Added new option --kernel-threads to manually override the automatically calculated number of threads

docs/changes.txt

@@ -6,6 +6,7 @@
 
 - Added support to use --stdout in brain-client mode
 - Added new option --stdin-timeout-abort, to set how long hashcat should wait for stdin input before aborting
+- Added new option --kernel-threads to manually override the automatically calculated number of threads
 
 ##
 ## Algorithms

extra/tab_completion/hashcat.sh

@@ -189,8 +189,8 @@ _hashcat ()
   local BUILD_IN_CHARSETS='?l ?u ?d ?a ?b ?s ?h ?H'
 
   local SHORT_OPTS="-m -a -V -v -h -b -t -o -p -c -d -w -n -u -j -k -r -g -1 -2 -3 -4 -i -I -s -l -O -S -z"
-  local LONG_OPTS="--hash-type --attack-mode --version --help --quiet --benchmark --benchmark-all --hex-salt --hex-wordlist --hex-charset --force --status --status-timer --machine-readable --loopback --markov-hcstat2 --markov-disable --markov-classic --markov-threshold --runtime --session --speed-only --progress-only --restore --restore-file-path --restore-disable --outfile --outfile-format --outfile-autohex-disable --outfile-check-timer --outfile-check-dir --wordlist-autohex-disable --separator --show --left --username --remove --remove-timer --potfile-disable --potfile-path --debug-mode --debug-file --induction-dir --segment-size --bitmap-min --bitmap-max --cpu-affinity --example-hashes --opencl-info --opencl-devices --opencl-platforms --opencl-device-types --opencl-vector-width --workload-profile --kernel-accel --kernel-loops --nvidia-spin-damp --gpu-temp-disable --gpu-temp-abort --skip --limit --keyspace --rule-left --rule-right --rules-file --generate-rules --generate-rules-func-min --generate-rules-func-max --generate-rules-seed --custom-charset1 --custom-charset2 --custom-charset3 --custom-charset4 --increment --increment-min --increment-max --logfile-disable --scrypt-tmto --truecrypt-keyfiles --veracrypt-keyfiles --veracrypt-pim --stdout --keep-guessing --hccapx-message-pair --nonce-error-corrections --encoding-from --encoding-to --optimized-kernel-enable --self-test-disable  --slow-candidates --brain-server --brain-client --brain-client-features --brain-host --brain-port --brain-session --brain-session-whitelist --brain-password"
+  local LONG_OPTS="--hash-type --attack-mode --version --help --quiet --benchmark --benchmark-all --hex-salt --hex-wordlist --hex-charset --force --status --status-timer --machine-readable --loopback --markov-hcstat2 --markov-disable --markov-classic --markov-threshold --runtime --session --speed-only --progress-only --restore --restore-file-path --restore-disable --outfile --outfile-format --outfile-autohex-disable --outfile-check-timer --outfile-check-dir --wordlist-autohex-disable --separator --show --left --username --remove --remove-timer --potfile-disable --potfile-path --debug-mode --debug-file --induction-dir --segment-size --bitmap-min --bitmap-max --cpu-affinity --example-hashes --opencl-info --opencl-devices --opencl-platforms --opencl-device-types --opencl-vector-width --workload-profile --kernel-accel --kernel-loops --kernel-threads --nvidia-spin-damp --gpu-temp-disable --gpu-temp-abort --skip --limit --keyspace --rule-left --rule-right --rules-file --generate-rules --generate-rules-func-min --generate-rules-func-max --generate-rules-seed --custom-charset1 --custom-charset2 --custom-charset3 --custom-charset4 --increment --increment-min --increment-max --logfile-disable --scrypt-tmto --truecrypt-keyfiles --veracrypt-keyfiles --veracrypt-pim --stdout --keep-guessing --hccapx-message-pair --nonce-error-corrections --encoding-from --encoding-to --optimized-kernel-enable --self-test-disable  --slow-candidates --brain-server --brain-client --brain-client-features --brain-host --brain-port --brain-session --brain-session-whitelist --brain-password"
-  local OPTIONS="-m -a -t -o -p -c -d -w -n -u -j -k -r -g -1 -2 -3 -4 -s -l --hash-type --attack-mode --status-timer --markov-hcstat2 --markov-threshold --runtime --session --timer --outfile --outfile-format --outfile-check-timer --outfile-check-dir --separator --remove-timer --potfile-path --restore-file-path --debug-mode --debug-file --induction-dir --segment-size --bitmap-min --bitmap-max --cpu-affinity --opencl-devices --opencl-platforms --opencl-device-types --opencl-vector-width --workload-profile --kernel-accel --kernel-loops --nvidia-spin-damp --gpu-temp-abort --skip --limit --rule-left --rule-right --rules-file --generate-rules --generate-rules-func-min --generate-rules-func-max --generate-rules-seed --custom-charset1 --custom-charset2 --custom-charset3 --custom-charset4 --increment-min --increment-max --scrypt-tmto --truecrypt-keyfiles --veracrypt-keyfiles --veracrypt-pim --hccapx-message-pair --nonce-error-corrections --encoding-from --encoding-to  --brain-host --brain-password --brain-port --brain-session --brain-whitelist-session --stdin-timeout-abort"
+  local OPTIONS="-m -a -t -o -p -c -d -w -n -u -j -k -r -g -1 -2 -3 -4 -s -l --hash-type --attack-mode --status-timer --markov-hcstat2 --markov-threshold --runtime --session --timer --outfile --outfile-format --outfile-check-timer --outfile-check-dir --separator --remove-timer --potfile-path --restore-file-path --debug-mode --debug-file --induction-dir --segment-size --bitmap-min --bitmap-max --cpu-affinity --opencl-devices --opencl-platforms --opencl-device-types --opencl-vector-width --workload-profile --kernel-accel --kernel-loops --kernel-threads --nvidia-spin-damp --gpu-temp-abort --skip --limit --rule-left --rule-right --rules-file --generate-rules --generate-rules-func-min --generate-rules-func-max --generate-rules-seed --custom-charset1 --custom-charset2 --custom-charset3 --custom-charset4 --increment-min --increment-max --scrypt-tmto --truecrypt-keyfiles --veracrypt-keyfiles --veracrypt-pim --hccapx-message-pair --nonce-error-corrections --encoding-from --encoding-to  --brain-host --brain-password --brain-port --brain-session --brain-whitelist-session --stdin-timeout-abort"
 
   COMPREPLY=()
   local cur="${COMP_WORDS[COMP_CWORD]}"

include/types.h

@@ -560,6 +560,7 @@ typedef enum user_options_defaults
   KEEP_GUESSING            = false,
   KERNEL_ACCEL             = 0,
   KERNEL_LOOPS             = 0,
+  KERNEL_THREADS           = 0,
   KEYSPACE                 = false,
   LEFT                     = false,
   LIMIT                    = 0,
@@ -654,6 +655,7 @@ typedef enum user_options_map
   IDX_KEEP_GUESSING             = 0xff1b,
   IDX_KERNEL_ACCEL              = 'n',
   IDX_KERNEL_LOOPS              = 'u',
+  IDX_KERNEL_THREADS            = 'T',
   IDX_KEYSPACE                  = 0xff1c,
   IDX_LEFT                      = 0xff1d,
   IDX_LIMIT                     = 'l',
@@ -1086,8 +1088,6 @@ typedef struct hc_device_param
   u64     kernel_local_mem_size_aux3;
   u64     kernel_local_mem_size_aux4;
 
-  u32     kernel_threads;
-
   u32     kernel_accel;
   u32     kernel_accel_prev;
   u32     kernel_accel_min;
@@ -1098,6 +1098,7 @@ typedef struct hc_device_param
   u32     kernel_loops_max;
   u32     kernel_loops_min_sav; // the _sav are required because each -i iteration
   u32     kernel_loops_max_sav; // needs to recalculate the kernel_loops_min/max based on the current amplifier count
+  u32     kernel_threads;
 
   u64     kernel_power;
   u64     hardware_power;
@@ -1684,6 +1685,7 @@ typedef struct user_options
   bool         increment_min_chgd;
   bool         kernel_accel_chgd;
   bool         kernel_loops_chgd;
+  bool         kernel_threads_chgd;
   bool         nonce_error_corrections_chgd;
   bool         nvidia_spin_damp_chgd;
   bool         opencl_vector_width_chgd;
@@ -1784,6 +1786,7 @@ typedef struct user_options
   u32          increment_min;
   u32          kernel_accel;
   u32          kernel_loops;
+  u32          kernel_threads;
   u32          markov_threshold;
   u32          nonce_error_corrections;
   u32          nvidia_spin_damp;

src/interface.c

@@ -28296,9 +28296,10 @@ void hashconfig_destroy (hashcat_ctx_t *hashcat_ctx)
 
 u32 hashconfig_forced_kernel_threads (hashcat_ctx_t *hashcat_ctx)
 {
-  hashconfig_t *hashconfig = hashcat_ctx->hashconfig;
+  const hashconfig_t   *hashconfig   = hashcat_ctx->hashconfig;
+  const user_options_t *user_options = hashcat_ctx->user_options;
 
-  u32 kernel_threads = 0;
+  u32 kernel_threads = user_options->kernel_threads;
 
   // this should have a kernel hint attribute in the kernel files
   // __attribute__((reqd_work_group_size(X, 1, 1)))

src/usage.c

@@ -95,6 +95,7 @@ static const char *const USAGE_BIG[] =
   " -w, --workload-profile         | Num  | Enable a specific workload profile, see pool below   | -w 3",
   " -n, --kernel-accel             | Num  | Manual workload tuning, set outerloop step size to X | -n 64",
   " -u, --kernel-loops             | Num  | Manual workload tuning, set innerloop step size to X | -u 256",
+  " -T, --kernel-thread            | Num  | Manual workload tuning, set thread count to X        | -T 64",
   "     --nvidia-spin-damp         | Num  | Workaround NVIDIAs CPU burning loop bug, in percent  | --nvidia-spin-damp=50",
   "     --gpu-temp-disable         |      | Disable temperature and fanspeed reads and triggers  |",
   "     --gpu-temp-abort           | Num  | Abort if GPU temperature reaches X degrees Celsius   | --gpu-temp-abort=100",

src/user_options.c

@@ -19,9 +19,9 @@
 #endif
 
 #ifdef WITH_BRAIN
-static const char *short_options = "hVvm:a:r:j:k:g:o:t:d:D:n:u:c:p:s:l:1:2:3:4:iIbw:OSz";
+static const char *short_options = "hVvm:a:r:j:k:g:o:t:d:D:n:u:T:c:p:s:l:1:2:3:4:iIbw:OSz";
 #else
-static const char *short_options = "hVvm:a:r:j:k:g:o:t:d:D:n:u:c:p:s:l:1:2:3:4:iIbw:OS";
+static const char *short_options = "hVvm:a:r:j:k:g:o:t:d:D:n:u:T:c:p:s:l:1:2:3:4:iIbw:OS";
 #endif
 
 static const struct option long_options[] =
@@ -62,6 +62,7 @@ static const struct option long_options[] =
   {"keep-guessing",             no_argument,       NULL, IDX_KEEP_GUESSING},
   {"kernel-accel",              required_argument, NULL, IDX_KERNEL_ACCEL},
   {"kernel-loops",              required_argument, NULL, IDX_KERNEL_LOOPS},
+  {"kernel-threads",            required_argument, NULL, IDX_KERNEL_THREADS},
   {"keyspace",                  no_argument,       NULL, IDX_KEYSPACE},
   {"left",                      no_argument,       NULL, IDX_LEFT},
   {"limit",                     required_argument, NULL, IDX_LIMIT},
@@ -186,6 +187,7 @@ int user_options_init (hashcat_ctx_t *hashcat_ctx)
   user_options->keep_guessing             = KEEP_GUESSING;
   user_options->kernel_accel              = KERNEL_ACCEL;
   user_options->kernel_loops              = KERNEL_LOOPS;
+  user_options->kernel_threads            = KERNEL_THREADS;
   user_options->keyspace                  = KEYSPACE;
   user_options->left                      = LEFT;
   user_options->limit                     = LIMIT;
@@ -302,6 +304,7 @@ int user_options_getopt (hashcat_ctx_t *hashcat_ctx, int argc, char **argv)
       case IDX_WORKLOAD_PROFILE:
       case IDX_KERNEL_ACCEL:
       case IDX_KERNEL_LOOPS:
+      case IDX_KERNEL_THREADS:
       case IDX_NVIDIA_SPIN_DAMP:
       case IDX_GPU_TEMP_ABORT:
       case IDX_HCCAPX_MESSAGE_PAIR:
@@ -427,6 +430,8 @@ int user_options_getopt (hashcat_ctx_t *hashcat_ctx, int argc, char **argv)
                                          user_options->kernel_accel_chgd         = true;                            break;
       case IDX_KERNEL_LOOPS:             user_options->kernel_loops              = hc_strtoul (optarg, NULL, 10);
                                          user_options->kernel_loops_chgd         = true;                            break;
+      case IDX_KERNEL_THREADS:           user_options->kernel_threads            = hc_strtoul (optarg, NULL, 10);
+                                         user_options->kernel_threads_chgd       = true;                            break;
       case IDX_NVIDIA_SPIN_DAMP:         user_options->nvidia_spin_damp          = hc_strtoul (optarg, NULL, 10);
                                          user_options->nvidia_spin_damp_chgd     = true;                            break;
       case IDX_GPU_TEMP_DISABLE:         user_options->gpu_temp_disable          = true;                            break;
@@ -804,6 +809,33 @@ int user_options_sanity (hashcat_ctx_t *hashcat_ctx)
     }
   }
 
+  if (user_options->kernel_threads_chgd == true)
+  {
+    if (user_options->force == false)
+    {
+      event_log_error (hashcat_ctx, "The manual use of the -T option (or --kernel-threads) is outdated.");
+
+      event_log_warning (hashcat_ctx, "You can use --force to override this, but do not report related errors.");
+      event_log_warning (hashcat_ctx, NULL);
+
+      return -1;
+    }
+
+    if (user_options->kernel_threads < 1)
+    {
+      event_log_error (hashcat_ctx, "Invalid kernel-threads specified.");
+
+      return -1;
+    }
+
+    if (user_options->kernel_threads > 1024)
+    {
+      event_log_error (hashcat_ctx, "Invalid kernel-threads specified.");
+
+      return -1;
+    }
+  }
+
   if ((user_options->workload_profile < 1) || (user_options->workload_profile > 4))
   {
     event_log_error (hashcat_ctx, "workload-profile %u is not available.", user_options->workload_profile);
@@ -1706,11 +1738,18 @@ void user_options_info (hashcat_ctx_t *hashcat_ctx)
       event_log_info (hashcat_ctx, "* --opencl-vector-width=%u", user_options->opencl_vector_width);
     }
 
-    if ((user_options->kernel_accel_chgd == true) || (user_options->kernel_loops_chgd == true))
+    if (user_options->kernel_accel_chgd == true)
     {
       event_log_info (hashcat_ctx, "* --kernel-accel=%u", user_options->kernel_accel);
+    }
+    else if (user_options->kernel_loops_chgd == true)
+    {
       event_log_info (hashcat_ctx, "* --kernel-loops=%u", user_options->kernel_loops);
     }
+    else if (user_options->kernel_threads_chgd == true)
+    {
+      event_log_info (hashcat_ctx, "* --kernel-threads=%u", user_options->kernel_threads);
+    }
     else
     {
       if (user_options->workload_profile_chgd == true)
@@ -1758,11 +1797,18 @@ void user_options_info (hashcat_ctx_t *hashcat_ctx)
       event_log_info (hashcat_ctx, "# option: --opencl-vector-width=%u", user_options->opencl_vector_width);
     }
 
-    if ((user_options->kernel_accel_chgd == true) || (user_options->kernel_loops_chgd == true))
+    if (user_options->kernel_accel_chgd == true)
     {
       event_log_info (hashcat_ctx, "# option: --kernel-accel=%u", user_options->kernel_accel);
+    }
+    else if (user_options->kernel_loops_chgd == true)
+    {
       event_log_info (hashcat_ctx, "# option: --kernel-loops=%u", user_options->kernel_loops);
     }
+    else if (user_options->kernel_threads_chgd == true)
+    {
+      event_log_info (hashcat_ctx, "# option: --kernel-threads=%u", user_options->kernel_threads);
+    }
     else
     {
       if (user_options->workload_profile_chgd == true)
@@ -2563,6 +2609,7 @@ void user_options_logger (hashcat_ctx_t *hashcat_ctx)
   logfile_top_uint   (user_options->keep_guessing);
   logfile_top_uint   (user_options->kernel_accel);
   logfile_top_uint   (user_options->kernel_loops);
+  logfile_top_uint   (user_options->kernel_threads);
   logfile_top_uint   (user_options->keyspace);
   logfile_top_uint   (user_options->left);
   logfile_top_uint   (user_options->logfile_disable);