You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
hashcat/src/ext_nvml.c

201 lines
4.8 KiB

/**
* Authors.....: Jens Steube <jens.steube@gmail.com>
* Gabriele Gristina <matrix@hashcat.net>
*
* License.....: MIT
*/
#include <ext_nvml.h>
int nvml_init (NVML_PTR *nvml)
{
if (!nvml) return (-1);
memset (nvml, 0, sizeof (NVML_PTR));
nvml->lib = hc_dlopen ("libnvidia-ml.so", RTLD_NOW);
if (!nvml->lib)
{
//if (data.quiet == 0)
// log_info ("WARNING: load NVML library failed, proceed without NVML HWMon enabled.");
return (-1);
}
HC_LOAD_FUNC(nvml, nvmlErrorString, NVML_ERROR_STRING, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlInit, NVML_INIT, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlShutdown, NVML_SHUTDOWN, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlDeviceGetName, NVML_DEVICE_GET_NAME, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlDeviceGetHandleByIndex, NVML_DEVICE_GET_HANDLE_BY_INDEX, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlDeviceGetTemperature, NVML_DEVICE_GET_TEMPERATURE, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlDeviceGetFanSpeed, NVML_DEVICE_GET_FAN_SPEED, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlDeviceGetPowerUsage, NVML_DEVICE_GET_POWER_USAGE, NVML, 0)
HC_LOAD_FUNC(nvml, nvmlDeviceGetUtilizationRates, NVML_DEVICE_GET_UTILIZATION_RATES, NVML, 0)
return 0;
}
void nvml_close (NVML_PTR *nvml)
{
if (nvml)
{
if (nvml->lib)
hc_dlclose (nvml->lib);
myfree (nvml);
}
}
const char *hm_NVML_nvmlErrorString (NVML_PTR *nvml, nvmlReturn_t nvml_rc)
{
if (!nvml) return NULL;
return nvml->nvmlErrorString (nvml_rc);
}
nvmlReturn_t hm_NVML_nvmlInit (NVML_PTR *nvml)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlInit ();
if (nvml_rc != NVML_SUCCESS)
{
const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
log_info ("WARN: %s %d %s\n", "nvmlInit()", nvml_rc, string);
}
return nvml_rc;
}
nvmlReturn_t hm_NVML_nvmlShutdown (NVML_PTR *nvml)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlShutdown ();
if (nvml_rc != NVML_SUCCESS)
{
const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
log_info ("WARN: %s %d %s\n", "nvmlShutdown()", nvml_rc, string);
}
return nvml_rc;
}
nvmlReturn_t hm_NVML_nvmlDeviceGetName (NVML_PTR *nvml, nvmlDevice_t device, char *name, unsigned int length)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlDeviceGetName (device, name, length);
if (nvml_rc != NVML_SUCCESS)
{
const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
log_info ("WARN: %s %d %s\n", "nvmlDeviceGetName()", nvml_rc, string);
}
return nvml_rc;
}
nvmlReturn_t hm_NVML_nvmlDeviceGetHandleByIndex (NVML_PTR *nvml, int skip_warnings, unsigned int index, nvmlDevice_t *device)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlDeviceGetHandleByIndex (index, device);
if (nvml_rc != NVML_SUCCESS)
{
if (skip_warnings == 0)
{
const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
log_info ("WARN: %s %d %s\n", "nvmlDeviceGetHandleByIndex()", nvml_rc, string);
}
}
return nvml_rc;
}
nvmlReturn_t hm_NVML_nvmlDeviceGetTemperature (NVML_PTR *nvml, nvmlDevice_t device, nvmlTemperatureSensors_t sensorType, unsigned int *temp)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlDeviceGetTemperature (device, sensorType, temp);
if (nvml_rc != NVML_SUCCESS)
{
*temp = -1;
//const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
//log_info ("WARN: %s %d %s\n", "nvmlDeviceGetTemperature()", nvml_rc, string);
}
return nvml_rc;
}
nvmlReturn_t hm_NVML_nvmlDeviceGetFanSpeed (NVML_PTR *nvml, int skip_warnings, nvmlDevice_t device, unsigned int *speed)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlDeviceGetFanSpeed (device, speed);
if (nvml_rc != NVML_SUCCESS)
{
*speed = -1;
if (skip_warnings == 0)
{
const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
log_info ("WARN: %s %d %s\n", "nvmlDeviceGetFanSpeed()", nvml_rc, string);
}
}
return nvml_rc;
}
/* only tesla following */
nvmlReturn_t hm_NVML_nvmlDeviceGetPowerUsage (NVML_PTR *nvml, nvmlDevice_t device, unsigned int *power)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlDeviceGetPowerUsage (device, power);
if (nvml_rc != NVML_SUCCESS)
{
*power = -1;
//const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
//log_info ("WARN: %s %d %s\n", "nvmlDeviceGetPowerUsage()", nvml_rc, string);
}
return nvml_rc;
}
nvmlReturn_t hm_NVML_nvmlDeviceGetUtilizationRates (NVML_PTR *nvml, nvmlDevice_t device, nvmlUtilization_t *utilization)
{
if (!nvml) return -1;
nvmlReturn_t nvml_rc = nvml->nvmlDeviceGetUtilizationRates (device, utilization);
if (nvml_rc != NVML_SUCCESS)
{
utilization->gpu = -1;
utilization->memory = -1;
//const char *string = hm_NVML_nvmlErrorString (nvml, nvml_rc);
//log_info ("WARN: %s %d %s\n", "nvmlDeviceGetUtilizationRates()", nvml_rc, string);
}
return nvml_rc;
}