X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;ds=sidebyside;f=lib%2Flibrte_eal%2Flinuxapp%2Feal%2Feal_pci.c;h=3b94b6fb54fd487cb728741cbe515db0bdd46d55;hb=629395b063e8278a05ea41908d1152fa68df098c;hp=29f17287472be3b6217b910032b4dba5617b030a;hpb=46a6fa87931a46153e4ed987e55e750607e03246;p=dpdk.git

diff --git a/lib/librte_eal/linuxapp/eal/eal_pci.c b/lib/librte_eal/linuxapp/eal/eal_pci.c
index 29f1728747..3b94b6fb54 100644
--- a/lib/librte_eal/linuxapp/eal/eal_pci.c
+++ b/lib/librte_eal/linuxapp/eal/eal_pci.c
@@ -32,8 +32,6 @@
  */
 
 #include <string.h>
-#include <sys/stat.h>
-#include <fcntl.h>
 #include <dirent.h>
 #include <sys/mman.h>
 
@@ -47,6 +45,7 @@
 #include "rte_pci_dev_ids.h"
 #include "eal_filesystem.h"
 #include "eal_private.h"
+#include "eal_pci_init.h"
 
 /**
  * @file
@@ -57,30 +56,7 @@
  * IGB_UIO driver (or doesn't initialize, if the device wasn't bound to it).
  */
 
-struct pci_map {
-	void *addr;
-	uint64_t offset;
-	uint64_t size;
-	uint64_t phaddr;
-};
-
-/*
- * For multi-process we need to reproduce all PCI mappings in secondary
- * processes, so save them in a tailq.
- */
-struct mapped_pci_resource {
-	TAILQ_ENTRY(mapped_pci_resource) next;
-
-	struct rte_pci_addr pci_addr;
-	char path[PATH_MAX];
-	int nb_maps;
-	struct pci_map maps[PCI_MAX_RESOURCE];
-};
-
-TAILQ_HEAD(mapped_pci_res_list, mapped_pci_resource);
-static struct mapped_pci_res_list *pci_res_list;
-
-static int pci_parse_sysfs_value(const char *filename, uint64_t *val);
+struct mapped_pci_res_list *pci_res_list = NULL;
 
 /* unbind kernel driver for this device */
 static int
@@ -122,7 +98,7 @@ error:
 }
 
 /* map a particular resource from a file */
-static void *
+void *
 pci_map_resource(void *requested_addr, int fd, off_t offset, size_t size)
 {
 	void *mapaddr;
@@ -147,342 +123,6 @@ fail:
 	return NULL;
 }
 
-#define OFF_MAX              ((uint64_t)(off_t)-1)
-static int
-pci_uio_get_mappings(const char *devname, struct pci_map maps[], int nb_maps)
-{
-	int i;
-	char dirname[PATH_MAX];
-	char filename[PATH_MAX];
-	uint64_t offset, size;
-
-	for (i = 0; i != nb_maps; i++) {
-
-		/* check if map directory exists */
-		rte_snprintf(dirname, sizeof(dirname),
-			"%s/maps/map%u", devname, i);
-
-		if (access(dirname, F_OK) != 0)
-			break;
-
-		/* get mapping offset */
-		rte_snprintf(filename, sizeof(filename),
-			"%s/offset", dirname);
-		if (pci_parse_sysfs_value(filename, &offset) < 0) {
-			RTE_LOG(ERR, EAL,
-				"%s(): cannot parse offset of %s\n",
-				__func__, dirname);
-			return (-1);
-		}
-
-		/* get mapping size */
-		rte_snprintf(filename, sizeof(filename),
-			"%s/size", dirname);
-		if (pci_parse_sysfs_value(filename, &size) < 0) {
-			RTE_LOG(ERR, EAL,
-				"%s(): cannot parse size of %s\n",
-				__func__, dirname);
-			return (-1);
-		}
-
-		/* get mapping physical address */
-		rte_snprintf(filename, sizeof(filename),
-			"%s/addr", dirname);
-		if (pci_parse_sysfs_value(filename, &maps[i].phaddr) < 0) {
-			RTE_LOG(ERR, EAL,
-				"%s(): cannot parse addr of %s\n",
-				__func__, dirname);
-			return (-1);
-		}
-
-		if ((offset > OFF_MAX) || (size > SIZE_MAX)) {
-			RTE_LOG(ERR, EAL,
-				"%s(): offset/size exceed system max value\n",
-				__func__);
-			return (-1);
-		}
-
-		maps[i].offset = offset;
-		maps[i].size = size;
-        }
-	return (i);
-}
-
-static int
-pci_uio_map_secondary(struct rte_pci_device *dev)
-{
-	int fd, i;
-	struct mapped_pci_resource *uio_res;
-
-	TAILQ_FOREACH(uio_res, pci_res_list, next) {
-
-		/* skip this element if it doesn't match our PCI address */
-		if (memcmp(&uio_res->pci_addr, &dev->addr, sizeof(dev->addr)))
-			continue;
-
-		for (i = 0; i != uio_res->nb_maps; i++) {
-			/*
-			 * open devname, to mmap it
-			 */
-			fd = open(uio_res->path, O_RDWR);
-			if (fd < 0) {
-				RTE_LOG(ERR, EAL, "Cannot open %s: %s\n",
-					uio_res->path, strerror(errno));
-				return -1;
-			}
-
-			if (pci_map_resource(uio_res->maps[i].addr, fd,
-					     (off_t)uio_res->maps[i].offset,
-					     (size_t)uio_res->maps[i].size)
-			    != uio_res->maps[i].addr) {
-				RTE_LOG(ERR, EAL,
-					"Cannot mmap device resource\n");
-				close(fd);
-				return (-1);
-			}
-			/* fd is not needed in slave process, close it */
-			close(fd);
-		}
-		return (0);
-	}
-
-	RTE_LOG(ERR, EAL, "Cannot find resource for device\n");
-	return -1;
-}
-
-static int
-pci_mknod_uio_dev(const char *sysfs_uio_path, unsigned uio_num)
-{
-	FILE *f;
-	char filename[PATH_MAX];
-	int ret;
-	unsigned major, minor;
-	dev_t dev;
-
-	/* get the name of the sysfs file that contains the major and minor
-	 * of the uio device and read its content */
-	rte_snprintf(filename, sizeof(filename), "%s/dev", sysfs_uio_path);
-
-	f = fopen(filename, "r");
-	if (f == NULL) {
-		RTE_LOG(ERR, EAL, "%s(): cannot open sysfs to get major:minor\n",
-			__func__);
-		return -1;
-	}
-
-	ret = fscanf(f, "%d:%d", &major, &minor);
-	if (ret != 2) {
-		RTE_LOG(ERR, EAL, "%s(): cannot parse sysfs to get major:minor\n",
-			__func__);
-		fclose(f);
-		return -1;
-	}
-	fclose(f);
-
-	/* create the char device "mknod /dev/uioX c major minor" */
-	rte_snprintf(filename, sizeof(filename), "/dev/uio%u", uio_num);
-	dev = makedev(major, minor);
-	ret = mknod(filename, S_IFCHR | S_IRUSR | S_IWUSR, dev);
-	if (f == NULL) {
-		RTE_LOG(ERR, EAL, "%s(): mknod() failed %s\n",
-			__func__, strerror(errno));
-		return -1;
-	}
-
-	return ret;
-}
-
-/*
- * Return the uioX char device used for a pci device. On success, return
- * the UIO number and fill dstbuf string with the path of the device in
- * sysfs. On error, return a negative value. In this case dstbuf is
- * invalid.
- */
-static int
-pci_get_uio_dev(struct rte_pci_device *dev, char *dstbuf,
-			   unsigned int buflen)
-{
-	struct rte_pci_addr *loc = &dev->addr;
-	unsigned int uio_num;
-	struct dirent *e;
-	DIR *dir;
-	char dirname[PATH_MAX];
-
-	/* depending on kernel version, uio can be located in uio/uioX
-	 * or uio:uioX */
-
-	rte_snprintf(dirname, sizeof(dirname),
-	         SYSFS_PCI_DEVICES "/" PCI_PRI_FMT "/uio",
-	         loc->domain, loc->bus, loc->devid, loc->function);
-
-	dir = opendir(dirname);
-	if (dir == NULL) {
-		/* retry with the parent directory */
-		rte_snprintf(dirname, sizeof(dirname),
-		         SYSFS_PCI_DEVICES "/" PCI_PRI_FMT,
-		         loc->domain, loc->bus, loc->devid, loc->function);
-		dir = opendir(dirname);
-
-		if (dir == NULL) {
-			RTE_LOG(ERR, EAL, "Cannot opendir %s\n", dirname);
-			return -1;
-		}
-	}
-
-	/* take the first file starting with "uio" */
-	while ((e = readdir(dir)) != NULL) {
-		/* format could be uio%d ...*/
-		int shortprefix_len = sizeof("uio") - 1;
-		/* ... or uio:uio%d */
-		int longprefix_len = sizeof("uio:uio") - 1;
-		char *endptr;
-
-		if (strncmp(e->d_name, "uio", 3) != 0)
-			continue;
-
-		/* first try uio%d */
-		errno = 0;
-		uio_num = strtoull(e->d_name + shortprefix_len, &endptr, 10);
-		if (errno == 0 && endptr != (e->d_name + shortprefix_len)) {
-			rte_snprintf(dstbuf, buflen, "%s/uio%u", dirname, uio_num);
-			break;
-		}
-
-		/* then try uio:uio%d */
-		errno = 0;
-		uio_num = strtoull(e->d_name + longprefix_len, &endptr, 10);
-		if (errno == 0 && endptr != (e->d_name + longprefix_len)) {
-			rte_snprintf(dstbuf, buflen, "%s/uio:uio%u", dirname, uio_num);
-			break;
-		}
-	}
-	closedir(dir);
-
-	/* No uio resource found */
-	if (e == NULL)
-		return -1;
-
-	/* create uio device if we've been asked to */
-	if (internal_config.create_uio_dev && pci_mknod_uio_dev(dstbuf, uio_num) < 0)
-		RTE_LOG(WARNING, EAL, "Cannot create /dev/uio%u\n", uio_num);
-
-	return uio_num;
-}
-
-/* map the PCI resource of a PCI device in virtual memory */
-static int
-pci_uio_map_resource(struct rte_pci_device *dev)
-{
-	int i, j;
-	char dirname[PATH_MAX];
-	char devname[PATH_MAX]; /* contains the /dev/uioX */
-	void *mapaddr;
-	int uio_num;
-	uint64_t phaddr;
-	uint64_t offset;
-	uint64_t pagesz;
-	int nb_maps;
-	struct rte_pci_addr *loc = &dev->addr;
-	struct mapped_pci_resource *uio_res;
-	struct pci_map *maps;
-
-	dev->intr_handle.fd = -1;
-	dev->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN;
-
-	/* secondary processes - use already recorded details */
-	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
-		return (pci_uio_map_secondary(dev));
-
-	/* find uio resource */
-	uio_num = pci_get_uio_dev(dev, dirname, sizeof(dirname));
-	if (uio_num < 0) {
-		RTE_LOG(WARNING, EAL, "  "PCI_PRI_FMT" not managed by UIO driver, "
-				"skipping\n", loc->domain, loc->bus, loc->devid, loc->function);
-		return -1;
-	}
-	rte_snprintf(devname, sizeof(devname), "/dev/uio%u", uio_num);
-
-	/* save fd if in primary process */
-	dev->intr_handle.fd = open(devname, O_RDWR);
-	if (dev->intr_handle.fd < 0) {
-		RTE_LOG(ERR, EAL, "Cannot open %s: %s\n",
-			devname, strerror(errno));
-		return -1;
-	}
-	dev->intr_handle.type = RTE_INTR_HANDLE_UIO;
-
-	/* allocate the mapping details for secondary processes*/
-	if ((uio_res = rte_zmalloc("UIO_RES", sizeof (*uio_res), 0)) == NULL) {
-		RTE_LOG(ERR, EAL,
-			"%s(): cannot store uio mmap details\n", __func__);
-		return (-1);
-	}
-
-	rte_snprintf(uio_res->path, sizeof(uio_res->path), "%s", devname);
-	memcpy(&uio_res->pci_addr, &dev->addr, sizeof(uio_res->pci_addr));
-
-	/* collect info about device mappings */
-	nb_maps = pci_uio_get_mappings(dirname, uio_res->maps,
-				       RTE_DIM(uio_res->maps));
-	if (nb_maps < 0) {
-		rte_free(uio_res);
-		return (nb_maps);
-	}
-
-	uio_res->nb_maps = nb_maps;
-
-	/* Map all BARs */
-	pagesz = sysconf(_SC_PAGESIZE);
-
-	maps = uio_res->maps;
-	for (i = 0; i != PCI_MAX_RESOURCE; i++) {
-		int fd;
-
-		/* skip empty BAR */
-		if ((phaddr = dev->mem_resource[i].phys_addr) == 0)
-			continue;
-
-		for (j = 0; j != nb_maps && (phaddr != maps[j].phaddr ||
-				dev->mem_resource[i].len != maps[j].size);
-				j++)
-			;
-
-		/* if matching map is found, then use it */
-		if (j != nb_maps) {
-			offset = j * pagesz;
-
-			/*
-			 * open devname, to mmap it
-			 */
-			fd = open(devname, O_RDWR);
-			if (fd < 0) {
-				RTE_LOG(ERR, EAL, "Cannot open %s: %s\n",
-					devname, strerror(errno));
-				return -1;
-			}
-
-			if (maps[j].addr != NULL ||
-			    (mapaddr = pci_map_resource(NULL, fd,
-							(off_t)offset,
-							(size_t)maps[j].size)
-			    ) == NULL) {
-				rte_free(uio_res);
-				close(fd);
-				return (-1);
-			}
-			close(fd);
-
-			maps[j].addr = mapaddr;
-			maps[j].offset = offset;
-			dev->mem_resource[i].addr = mapaddr;
-		}
-	}
-
-	TAILQ_INSERT_TAIL(pci_res_list, uio_res, next);
-
-	return (0);
-}
-
 /* parse the "resource" sysfs file */
 #define IORESOURCE_MEM  0x00000200
 
@@ -546,41 +186,6 @@ error:
 	return -1;
 }
 
-/*
- * parse a sysfs file containing one integer value
- * different to the eal version, as it needs to work with 64-bit values
- */
-static int
-pci_parse_sysfs_value(const char *filename, uint64_t *val)
-{
-        FILE *f;
-        char buf[BUFSIZ];
-        char *end = NULL;
-
-        f = fopen(filename, "r");
-        if (f == NULL) {
-                RTE_LOG(ERR, EAL, "%s(): cannot open sysfs value %s\n",
-                        __func__, filename);
-                return -1;
-        }
-
-        if (fgets(buf, sizeof(buf), f) == NULL) {
-                RTE_LOG(ERR, EAL, "%s(): cannot read sysfs value %s\n",
-                        __func__, filename);
-                fclose(f);
-                return -1;
-        }
-        *val = strtoull(buf, &end, 0);
-        if ((buf[0] == '\0') || (end == NULL) || (*end != '\n')) {
-                RTE_LOG(ERR, EAL, "%s(): cannot parse sysfs value %s\n",
-                                __func__, filename);
-                fclose(f);
-                return -1;
-        }
-        fclose(f);
-        return 0;
-}
-
 /* Compare two PCI device addresses. */
 static int
 pci_addr_comparison(struct rte_pci_addr *addr, struct rte_pci_addr *addr2)
@@ -788,6 +393,30 @@ error:
 	return -1;
 }
 
+static int
+pci_map_device(struct rte_pci_device *dev)
+{
+	int ret, mapped = 0;
+
+	/* try mapping the NIC resources using VFIO if it exists */
+#ifdef VFIO_PRESENT
+	if (pci_vfio_is_enabled()) {
+		ret = pci_vfio_map_resource(dev);
+		if (ret == 0)
+			mapped = 1;
+		else if (ret < 0)
+			return ret;
+	}
+#endif
+	/* map resources for devices that use igb_uio */
+	if (!mapped) {
+		ret = pci_uio_map_resource(dev);
+		if (ret != 0)
+			return ret;
+	}
+	return 0;
+}
+
 /*
  * If vendor/device ID match, call the devinit() function of the
  * driver.
@@ -795,6 +424,7 @@ error:
 int
 rte_eal_pci_probe_one_driver(struct rte_pci_driver *dr, struct rte_pci_device *dev)
 {
+	int ret;
 	struct rte_pci_id *id_table;
 
 	for (id_table = dr->id_table ; id_table->vendor_id != 0; id_table++) {
@@ -826,13 +456,14 @@ rte_eal_pci_probe_one_driver(struct rte_pci_driver *dr, struct rte_pci_device *d
 		if (dev->devargs != NULL &&
 			dev->devargs->type == RTE_DEVTYPE_BLACKLISTED_PCI) {
 			RTE_LOG(DEBUG, EAL, "  Device is blacklisted, not initializing\n");
-			return 0;
+			return 1;
 		}
 
-		if (dr->drv_flags & RTE_PCI_DRV_NEED_IGB_UIO) {
+		if (dr->drv_flags & RTE_PCI_DRV_NEED_MAPPING) {
 			/* map resources for devices that use igb_uio */
-			if (pci_uio_map_resource(dev) < 0)
-				return -1;
+			ret = pci_map_device(dev);
+			if (ret != 0)
+				return ret;
 		} else if (dr->drv_flags & RTE_PCI_DRV_FORCE_UNBIND &&
 		           rte_eal_process_type() == RTE_PROC_PRIMARY) {
 			/* unbind current driver */
@@ -867,5 +498,21 @@ rte_eal_pci_init(void)
 		RTE_LOG(ERR, EAL, "%s(): Cannot scan PCI bus\n", __func__);
 		return -1;
 	}
+#ifdef VFIO_PRESENT
+	pci_vfio_enable();
+
+	if (pci_vfio_is_enabled()) {
+
+		/* if we are primary process, create a thread to communicate with
+		 * secondary processes. the thread will use a socket to wait for
+		 * requests from secondary process to send open file descriptors,
+		 * because VFIO does not allow multiple open descriptors on a group or
+		 * VFIO container.
+		 */
+		if (internal_config.process_type == RTE_PROC_PRIMARY &&
+				pci_vfio_mp_sync_setup() < 0)
+			return -1;
+	}
+#endif
 	return 0;
 }