Use dedicated variables to control shutdown of parallel threads like keypress and...

[hashcat.git] / src / hashcat.c

diff --git a/src/hashcat.c b/src/hashcat.c
index fb38398..1effe6a 100644 (file)
--- a/src/hashcat.c
+++ b/src/hashcat.c
@@ -1,4 +1,4 @@
- /**
+/**
  * Authors.....: Jens Steube <jens.steube@gmail.com>
  *               Gabriele Gristina <matrix@hashcat.net>
  *               magnum <john.magnum@hushmail.com>
@@ -3045,6 +3045,7 @@ static void choose_kernel (hc_device_param_t *device_param, const uint attack_ex
       if (data.devices_status == STATUS_CRACKED) break;
       if (data.devices_status == STATUS_ABORTED) break;
       if (data.devices_status == STATUS_QUIT)    break;
+      if (data.devices_status == STATUS_BYPASS)  break;
 
       /**
        * speed
@@ -3701,6 +3702,7 @@ static void run_cracker (hc_device_param_t *device_param, const uint pws_cnt)
       if (data.devices_status == STATUS_CRACKED) break;
       if (data.devices_status == STATUS_ABORTED) break;
       if (data.devices_status == STATUS_QUIT)    break;
+      if (data.devices_status == STATUS_BYPASS)  break;
 
       /**
        * result
@@ -4163,7 +4165,7 @@ static void *thread_monitor (void *p)
     return (p);
   }
 
-  while ((data.devices_status != STATUS_EXHAUSTED) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+  while (data.shutdown_inner == 0)
   {
     hc_sleep (sleep_time);
 
@@ -4452,7 +4454,7 @@ static void *thread_outfile_remove (void *p)
 
   uint check_left = outfile_check_timer; // or 1 if we want to check it at startup
 
-  while ((data.devices_status != STATUS_EXHAUSTED) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+  while (data.shutdown_inner == 0)
   {
     hc_sleep (1);
 
@@ -4830,7 +4832,7 @@ static void *thread_calc_stdin (void *p)
 
   const uint attack_kern = data.attack_kern;
 
-  while ((data.devices_status != STATUS_EXHAUSTED) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+  while ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
   {
     hc_thread_mutex_lock (mux_dispatcher);
 
@@ -4958,7 +4960,7 @@ static void *thread_calc (void *p)
 
   if (attack_mode == ATTACK_MODE_BF)
   {
-    while ((data.devices_status != STATUS_EXHAUSTED) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+    while ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
     {
       const uint work = get_work (device_param, -1);
 
@@ -5070,7 +5072,7 @@ static void *thread_calc (void *p)
 
     u64 words_cur = 0;
 
-    while ((data.devices_status != STATUS_EXHAUSTED) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+    while ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
     {
       u64 words_off = 0;
       u64 words_fin = 0;
@@ -7362,6 +7364,8 @@ int main (int argc, char **argv)
 
   hc_thread_t *outer_threads = (hc_thread_t *) mycalloc (10, sizeof (hc_thread_t));
 
+  data.shutdown_outer = 0;
+
   if (keyspace == 0 && benchmark == 0 && stdout_flag == 0)
   {
     if ((data.wordlist_mode == WL_MODE_FILE) || (data.wordlist_mode == WL_MODE_MASK))
@@ -7370,10 +7374,6 @@ int main (int argc, char **argv)
 
       outer_threads_cnt++;
     }
-
-    hc_thread_create (outer_threads[outer_threads_cnt], thread_monitor, NULL);
-
-    outer_threads_cnt++;
   }
 
   /**
@@ -14188,7 +14188,7 @@ int main (int argc, char **argv)
 
         device_param->device_name_chksum = device_name_chksum;
 
-        // device_processor_cores
+        // vendor specific
 
         if (device_param->device_type & CL_DEVICE_TYPE_GPU)
         {
@@ -14211,28 +14211,9 @@ int main (int argc, char **argv)
           }
         }
 
-        // device_processor_cores
-
-        if (device_type & CL_DEVICE_TYPE_CPU)
-        {
-          cl_uint device_processor_cores = 1;
-
-          device_param->device_processor_cores = device_processor_cores;
-        }
-
         if (device_type & CL_DEVICE_TYPE_GPU)
         {
-          if (device_vendor_id == VENDOR_ID_AMD)
-          {
-            cl_uint device_processor_cores = 0;
-
-            #define CL_DEVICE_WAVEFRONT_WIDTH_AMD               0x4043
-
-            hc_clGetDeviceInfo (data.ocl, device_param->device, CL_DEVICE_WAVEFRONT_WIDTH_AMD, sizeof (device_processor_cores), &device_processor_cores, NULL);
-
-            device_param->device_processor_cores = device_processor_cores;
-          }
-          else if (device_vendor_id == VENDOR_ID_NV)
+          if (device_vendor_id == VENDOR_ID_NV)
           {
             cl_uint kernel_exec_timeout = 0;
 
@@ -14242,14 +14223,6 @@ int main (int argc, char **argv)
 
             device_param->kernel_exec_timeout = kernel_exec_timeout;
 
-            cl_uint device_processor_cores = 0;
-
-            #define CL_DEVICE_WARP_SIZE_NV                      0x4003
-
-            hc_clGetDeviceInfo (data.ocl, device_param->device, CL_DEVICE_WARP_SIZE_NV, sizeof (device_processor_cores), &device_processor_cores, NULL);
-
-            device_param->device_processor_cores = device_processor_cores;
-
             cl_uint sm_minor = 0;
             cl_uint sm_major = 0;
 
@@ -14285,12 +14258,6 @@ int main (int argc, char **argv)
 
             device_param->nvidia_spin_damp /= 100;
           }
-          else
-          {
-            cl_uint device_processor_cores = 1;
-
-            device_param->device_processor_cores = device_processor_cores;
-          }
         }
 
         // display results
@@ -15019,7 +14986,6 @@ int main (int argc, char **argv)
 
       const char *device_name_chksum      = device_param->device_name_chksum;
       const u32   device_processors       = device_param->device_processors;
-      const u32   device_processor_cores  = device_param->device_processor_cores;
 
       /**
        * create context for each device
@@ -15053,7 +15019,9 @@ int main (int argc, char **argv)
       if (hash_mode ==  3000) kernel_threads = 64; // DES
       if (hash_mode ==  3200) kernel_threads = 8;  // Blowfish
       if (hash_mode ==  7500) kernel_threads = 64; // RC4
+      if (hash_mode ==  8900) kernel_threads = 32; // scrypt
       if (hash_mode ==  9000) kernel_threads = 8;  // Blowfish
+      if (hash_mode ==  9300) kernel_threads = 32; // scrypt
       if (hash_mode ==  9700) kernel_threads = 64; // RC4
       if (hash_mode ==  9710) kernel_threads = 64; // RC4
       if (hash_mode ==  9800) kernel_threads = 64; // RC4
@@ -15140,6 +15108,9 @@ int main (int argc, char **argv)
           }
         }
 
+        device_param->kernel_accel_min = 1;
+        device_param->kernel_accel_max = 8;
+
         for (uint tmto = tmto_start; tmto < tmto_stop; tmto++)
         {
           // TODO: in theory the following calculation needs to be done per salt, not global
@@ -15149,7 +15120,7 @@ int main (int argc, char **argv)
 
           size_scryptV /= 1 << tmto;
 
-          size_scryptV *= device_processors * device_processor_cores;
+          size_scryptV *= device_param->device_processors * device_param->kernel_threads * device_param->kernel_accel_max;
 
           if (size_scryptV > device_param->device_maxmem_alloc)
           {
@@ -15161,7 +15132,7 @@ int main (int argc, char **argv)
           for (uint salts_pos = 0; salts_pos < data.salts_cnt; salts_pos++)
           {
             data.salts_buf[salts_pos].scrypt_tmto = tmto;
-            data.salts_buf[salts_pos].scrypt_phy  = device_processors * device_processor_cores;
+            data.salts_buf[salts_pos].scrypt_phy  = device_param->device_processors * device_param->kernel_threads * device_param->kernel_accel_max;
           }
 
           break;
@@ -15491,7 +15462,7 @@ int main (int argc, char **argv)
 
       char build_opts_new[1024] = { 0 };
 
-      snprintf (build_opts_new, sizeof (build_opts_new) - 1, "%s -D VENDOR_ID=%u -D CUDA_ARCH=%d -D VECT_SIZE=%u -D DEVICE_TYPE=%u -D KERN_TYPE=%u -D _unroll -cl-std=CL1.1", build_opts, device_param->device_vendor_id, (device_param->sm_major * 100) + device_param->sm_minor, device_param->vector_width, (u32) device_param->device_type, kern_type);
+      snprintf (build_opts_new, sizeof (build_opts_new) - 1, "%s -D VENDOR_ID=%u -D CUDA_ARCH=%d -D VECT_SIZE=%u -D DEVICE_TYPE=%u -D DGST_R0=%u -D DGST_R1=%u -D DGST_R2=%u -D DGST_R3=%u -D DGST_ELEM=%u -D KERN_TYPE=%u -D _unroll -cl-std=CL1.1", build_opts, device_param->device_vendor_id, (device_param->sm_major * 100) + device_param->sm_minor, device_param->vector_width, (u32) device_param->device_type, data.dgst_pos0, data.dgst_pos1, data.dgst_pos2, data.dgst_pos3, data.dgst_size / 4, kern_type);
 
       strncpy (build_opts, build_opts_new, sizeof (build_opts));
 
@@ -17232,7 +17203,7 @@ int main (int argc, char **argv)
      * status and monitor threads
      */
 
-    if ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+    if ((data.devices_status != STATUS_BYPASS) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
     {
       data.devices_status = STATUS_STARTING;
     }
@@ -17241,12 +17212,18 @@ int main (int argc, char **argv)
 
     hc_thread_t *inner_threads = (hc_thread_t *) mycalloc (10, sizeof (hc_thread_t));
 
+    data.shutdown_inner = 0;
+
     /**
       * Outfile remove
       */
 
     if (keyspace == 0 && benchmark == 0 && stdout_flag == 0)
     {
+      hc_thread_create (inner_threads[inner_threads_cnt], thread_monitor, NULL);
+
+      inner_threads_cnt++;
+
       if (outfile_check_timer != 0)
       {
         if (data.outfile_check_directory != NULL)
@@ -17559,7 +17536,7 @@ int main (int argc, char **argv)
 
         logfile_sub_msg ("START");
 
-        if ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+        if ((data.devices_status != STATUS_BYPASS) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
         {
           data.devices_status = STATUS_INIT;
         }
@@ -18090,7 +18067,7 @@ int main (int argc, char **argv)
 
         hc_thread_t *c_threads = (hc_thread_t *) mycalloc (data.devices_cnt, sizeof (hc_thread_t));
 
-        if ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+        if ((data.devices_status != STATUS_BYPASS) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
         {
           data.devices_status = STATUS_AUTOTUNE;
         }
@@ -18147,7 +18124,7 @@ int main (int argc, char **argv)
          * create cracker threads
          */
 
-        if ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+        if ((data.devices_status != STATUS_BYPASS) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
         {
           data.devices_status = STATUS_RUNNING;
         }
@@ -18199,6 +18176,11 @@ int main (int argc, char **argv)
 
         local_free (c_threads);
 
+        if ((data.devices_status != STATUS_BYPASS) && (data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
+        {
+          data.devices_status = STATUS_EXHAUSTED;
+        }
+
         logfile_sub_var_uint ("status-after-work", data.devices_status);
 
         data.restore = 0;
@@ -18265,6 +18247,16 @@ int main (int argc, char **argv)
           data.devices_status = STATUS_RUNNING;
         }
 
+        // and overwrite benchmark aborts as well
+
+        if (data.benchmark == 1)
+        {
+          if (data.devices_status == STATUS_ABORTED)
+          {
+            data.devices_status = STATUS_RUNNING;
+          }
+        }
+
         // finalize task
 
         if (data.devices_status == STATUS_CRACKED) break;
@@ -18328,12 +18320,9 @@ int main (int argc, char **argv)
       }
     }
 
-    // wait for non-interactive threads
+    // wait for inner threads
 
-    if ((data.devices_status != STATUS_CRACKED) && (data.devices_status != STATUS_ABORTED) && (data.devices_status != STATUS_QUIT))
-    {
-      data.devices_status = STATUS_EXHAUSTED;
-    }
+    data.shutdown_inner = 1;
 
     for (uint thread_idx = 0; thread_idx < inner_threads_cnt; thread_idx++)
     {
@@ -18435,7 +18424,7 @@ int main (int argc, char **argv)
     #ifdef HAVE_HWMON
     if (gpu_temp_disable == 0)
     {
-      if (gpu_temp_retain != 0) // VENDOR_ID_AMD is implied here
+      if (gpu_temp_retain != 0)
       {
         hc_thread_mutex_lock (mux_adl);
 
@@ -18474,7 +18463,7 @@ int main (int argc, char **argv)
 
     // reset power tuning
 
-    if (powertune_enable == 1) // VENDOR_ID_AMD is implied here
+    if (powertune_enable == 1)
     {
       hc_thread_mutex_lock (mux_adl);
 
@@ -18655,7 +18644,9 @@ int main (int argc, char **argv)
     if (data.devices_status == STATUS_QUIT) break;
   }
 
-  // wait for interactive threads
+  // wait for outer threads
+
+  data.shutdown_outer = 1;
 
   for (uint thread_idx = 0; thread_idx < outer_threads_cnt; thread_idx++)
   {