Fix in fan error handling

README.md

@@ -3,7 +3,7 @@
 This program allows you to read temperature sensors off of supported EVGA 30-series iCX3 video cards, as well as control the fans individually.
 
 ## Prerequisites
-A supported EVGA 30-series iCX3 card.  I have not done extensive testing but belive this is every model of their:
+A supported EVGA 30-series card with iCX3.  This includes:
 * RTX 3060 Ti
 * RTX 3070
 * RTX 3070 Ti
@@ -14,6 +14,8 @@ A supported EVGA 30-series iCX3 card.  I have not done extensive testing but bel
 
 The number of fans supported depends, of course, on your particular model.
 
+You must have the `i2c-dev` kernel module loaded with `modprobe i2c-dev`
+
 Access to the `/dev/i2c` device files, which means either:
 * Run as root, or
 * Install udev rules to allow user access. If you have the OpenRGB udev rules installed to control the LEDs you already have this set up.
@@ -39,6 +41,7 @@ Note that when controlling fans directly through iCX3 they will fall offline fro
 
 ```text
 Available options:
+--i2c N      : Only probe I2C bus N instead of all (may help with stuttering or freezing when probing I2C devices)
 --gpu N      : Control only GPU N instead of all supported cards
 --fan SPEED  : Set all fans at once to SPEED (see below)
 --fanN SPEED : Set fan N (0-3) to SPEED

evga-card.c

@@ -8,7 +8,7 @@
 
 /* Search all i2c device files for ones are on a PCI device of a supported GPU,
    and respond with the correct iCX3 version information */
-int find_evga_gpu_i2cs(struct card_info *infos, int max_gpus)
+int find_evga_gpu_i2cs(struct card_info *infos, int max_gpus, int i2c_bus)
 {
 	char i2c_devices_path[NAME_MAX];
 	char device_path[NAME_MAX];
@@ -20,6 +20,7 @@ int find_evga_gpu_i2cs(struct card_info *infos, int max_gpus)
 	struct dirent *ent;
 	
 	int num_gpus = 0;
+	int current_i2c_bus = -1;
 	unsigned short pci_vendor, pci_device, pci_subsystem_vendor, pci_subsystem_device = 0;
 	
 	/* Start looking for I2C adapters in /sys/bus/i2c/devices/ */
@@ -37,6 +38,14 @@ int find_evga_gpu_i2cs(struct card_info *infos, int max_gpus)
 		if(strncmp(ent->d_name, "i2c-", 4) != 0) 
 			continue;
 		
+		/* Only probe the specific device given (if provided) */
+		if (i2c_bus >= 0) {
+			sscanf(ent->d_name, "i2c-%i", &current_i2c_bus);
+			if (current_i2c_bus != i2c_bus)
+				continue;
+		}
+		
+		
 		strcpy(device_path, i2c_devices_path);
 		strcat(device_path, ent->d_name);
 		

evga-card.h

@@ -91,6 +91,9 @@ struct card_info {
 	int i2c_fd;							/* File descriptor for the i2c device file, for re-use */
 	int product_id;						/* EVGA internal product ID, as reported by the iCX3 controller */
 	unsigned int bar0;					/* Address of the card's PCI base address register */
+	void *nvml_device;					/* Pointer to nvmlDevice_t for use in NVML calls */
+	void *vram_addr;					/* Memory mapping for GDDR6 temps */
+	void *hotspot_addr;					/* Memory mapping for hotspot temperature */
 };
 
 struct gpu_pci_info {
@@ -162,7 +165,7 @@ static struct gpu_pci_info evga_pci_ids[] =
 	{"EVGA GeForce RTX 3090 Ti FTW3 Ultra Gaming"          , NVIDIA_VEN,   NVIDIA_RTX3090TI_DEV,       EVGA_SUB_VEN,   EVGA_RTX3090TI_FTW3_ULTRA_GAMING_SUB_DEV      }
 };
 
-int find_evga_gpu_i2cs(struct card_info *infos, int max_gpus);
+int find_evga_gpu_i2cs(struct card_info *infos, int max_gpus, int i2c_bus);
 unsigned short read_pci_id(char *device_path, char *field);
 char *read_nvidia_pci_address(char *device_path);
 

evga-icx.c

@@ -24,7 +24,8 @@ char *header_start = "";
 char *header_end = "";
 
 static const char helpstring[] = "Available options:\n"
-                                 "--gpu N      : Control only GPU N instead of all supported cards\n"
+								 "--i2c N      : Only probe I2C bus N instead of all (may help with stuttering or freezing when probing I2C devices)\n"
+								 "--gpu N      : Control only GPU N instead of all supported cards\n"
 								 "--fan SPEED  : Set all fans at once to SPEED (see below)\n"
 								 "--fanN SPEED : Set fan N (0-3) to SPEED\n"
 								 "    SPEED may be one of the following:\n"
@@ -47,13 +48,22 @@ int main (int argc, char **argv)
 	int print_info = 0;
 	int compact = 0;
 	int gpu_num = -1;    /* Card to control */
+	int i2c_bus = -1;
 	int overwrite = 0;
 	unsigned int watch = 0;
 	char *fan_speed[ICX3_MAX_FANS] = {NULL}; 
 	
 	/* Input parsing */
 	for (int i = 1; i < argc; i++){
-		if (strcmp(argv[i], "--gpu") == 0) {
+		if (strcmp(argv[i], "--i2c") == 0) {
+			i++;
+			if (i < argc) {
+				i2c_bus = atoi(argv[i]);
+			} else {
+				printf(helpstring);
+				return -1;
+			}
+		} else if (strcmp(argv[i], "--gpu") == 0) {
 			i++;
 			if (i < argc) {
 				gpu_num = atoi(argv[i]);
@@ -119,7 +129,7 @@ int main (int argc, char **argv)
 	if (overwrite && !compact)
 		overwrite = 0;
 	
-	gpu_count = find_evga_gpu_i2cs(gpus, MAX_GPUS);
+	gpu_count = find_evga_gpu_i2cs(gpus, MAX_GPUS, i2c_bus);
 
 	if (gpu_count == -1) {
 		printf("Error scanning I2C devices\n");
@@ -153,6 +163,8 @@ int main (int argc, char **argv)
 	/* NVML init */
 #ifdef USE_NVML
     init_nvml();
+	for (int i = 0; i < gpu_count; i++)
+		get_nvml_handle(&gpus[i]);
 #endif
 
 	/* PCI init for VRAM/hotspot temps */
@@ -164,7 +176,8 @@ int main (int argc, char **argv)
 	/* print sensor info */
 	if (print_info) {
 		do {
-			printf("\x1b[K");	/* Clear current console line (really just for overwrite mode) */
+			if (overwrite)
+				printf("\x1b[K");	/* Clear current console line */
 			
 			if (gpu_num == -1) {
 				/* No GPU specified on command line, loop over all supported GPUs */

gddr6.c

@@ -36,70 +36,69 @@ void init_gddr6(struct card_info *card)
 	card->bar0 = (pci_dev->base_addr[0] & 0xFFFFFFFF);
 
 	pci_cleanup(pacc);
-}
-
-float get_vram_temp(struct card_info *card)
-{
-	int fd;
-	float temp = 0.0;
 	
+	/* Open our memory mappings */
+	card->vram_addr = NULL;
+	card->hotspot_addr = NULL;
+	
+	int fd;
 	if ((fd = open("/dev/mem", O_RDWR | O_SYNC)) == -1)
     {
-        printf("Can't read memory. If you are root, enable kernel parameter iomem=relaxed\n");
+        printf("Can't read memory for VRAM and Hotspot temperatures. If you are root, enable kernel parameter iomem=relaxed\n");
-		return 0;
+		return;
     }
 	
+	unsigned int phys_addr, base_offset;
+	void *map_base;
+	
 	for (int i = 0; i < sizeof(device_offset_info) / sizeof(struct device_offset); i++) {
 		if (card->pci_device_id == device_offset_info[i].device_id){
-			unsigned int phys_addr = (card->bar0 + device_offset_info[i].vram_offset);
+			/* Map for VRAM */
-			unsigned int base_offset = phys_addr & ~(PG_SZ-1);
+			phys_addr = (card->bar0 + device_offset_info[i].vram_offset);
-			void *map_base = mmap(0, PG_SZ, PROT_READ, MAP_SHARED, fd, base_offset);
+			base_offset = phys_addr & ~(PG_SZ-1);
+			map_base = mmap(0, PG_SZ, PROT_READ, MAP_SHARED, fd, base_offset);
 			if(map_base == (void *) -1)
-			{
+				printf("Can't map memory for VRAM temperature. If you are root, enable kernel parameter iomem=relaxed\n");
-				if (fd != -1)
+			else
-					close(fd);
+				card->vram_addr = (void *) map_base + (phys_addr - base_offset);
-				printf("Can't read memory for VRAM temperature. If you are root, enable kernel parameter iomem=relaxed\n");
+			
-			}
+			
-			void *virt_addr = (char *) map_base + (phys_addr - base_offset);
+			/* Map for hotspot */
-			int read_result = *((unsigned int *) virt_addr);
+			phys_addr = (card->bar0 + device_offset_info[i].hotspot_offset);
-			temp = ((read_result & 0x00000fff) / 0x20);
+			base_offset = phys_addr & ~(PG_SZ-1);
-			munmap(map_base, PG_SZ);
+			map_base = mmap(0, PG_SZ, PROT_READ, MAP_SHARED, fd, base_offset);
+			if(map_base == (void *) -1)
+				printf("Can't map memory for Hotspot temperature. If you are root, enable kernel parameter iomem=relaxed\n");
+			else
+				card->hotspot_addr = (void *) map_base + (phys_addr - base_offset);
 		}
 	}
 	
 	close(fd);
+}
+
+float get_vram_temp(struct card_info *card)
+{
+	float temp = 0.0;
+	
+	if(card->vram_addr == NULL)
+		return 0.0;
+
+	int read_result = *((unsigned int *) card->vram_addr);
+	temp = ((read_result & 0x00000fff) / 0x20);
+
 	return temp;
 }
 
 float get_hotspot_temp(struct card_info *card)
 {
-	int fd;
 	float temp = 0.0;
 	
-	if ((fd = open("/dev/mem", O_RDWR | O_SYNC)) == -1)
+	if(card->hotspot_addr == NULL)
-    {
+		return 0.0;
-        printf("Can't read memory. If you are root, enable kernel parameter iomem=relaxed\n");
+
-		return 0;
+	int read_result = *((unsigned int *) card->hotspot_addr);
-    }
+	temp = (read_result >> 8) & 0xff;
-	
+
-	for (int i = 0; i < sizeof(device_offset_info) / sizeof(struct device_offset); i++) {
-		if (card->pci_device_id == device_offset_info[i].device_id){
-			unsigned int phys_addr = (card->bar0 + device_offset_info[i].hotspot_offset);
-			unsigned int base_offset = phys_addr & ~(PG_SZ-1);
-			void *map_base = mmap(0, PG_SZ, PROT_READ, MAP_SHARED, fd, base_offset);
-			if(map_base == (void *) -1)
-			{
-				if (fd != -1)
-					close(fd);
-				printf("Can't read memory for hotspot. If you are root, enable kernel parameter iomem=relaxed\n");
-			}
-			void *virt_addr = (char *) map_base + (phys_addr - base_offset);
-			int read_result = *((unsigned int *) virt_addr);
-			temp = (read_result >> 8) & 0xff;
-			munmap(map_base, PG_SZ);
-		}
-	}
-	
-	close(fd);
 	return temp;
 }

icx3.c

@@ -237,6 +237,6 @@ void set_fan(int fan, char *setting, struct card_info *card)
 	if (fan_readback.fanmode    != fan_control.fanmode ||
 		fan_readback.rpm_offset != fan_control.rpm_offset ||
 		fan_readback.duty       != fan_control.duty)
-		printf("Error setting fan %d on %s\n", fan, card->i2c_fd);
+		printf("Error setting fan %d on %s\n", fan, card->i2c_dev_path);
 }
 

nvidia-sensors.c

@@ -10,15 +10,14 @@ void init_nvml()
 		printf("Could not init NVML: %s\n", nvmlErrorString(result));
 }
 
-int get_nvml_handle(nvmlDevice_t *device, struct card_info *card)
+void get_nvml_handle(struct card_info *card)
 {
 	nvmlReturn_t result;
-	result = nvmlDeviceGetHandleByPciBusId_v2(card->pci_id, device);
+	result = nvmlDeviceGetHandleByPciBusId_v2(card->pci_id, card->nvml_device);
 	if (result != NVML_SUCCESS) {
-		printf("Failed to get device handle for card at %s: %s\n", card->pci_id, nvmlErrorString(result));
+		printf("Failed to get NVML device handle for card at %s: %s\n", card->pci_id, nvmlErrorString(result));
-		return 0;
+		card->nvml_device = NULL;
 	}
-	return 1;
 }
 
 void print_nvml_clock_reason(int compact, struct card_info *card)
@@ -52,12 +51,11 @@ void print_nvml_clock_reason(int compact, struct card_info *card)
 
 unsigned int get_nvml_temp(struct card_info *card)
 {
-	nvmlDevice_t nvml_device;
+	if (card->nvml_device == NULL)
-	if (!get_nvml_handle(&nvml_device, card))
 		return 0;
 	
 	unsigned int temp;
-	nvmlReturn_t result = nvmlDeviceGetTemperature(nvml_device, NVML_TEMPERATURE_GPU, &temp); 
+	nvmlReturn_t result = nvmlDeviceGetTemperature(*(nvmlDevice_t*)(card->nvml_device), NVML_TEMPERATURE_GPU, &temp); 
 	if (result != NVML_SUCCESS) {
 		printf("Failed to get temperature for card at %s: %s\n", card->pci_id, nvmlErrorString(result));
 		return 0;
@@ -67,12 +65,11 @@ unsigned int get_nvml_temp(struct card_info *card)
 
 unsigned long long get_nvml_clock_reasons(struct card_info *card)
 {
-	nvmlDevice_t nvml_device;
+	if (card->nvml_device == NULL)
-	if (!get_nvml_handle(&nvml_device, card))
 		return 0;
 	
 	unsigned long long reasons;
-	nvmlReturn_t result = nvmlDeviceGetCurrentClocksEventReasons(nvml_device, &reasons) ; 
+	nvmlReturn_t result = nvmlDeviceGetCurrentClocksEventReasons(*(nvmlDevice_t*)(card->nvml_device), &reasons) ; 
 	if (result != NVML_SUCCESS) {
 		printf("Failed to get clock reasons for card at %s: %s\n", card->pci_id, nvmlErrorString(result));
 		return 0;
@@ -83,12 +80,11 @@ unsigned long long get_nvml_clock_reasons(struct card_info *card)
 
 unsigned int get_nvml_mem_util(struct card_info *card)
 {
-	nvmlDevice_t nvml_device;
+	if (card->nvml_device == NULL)
-	if (!get_nvml_handle(&nvml_device, card))
 		return 0;
 	
 	nvmlUtilization_t util;
-	nvmlReturn_t result = nvmlDeviceGetUtilizationRates(nvml_device, &util);
+	nvmlReturn_t result = nvmlDeviceGetUtilizationRates(*(nvmlDevice_t*)(card->nvml_device), &util);
 	if (result != NVML_SUCCESS) {
 		printf("Failed to get clock reasons for card at %s: %s\n", card->pci_id, nvmlErrorString(result));
 		return 0;

nvidia-sensors.h

@@ -22,7 +22,7 @@ static struct clock_reason clock_reason_names[] =
 };
 
 void init_nvml();
-int get_nvml_handle(nvmlDevice_t *device, struct card_info *card);
+void get_nvml_handle(struct card_info *card);
 void print_nvml_clock_reason(int compact, struct card_info *card);
 unsigned int get_nvml_temp(struct card_info *card);
 unsigned long long get_nvml_clock_reasons(struct card_info *card);