTF16: Replaced naive UTF8 to UTF16 conversion with true conversion for RAR3, AES Crypt, MultiBit HD (scrypt) and Umbraco HMAC-SHA1 2021-05-22 11:14:51 +02:00
Jens Steube
0457016834 UTF8-to-UTF16: Replaced naive UTF8 to UTF16 conversion with true conversion for RAR3, AES Crypt and MultiBit HD (scrypt) 2021-05-22 10:37:51 +02:00
Jens Steube
0a6080505b AES Crypt Plugin: Replaced naive with true UTF8 to UTF16 conversion, reduced max password length to 128 and improved performance 2021-05-21 22:12:32 +02:00
Jens Steube
df8a773341 Restore maximum password length 128 for pure RAR3 kernels 2021-05-21 22:04:24 +02:00
Jens Steube
fd51086395 RAR3 Plugins: Replaced naive with true UTF8 to UTF16 conversion in optimized and pure kernels 2021-05-21 12:46:14 +02:00
Jens Steube
23bbe457a0 RAR3-hp Plugin: Replaced naive with true UTF8 to UTF16 conversion in optimized and pure kernel 2021-05-21 11:24:43 +02:00
Jens Steube
5c6501444a Kernels: Add standalone true UTF8 to UTF16 converter kernel that runs after amplifier. Use OPTS_TYPE_POST_AMP_UTF16LE from plugin 2021-05-20 14:34:24 +02:00
Jens Steube
9028938998 Rename -m 3202 to -m 25800 2021-05-18 19:25:47 +02:00
Jens Steube
9bb45b4f4d
Merge branch 'master' into bcryptsha1 2021-05-18 19:20:28 +02:00
Jens Steube
aa7c579736 Added hash-mode: MurmurHash 2021-05-18 10:58:13 +02:00
Jens Steube
307cb63661
Merge pull request #2695 from matrix/cryptoapi
Added Linux Kernel Crypto API (2.4) KPA
2021-05-17 23:41:10 +02:00
Jens Steube
751d3a6454 Rename -m 3201 to -m 25600 2021-05-17 13:23:50 +02:00
Jens Steube
ae2580a542
Merge pull request #2637 from matrix/bcryptmd5
Added hash-mode 3201 - bcrypt(md5(pass)) / bcryptmd5
2021-05-17 13:17:25 +02:00
Jens Steube
bb402b784a Update module_unstable_warning for benchmark short selection on macOS for CPU and GPU; Allow use of GPU without --force testwise 2021-05-10 14:36:41 +02:00
Jens Steube
78a053800f Remove module_unstable_warning() entries for AMD (legacy) driver after workaround with zero check on context length is in use 2021-05-09 10:33:05 +00:00
Jens Steube
9bf0f36d0a Get rid of MAYBE_VOLATILE for context position by replacing it with zero length check 2021-05-09 11:43:32 +02:00
Jens Steube
e4dab0f1bf OpenCL Runtime: Workaround JiT compiler segfault on legacy AMDGPU driver compiling RAR3 OpenCL kernel 2021-05-09 07:38:22 +00:00
Jens Steube
9813811493 Remove truncation of buffer in hc_enc_next() to workaround AMD JiT compiler (legacy) issue 2021-05-08 16:38:48 +02:00
Gabriele Gristina
e30febc8b8 cryptoapi opts (2) and test unit 2021-05-08 02:06:05 +02:00
Gabriele Gristina
17a83a50af cryptoapi opts (1) 2021-05-06 23:24:46 +02:00
Jens Steube
a02b2ccd5f Fix type of address space of pointer in sha512_hmac_init_global_utf16le_swap() 2021-05-01 14:43:10 +02:00
Jens Steube
0439f0c4a1 Refactor UTF8 to UTF16 conversion from fixed size to a dynamic size using a context struct. This allows handle buffer sizes of arbitrary length for conversion 2021-05-01 12:49:43 +02:00
Jens Steube
f8ea1d5e78 Improve performance of test_any_8th_bit() by manually unrolling a few first steps 2021-04-30 17:22:31 +02:00
Jens Steube
b7dffd9259 Improve performance for UTF8->UTF16 conversion
Reverted d343e2c4a0 and ee26805138
Adds a test to decide whatever conversion technique to use. If all UTF8 characters are 7 bit, there's no need for regular conversion and we can stick to naive conversion.
2021-04-30 16:55:30 +02:00
Jens Steube
736290151d
Merge pull request #2638 from matrix/20710_fix
Fixed hash-mode 20710: removed unused code and fixed module_constraints
2021-04-29 16:08:00 +02:00
Jens Steube
17b9fcb925 Fix unicode conversion in -m 24800 2021-04-26 09:24:06 +02:00
Jens Steube
1879cbefd2 Fixed vector datatype support in -m 21100 only -P mode and only -a 3 mode were affected 2021-04-25 21:25:28 +02:00
Jens Steube
092e838cf4 Fixed buffer overflow in -m 4710 in -P mode and only in single hash mode if salt length was larger than 32 byte 2021-04-25 20:50:59 +02:00
Jens Steube
7a5f3610ca Fixed buffer overflow in -m 1800 in -O mode which is optimized to handle only password candidates up to length 15 2021-04-25 17:46:03 +02:00
Jens Steube
146a5237b5 Improved SCRYPT performance and updated hashcat.hctune 2021-04-24 14:01:13 +02:00
Jens Steube
1dac869cb7 Removed unnecessary swaps in SCRYPT based algorithms 2021-04-23 20:55:13 +02:00
Jens Steube
6c54314c2d Update -a 3 kernels to make use of new parameter salt_repeat 2021-04-22 19:42:49 +02:00
Jens Steube
2b069a94fd Reduce some register allocation in SCRYPT based kernels to improve speed on GPU with low resources 2021-04-22 00:26:09 +02:00
Jens Steube
15f35fa68c Scrypt Kernels: Reduced kernel wait times by making it a true split kernel where iteration count = N value 2021-04-21 15:59:14 +02:00
Jens Steube
62fc3601bb Wrap atomic functions with hc_ prefix to have better platform control 2021-04-20 17:47:44 +02:00
Jens Steube
73cc3170f4 Fixed both false negative and false positive result in -m 3000 in -a 3 (affected only NVIDIA GPU) 2021-04-20 17:14:13 +02:00
Jens Steube
4ddbd7e047 Fix false positives in -m 3000 in -a 3 mode 2021-04-20 11:14:28 +02:00
Jens Steube
5e9adac1c2 Update module_unstable_warning() results after Crypt::CBC has been fixed 2021-04-19 07:51:26 +02:00
Jens Steube
dc79983bc9 Fix out-of-boundary reads in -m 24800 pure kernels 2021-04-18 15:19:14 +02:00