X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;f=lib%2Flibrte_eal%2Flinuxapp%2Feal%2Feal_pci.c;h=62da4d430a21177d30301ebef519ef0935d3c94d;hb=affe1cdc009cd35a8122d8b1d5975f445d8f3d7a;hp=af9415dd15b710891c2357a86e495777c0a4f486;hpb=148f963fb5323c1c6b6d5cea95084deb25cc73f8;p=dpdk.git diff --git a/lib/librte_eal/linuxapp/eal/eal_pci.c b/lib/librte_eal/linuxapp/eal/eal_pci.c index af9415dd15..62da4d430a 100644 --- a/lib/librte_eal/linuxapp/eal/eal_pci.c +++ b/lib/librte_eal/linuxapp/eal/eal_pci.c @@ -1,13 +1,13 @@ /*- * BSD LICENSE - * + * * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. * All rights reserved. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: - * + * * * Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * * Redistributions in binary form must reproduce the above copyright @@ -17,7 +17,7 @@ * * Neither the name of Intel Corporation nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. - * + * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR @@ -31,213 +31,31 @@ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ -#include -#include -#include #include -#include -#include -#include -#include -#include -#include -#include -#include #include -#include -#include -#include -#include -#include #include #include -#include -#include -#include -#include -#include -#include #include -#include -#include #include -#include -#include +#include +#include -#include "rte_pci_dev_ids.h" #include "eal_filesystem.h" #include "eal_private.h" +#include "eal_pci_init.h" /** * @file * PCI probing under linux * - * This code is used to simulate a PCI probe by parsing information in - * sysfs. Moreover, when a registered driver matches a device, the - * kernel driver currently using it is unloaded and replaced by - * igb_uio module, which is a very minimal userland driver for Intel - * network card, only providing access to PCI BAR to applications, and - * enabling bus master. - */ - -struct uio_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 uio_resource { - TAILQ_ENTRY(uio_resource) next; - - struct rte_pci_addr pci_addr; - char path[PATH_MAX]; - size_t nb_maps; - struct uio_map maps[PCI_MAX_RESOURCE]; -}; - -TAILQ_HEAD(uio_res_list, uio_resource); - -static struct uio_res_list *uio_res_list = NULL; -static int pci_parse_sysfs_value(const char *filename, uint64_t *val); - -/* forward prototype of function called in pci_switch_module below */ -static int pci_uio_map_resource(struct rte_pci_device *dev); - -#ifdef RTE_EAL_UNBIND_PORTS -#define PROC_MODULES "/proc/modules" - -#define IGB_UIO_NAME "igb_uio" - -#define UIO_DRV_PATH "/sys/bus/pci/drivers/%s" - -/* maximum time to wait that /dev/uioX appears */ -#define UIO_DEV_WAIT_TIMEOUT 3 /* seconds */ - -/* - * Check that a kernel module is loaded. Returns 0 on success, or if the - * parameter is NULL, or -1 if the module is not loaded. + * This code is used to simulate a PCI probe by parsing information in sysfs. + * When a registered device matches a driver, it is then initialized with + * IGB_UIO driver (or doesn't initialize, if the device wasn't bound to it). */ -static int -pci_uio_check_module(const char *module_name) -{ - FILE *f; - unsigned i; - char buf[BUFSIZ]; - - if (module_name == NULL) - return 0; - - f = fopen(PROC_MODULES, "r"); - if (f == NULL) { - RTE_LOG(ERR, EAL, "Cannot open "PROC_MODULES": %s\n", - strerror(errno)); - return -1; - } - - while(fgets(buf, sizeof(buf), f) != NULL) { - - for (i = 0; i < sizeof(buf) && buf[i] != '\0'; i++) { - if (isspace(buf[i])) - buf[i] = '\0'; - } - - if (strncmp(buf, module_name, sizeof(buf)) == 0) { - fclose(f); - return 0; - } - } - fclose(f); - return -1; -} - -/* bind a PCI to the kernel module driver */ -static int -pci_bind_device(struct rte_pci_device *dev, char dr_path[]) -{ - FILE *f; - int n; - char buf[BUFSIZ]; - char dev_bind[PATH_MAX]; - struct rte_pci_addr *loc = &dev->addr; - - n = rte_snprintf(dev_bind, sizeof(dev_bind), "%s/bind", dr_path); - if ((n < 0) || (n >= (int)sizeof(buf))) { - RTE_LOG(ERR, EAL, "Cannot rte_snprintf device bind path\n"); - return -1; - } - - f = fopen(dev_bind, "w"); - if (f == NULL) { - RTE_LOG(ERR, EAL, "Cannot open %s\n", dev_bind); - return -1; - } - n = rte_snprintf(buf, sizeof(buf), PCI_PRI_FMT "\n", - loc->domain, loc->bus, loc->devid, loc->function); - if ((n < 0) || (n >= (int)sizeof(buf))) { - RTE_LOG(ERR, EAL, "Cannot rte_snprintf PCI infos\n"); - fclose(f); - return -1; - } - if (fwrite(buf, n, 1, f) == 0) { - fclose(f); - return -1; - } - - fclose(f); - return 0; -} - -static int -pci_uio_bind_device(struct rte_pci_device *dev, const char *module_name) -{ - FILE *f; - int n; - char buf[BUFSIZ]; - char uio_newid[PATH_MAX]; - char uio_bind[PATH_MAX]; - - n = rte_snprintf(uio_newid, sizeof(uio_newid), UIO_DRV_PATH "/new_id", module_name); - if ((n < 0) || (n >= (int)sizeof(uio_newid))) { - RTE_LOG(ERR, EAL, "Cannot rte_snprintf uio_newid name\n"); - return -1; - } - - n = rte_snprintf(uio_bind, sizeof(uio_bind), UIO_DRV_PATH, module_name); - if ((n < 0) || (n >= (int)sizeof(uio_bind))) { - RTE_LOG(ERR, EAL, "Cannot rte_snprintf uio_bind name\n"); - return -1; - } - - n = rte_snprintf(buf, sizeof(buf), "%x %x\n", - dev->id.vendor_id, dev->id.device_id); - if ((n < 0) || (n >= (int)sizeof(buf))) { - RTE_LOG(ERR, EAL, "Cannot rte_snprintf vendor_id/device_id\n"); - return -1; - } - - f = fopen(uio_newid, "w"); - if (f == NULL) { - RTE_LOG(ERR, EAL, "Cannot open %s\n", uio_newid); - return -1; - } - if (fwrite(buf, n, 1, f) == 0) { - fclose(f); - return -1; - } - fclose(f); - - pci_bind_device(dev, uio_bind); - return 0; -} /* unbind kernel driver for this device */ -static int +int pci_unbind_kernel_driver(struct rte_pci_device *dev) { int n; @@ -247,18 +65,18 @@ pci_unbind_kernel_driver(struct rte_pci_device *dev) struct rte_pci_addr *loc = &dev->addr; /* open /sys/bus/pci/devices/AAAA:BB:CC.D/driver */ - rte_snprintf(filename, sizeof(filename), - SYSFS_PCI_DEVICES "/" PCI_PRI_FMT "/driver/unbind", - loc->domain, loc->bus, loc->devid, loc->function); + snprintf(filename, sizeof(filename), + "%s/" PCI_PRI_FMT "/driver/unbind", pci_get_sysfs_path(), + loc->domain, loc->bus, loc->devid, loc->function); f = fopen(filename, "w"); if (f == NULL) /* device was not bound */ return 0; - n = rte_snprintf(buf, sizeof(buf), PCI_PRI_FMT "\n", + n = snprintf(buf, sizeof(buf), PCI_PRI_FMT "\n", loc->domain, loc->bus, loc->devid, loc->function); if ((n < 0) || (n >= (int)sizeof(buf))) { - RTE_LOG(ERR, EAL, "%s(): rte_snprintf failed\n", __func__); + RTE_LOG(ERR, EAL, "%s(): snprintf failed\n", __func__); goto error; } if (fwrite(buf, n, 1, f) == 0) { @@ -275,375 +93,143 @@ error: return -1; } - static int -pci_switch_module(struct rte_pci_driver *dr, struct rte_pci_device *dev, - int uio_status, const char *module_name) +pci_get_kernel_driver_by_path(const char *filename, char *dri_name) { - if (rte_eal_process_type() == RTE_PROC_PRIMARY) { - /* check that our driver is loaded */ - if (uio_status != 0 && - (uio_status = pci_uio_check_module(module_name)) != 0) - rte_exit(EXIT_FAILURE, "The %s module is required by the " - "%s driver\n", module_name, dr->name); - - /* unbind current driver, bind ours */ - if (pci_unbind_kernel_driver(dev) < 0) - return -1; - if (pci_uio_bind_device(dev, module_name) < 0) - return -1; - } - /* map the NIC resources */ - if (pci_uio_map_resource(dev) < 0) + int count; + char path[PATH_MAX]; + char *name; + + if (!filename || !dri_name) return -1; - return 0; -} + count = readlink(filename, path, PATH_MAX); + if (count >= PATH_MAX) + return -1; -#endif /* ifdef EAL_UNBIND_PORTS */ + /* For device does not have a driver */ + if (count < 0) + return 1; -/* map a particular resource from a file */ -static void * -pci_map_resource(struct rte_pci_device *dev, void *requested_addr, - const char *devname, off_t offset, size_t size) -{ - int fd; - void *mapaddr; - -#ifdef RTE_EAL_UNBIND_PORTS - /* - * open devname, and mmap it: it can take some time to - * appear, so we wait some time before returning an error - */ - unsigned n; - fd = dev->intr_handle.fd; - for (n = 0; n < UIO_DEV_WAIT_TIMEOUT*10 && fd < 0; n++) { - errno = 0; - if ((fd = open(devname, O_RDWR)) < 0 && errno != ENOENT) - break; - usleep(100000); - } -#else - /* - * open devname, to mmap it - */ - fd = open(devname, O_RDWR); -#endif - if (fd < 0) { - RTE_LOG(ERR, EAL, "Cannot open %s: %s\n", - devname, strerror(errno)); - goto fail; - } + path[count] = '\0'; - /* Map the PCI memory resource of device */ - mapaddr = mmap(requested_addr, size, PROT_READ | PROT_WRITE, - MAP_SHARED, fd, offset); - if (mapaddr == MAP_FAILED || - (requested_addr != NULL && mapaddr != requested_addr)) { - RTE_LOG(ERR, EAL, "%s(): cannot mmap(%s(%d), %p, 0x%lx, 0x%lx):" - " %s (%p)\n", __func__, devname, fd, requested_addr, - (unsigned long)size, (unsigned long)offset, - strerror(errno), mapaddr); - close(fd); - goto fail; - } - if (rte_eal_process_type() == RTE_PROC_PRIMARY) { - /* save fd if in primary process */ - dev->intr_handle.fd = fd; - dev->intr_handle.type = RTE_INTR_HANDLE_UIO; - } else { - /* fd is not needed in slave process, close it */ - dev->intr_handle.fd = -1; - dev->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN; - close(fd); + name = strrchr(path, '/'); + if (name) { + strncpy(dri_name, name + 1, strlen(name + 1) + 1); + return 0; } - RTE_LOG(DEBUG, EAL, " PCI memory mapped at %p\n", mapaddr); - - return mapaddr; - -fail: - dev->intr_handle.fd = -1; - dev->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN; - - return NULL; + return -1; } -#define OFF_MAX ((uint64_t)(off_t)-1) -static ssize_t -pci_uio_get_mappings(const char *devname, struct uio_map maps[], size_t nb_maps) +/* Map pci device */ +int +rte_eal_pci_map_device(struct rte_pci_device *dev) { - size_t 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); + int ret = -1; + + /* try mapping the NIC resources using VFIO if it exists */ + switch (dev->kdrv) { + case RTE_KDRV_VFIO: +#ifdef VFIO_PRESENT + if (pci_vfio_is_enabled()) + ret = pci_vfio_map_resource(dev); +#endif + break; + case RTE_KDRV_IGB_UIO: + case RTE_KDRV_UIO_GENERIC: + /* map resources for devices that use uio */ + ret = pci_uio_map_resource(dev); + break; + default: + RTE_LOG(DEBUG, EAL, + " Not managed by a supported kernel driver, skipped\n"); + ret = 1; + break; + } + + return ret; } -static int -pci_uio_map_secondary(struct rte_pci_device *dev) +/* Unmap pci device */ +void +rte_eal_pci_unmap_device(struct rte_pci_device *dev) { - size_t i; - struct uio_resource *uio_res; - - TAILQ_FOREACH(uio_res, uio_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++) { - if (pci_map_resource(dev, uio_res->maps[i].addr, - uio_res->path, - (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"); - return (-1); - } - } - return (0); + /* try unmapping the NIC resources using VFIO if it exists */ + switch (dev->kdrv) { + case RTE_KDRV_VFIO: + RTE_LOG(ERR, EAL, "Hotplug doesn't support vfio yet\n"); + break; + case RTE_KDRV_IGB_UIO: + case RTE_KDRV_UIO_GENERIC: + /* unmap resources for devices that use uio */ + pci_uio_unmap_resource(dev); + break; + default: + RTE_LOG(DEBUG, EAL, + " Not managed by a supported kernel driver, skipped\n"); + break; } - - RTE_LOG(ERR, EAL, "Cannot find resource for device\n"); - return -1; } -/* map the PCI resource of a PCI device in virtual memory */ -static int -pci_uio_map_resource(struct rte_pci_device *dev) +void * +pci_find_max_end_va(void) { - int i, j; - struct dirent *e; - DIR *dir; - char dirname[PATH_MAX]; - char filename[PATH_MAX]; - char dirname2[PATH_MAX]; - char devname[PATH_MAX]; /* contains the /dev/uioX */ - void *mapaddr; - unsigned uio_num; - unsigned long start,size; - uint64_t phaddr; - uint64_t offset; - uint64_t pagesz; - ssize_t nb_maps; - struct rte_pci_addr *loc = &dev->addr; - struct uio_resource *uio_res; - struct uio_map *maps; - - dev->intr_handle.fd = -1; + const struct rte_memseg *seg = rte_eal_get_physmem_layout(); + const struct rte_memseg *last = seg; + unsigned i = 0; - /* secondary processes - use already recorded details */ - if ((rte_eal_process_type() != RTE_PROC_PRIMARY) && - (dev->id.vendor_id != PCI_VENDOR_ID_QUMRANET)) - return (pci_uio_map_secondary(dev)); - - /* 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(dirname2, sizeof(dirname2), - "%s/uio%u", dirname, uio_num); + for (i = 0; i < RTE_MAX_MEMSEG; i++, seg++) { + if (seg->addr == NULL) 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(dirname2, sizeof(dirname2), - "%s/uio:uio%u", dirname, uio_num); - break; - } - } - closedir(dir); + if (seg->addr > last->addr) + last = seg; - /* No uio resource found */ - if (e == NULL) { - RTE_LOG(WARNING, EAL, " "PCI_PRI_FMT" not managed by UIO driver, " - "skipping\n", loc->domain, loc->bus, loc->devid, loc->function); - return -1; } + return RTE_PTR_ADD(last->addr, last->len); +} - if(dev->id.vendor_id == PCI_VENDOR_ID_QUMRANET) { - /* get portio size */ - rte_snprintf(filename, sizeof(filename), - "%s/portio/port0/size", dirname2); - if (eal_parse_sysfs_value(filename, &size) < 0) { - RTE_LOG(ERR, EAL, "%s(): cannot parse size\n", - __func__); - return -1; - } +/* parse one line of the "resource" sysfs file (note that the 'line' + * string is modified) + */ +int +pci_parse_one_sysfs_resource(char *line, size_t len, uint64_t *phys_addr, + uint64_t *end_addr, uint64_t *flags) +{ + union pci_resource_info { + struct { + char *phys_addr; + char *end_addr; + char *flags; + }; + char *ptrs[PCI_RESOURCE_FMT_NVAL]; + } res_info; - /* get portio start */ - rte_snprintf(filename, sizeof(filename), - "%s/portio/port0/start", dirname2); - if (eal_parse_sysfs_value(filename, &start) < 0) { - RTE_LOG(ERR, EAL, "%s(): cannot parse portio start\n", - __func__); - return -1; - } - dev->mem_resource[0].addr = (void *)(uintptr_t)start; - dev->mem_resource[0].len = (uint64_t)size; - RTE_LOG(DEBUG, EAL, "PCI Port IO found start=0x%lx with size=0x%lx\n", start, size); - /* rte_virtio_pmd does not need any other bar even if available */ - return (0); - } - - /* allocate the mapping details for secondary processes*/ - if ((uio_res = rte_zmalloc("UIO_RES", sizeof (*uio_res), 0)) == NULL) { + if (rte_strsplit(line, len, res_info.ptrs, 3, ' ') != 3) { RTE_LOG(ERR, EAL, - "%s(): cannot store uio mmap details\n", __func__); - return (-1); + "%s(): bad resource format\n", __func__); + return -1; } - - rte_snprintf(devname, sizeof(devname), "/dev/uio%u", uio_num); - 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 */ - if ((nb_maps = pci_uio_get_mappings(dirname2, uio_res->maps, - sizeof (uio_res->maps) / sizeof (uio_res->maps[0]))) - < 0) - 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++) { - - /* 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; - if (maps[j].addr != NULL || - (mapaddr = pci_map_resource(dev, - NULL, devname, (off_t)offset, - (size_t)maps[j].size)) == NULL) { - return (-1); - } - - maps[j].addr = mapaddr; - maps[j].offset = offset; - dev->mem_resource[i].addr = mapaddr; - } + errno = 0; + *phys_addr = strtoull(res_info.phys_addr, NULL, 16); + *end_addr = strtoull(res_info.end_addr, NULL, 16); + *flags = strtoull(res_info.flags, NULL, 16); + if (errno != 0) { + RTE_LOG(ERR, EAL, + "%s(): bad resource format\n", __func__); + return -1; } - TAILQ_INSERT_TAIL(uio_res_list, uio_res, next); - - return (0); + return 0; } /* parse the "resource" sysfs file */ -#define IORESOURCE_MEM 0x00000200 - static int pci_parse_sysfs_resource(const char *filename, struct rte_pci_device *dev) { FILE *f; char buf[BUFSIZ]; - union pci_resource_info { - struct { - char *phys_addr; - char *end_addr; - char *flags; - }; - char *ptrs[PCI_RESOURCE_FMT_NVAL]; - } res_info; int i; uint64_t phys_addr, end_addr, flags; @@ -656,31 +242,19 @@ pci_parse_sysfs_resource(const char *filename, struct rte_pci_device *dev) for (i = 0; imem_resource[i].phys_addr = phys_addr; dev->mem_resource[i].len = end_addr - phys_addr + 1; /* not mapped for now */ - dev->mem_resource[i].addr = NULL; + dev->mem_resource[i].addr = NULL; } } fclose(f); @@ -691,55 +265,6 @@ 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) -{ - uint64_t dev_addr = (addr->domain << 24) + (addr->bus << 16) + (addr->devid << 8) + addr->function; - uint64_t dev_addr2 = (addr2->domain << 24) + (addr2->bus << 16) + (addr2->devid << 8) + addr2->function; - - if (dev_addr > dev_addr2) - return 1; - else - return 0; -} - - /* Scan one pci sysfs entry, and fill the devices list from it. */ static int pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, @@ -748,11 +273,12 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, char filename[PATH_MAX]; unsigned long tmp; struct rte_pci_device *dev; + char driver[PATH_MAX]; + int ret; dev = malloc(sizeof(*dev)); - if (dev == NULL) { + if (dev == NULL) return -1; - } memset(dev, 0, sizeof(*dev)); dev->addr.domain = domain; @@ -761,7 +287,7 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, dev->addr.function = function; /* get vendor id */ - rte_snprintf(filename, sizeof(filename), "%s/vendor", dirname); + snprintf(filename, sizeof(filename), "%s/vendor", dirname); if (eal_parse_sysfs_value(filename, &tmp) < 0) { free(dev); return -1; @@ -769,7 +295,7 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, dev->id.vendor_id = (uint16_t)tmp; /* get device id */ - rte_snprintf(filename, sizeof(filename), "%s/device", dirname); + snprintf(filename, sizeof(filename), "%s/device", dirname); if (eal_parse_sysfs_value(filename, &tmp) < 0) { free(dev); return -1; @@ -777,7 +303,7 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, dev->id.device_id = (uint16_t)tmp; /* get subsystem_vendor id */ - rte_snprintf(filename, sizeof(filename), "%s/subsystem_vendor", + snprintf(filename, sizeof(filename), "%s/subsystem_vendor", dirname); if (eal_parse_sysfs_value(filename, &tmp) < 0) { free(dev); @@ -786,7 +312,7 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, dev->id.subsystem_vendor_id = (uint16_t)tmp; /* get subsystem_device id */ - rte_snprintf(filename, sizeof(filename), "%s/subsystem_device", + snprintf(filename, sizeof(filename), "%s/subsystem_device", dirname); if (eal_parse_sysfs_value(filename, &tmp) < 0) { free(dev); @@ -794,20 +320,37 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, } dev->id.subsystem_device_id = (uint16_t)tmp; + /* get class_id */ + snprintf(filename, sizeof(filename), "%s/class", + dirname); + if (eal_parse_sysfs_value(filename, &tmp) < 0) { + free(dev); + return -1; + } + /* the least 24 bits are valid: class, subclass, program interface */ + dev->id.class_id = (uint32_t)tmp & RTE_CLASS_ANY_ID; + /* get max_vfs */ dev->max_vfs = 0; - rte_snprintf(filename, sizeof(filename), "%s/max_vfs", dirname); - if (!access(filename, F_OK) && - eal_parse_sysfs_value(filename, &tmp) == 0) { + snprintf(filename, sizeof(filename), "%s/max_vfs", dirname); + if (!access(filename, F_OK) && + eal_parse_sysfs_value(filename, &tmp) == 0) dev->max_vfs = (uint16_t)tmp; + else { + /* for non igb_uio driver, need kernel version >= 3.8 */ + snprintf(filename, sizeof(filename), + "%s/sriov_numvfs", dirname); + if (!access(filename, F_OK) && + eal_parse_sysfs_value(filename, &tmp) == 0) + dev->max_vfs = (uint16_t)tmp; } /* get numa node */ - rte_snprintf(filename, sizeof(filename), "%s/numa_node", + snprintf(filename, sizeof(filename), "%s/numa_node", dirname); if (access(filename, R_OK) != 0) { - /* if no NUMA support just set node to 0 */ - dev->numa_node = -1; + /* if no NUMA support, set default to 0 */ + dev->numa_node = 0; } else { if (eal_parse_sysfs_value(filename, &tmp) < 0) { free(dev); @@ -817,34 +360,76 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, } /* parse resources */ - rte_snprintf(filename, sizeof(filename), "%s/resource", dirname); + snprintf(filename, sizeof(filename), "%s/resource", dirname); if (pci_parse_sysfs_resource(filename, dev) < 0) { RTE_LOG(ERR, EAL, "%s(): cannot parse resource\n", __func__); free(dev); return -1; } + /* parse driver */ + snprintf(filename, sizeof(filename), "%s/driver", dirname); + ret = pci_get_kernel_driver_by_path(filename, driver); + if (ret < 0) { + RTE_LOG(ERR, EAL, "Fail to get kernel driver\n"); + free(dev); + return -1; + } + + if (!ret) { + if (!strcmp(driver, "vfio-pci")) + dev->kdrv = RTE_KDRV_VFIO; + else if (!strcmp(driver, "igb_uio")) + dev->kdrv = RTE_KDRV_IGB_UIO; + else if (!strcmp(driver, "uio_pci_generic")) + dev->kdrv = RTE_KDRV_UIO_GENERIC; + else + dev->kdrv = RTE_KDRV_UNKNOWN; + } else + dev->kdrv = RTE_KDRV_NONE; + /* device is valid, add in list (sorted) */ - if (TAILQ_EMPTY(&device_list)) { - TAILQ_INSERT_TAIL(&device_list, dev, next); - } - else { - struct rte_pci_device *dev2 = NULL; + if (TAILQ_EMPTY(&pci_device_list)) { + TAILQ_INSERT_TAIL(&pci_device_list, dev, next); + } else { + struct rte_pci_device *dev2; + int ret; - TAILQ_FOREACH(dev2, &device_list, next) { - if (pci_addr_comparison(&dev->addr, &dev2->addr)) + TAILQ_FOREACH(dev2, &pci_device_list, next) { + ret = rte_eal_compare_pci_addr(&dev->addr, &dev2->addr); + if (ret > 0) continue; - else { + + if (ret < 0) { TAILQ_INSERT_BEFORE(dev2, dev, next); - return 0; + } else { /* already registered */ + dev2->kdrv = dev->kdrv; + dev2->max_vfs = dev->max_vfs; + memmove(dev2->mem_resource, dev->mem_resource, + sizeof(dev->mem_resource)); + free(dev); } + return 0; } - TAILQ_INSERT_TAIL(&device_list, dev, next); + TAILQ_INSERT_TAIL(&pci_device_list, dev, next); } - + return 0; } +int +pci_update_device(const struct rte_pci_addr *addr) +{ + char filename[PATH_MAX]; + + snprintf(filename, sizeof(filename), "%s/" PCI_PRI_FMT, + pci_get_sysfs_path(), addr->domain, addr->bus, addr->devid, + addr->function); + + return pci_scan_one(filename, addr->domain, addr->bus, addr->devid, + addr->function); +} + /* * split up a pci address into its constituent parts. */ @@ -896,8 +481,8 @@ error: * Scan the content of the PCI bus, and the devices in the devices * list */ -static int -pci_scan(void) +int +rte_eal_pci_scan(void) { struct dirent *e; DIR *dir; @@ -905,7 +490,7 @@ pci_scan(void) uint16_t domain; uint8_t bus, devid, function; - dir = opendir(SYSFS_PCI_DEVICES); + dir = opendir(pci_get_sysfs_path()); if (dir == NULL) { RTE_LOG(ERR, EAL, "%s(): opendir failed: %s\n", __func__, strerror(errno)); @@ -920,8 +505,8 @@ pci_scan(void) &bus, &devid, &function) != 0) continue; - rte_snprintf(dirname, sizeof(dirname), "%s/%s", SYSFS_PCI_DEVICES, - e->d_name); + snprintf(dirname, sizeof(dirname), "%s/%s", + pci_get_sysfs_path(), e->d_name); if (pci_scan_one(dirname, domain, bus, devid, function) < 0) goto error; } @@ -933,98 +518,252 @@ error: return -1; } -/* - * If vendor/device ID match, call the devinit() function of the - * driver. - */ -int -rte_eal_pci_probe_one_driver(struct rte_pci_driver *dr, struct rte_pci_device *dev) +/* Read PCI config space. */ +int rte_eal_pci_read_config(const struct rte_pci_device *device, + void *buf, size_t len, off_t offset) { - struct rte_pci_id *id_table; -#ifdef RTE_EAL_UNBIND_PORTS - const char *module_name = NULL; - int uio_status = -1; + const struct rte_intr_handle *intr_handle = &device->intr_handle; + + switch (intr_handle->type) { + case RTE_INTR_HANDLE_UIO: + case RTE_INTR_HANDLE_UIO_INTX: + return pci_uio_read_config(intr_handle, buf, len, offset); + +#ifdef VFIO_PRESENT + case RTE_INTR_HANDLE_VFIO_MSIX: + case RTE_INTR_HANDLE_VFIO_MSI: + case RTE_INTR_HANDLE_VFIO_LEGACY: + return pci_vfio_read_config(intr_handle, buf, len, offset); +#endif + default: + RTE_LOG(ERR, EAL, + "Unknown handle type of fd %d\n", + intr_handle->fd); + return -1; + } +} - if (dr->drv_flags & RTE_PCI_DRV_NEED_IGB_UIO) - module_name = IGB_UIO_NAME; +/* Write PCI config space. */ +int rte_eal_pci_write_config(const struct rte_pci_device *device, + const void *buf, size_t len, off_t offset) +{ + const struct rte_intr_handle *intr_handle = &device->intr_handle; + + switch (intr_handle->type) { + case RTE_INTR_HANDLE_UIO: + case RTE_INTR_HANDLE_UIO_INTX: + return pci_uio_write_config(intr_handle, buf, len, offset); + +#ifdef VFIO_PRESENT + case RTE_INTR_HANDLE_VFIO_MSIX: + case RTE_INTR_HANDLE_VFIO_MSI: + case RTE_INTR_HANDLE_VFIO_LEGACY: + return pci_vfio_write_config(intr_handle, buf, len, offset); #endif + default: + RTE_LOG(ERR, EAL, + "Unknown handle type of fd %d\n", + intr_handle->fd); + return -1; + } +} - for (id_table = dr->id_table ; id_table->vendor_id != 0; id_table++) { +#if defined(RTE_ARCH_X86) +static int +pci_ioport_map(struct rte_pci_device *dev, int bar __rte_unused, + struct rte_pci_ioport *p) +{ + uint16_t start, end; + FILE *fp; + char *line = NULL; + char pci_id[16]; + int found = 0; + size_t linesz; + + snprintf(pci_id, sizeof(pci_id), PCI_PRI_FMT, + dev->addr.domain, dev->addr.bus, + dev->addr.devid, dev->addr.function); + + fp = fopen("/proc/ioports", "r"); + if (fp == NULL) { + RTE_LOG(ERR, EAL, "%s(): can't open ioports\n", __func__); + return -1; + } - /* check if device's identifiers match the driver's ones */ - if (id_table->vendor_id != dev->id.vendor_id && - id_table->vendor_id != PCI_ANY_ID) - continue; - if (id_table->device_id != dev->id.device_id && - id_table->device_id != PCI_ANY_ID) - continue; - if (id_table->subsystem_vendor_id != dev->id.subsystem_vendor_id && - id_table->subsystem_vendor_id != PCI_ANY_ID) - continue; - if (id_table->subsystem_device_id != dev->id.subsystem_device_id && - id_table->subsystem_device_id != PCI_ANY_ID) - continue; + while (getdelim(&line, &linesz, '\n', fp) > 0) { + char *ptr = line; + char *left; + int n; - struct rte_pci_addr *loc = &dev->addr; + n = strcspn(ptr, ":"); + ptr[n] = 0; + left = &ptr[n + 1]; - RTE_LOG(DEBUG, EAL, "PCI device "PCI_PRI_FMT" on NUMA socket %i\n", - loc->domain, loc->bus, loc->devid, loc->function, - dev->numa_node); + while (*left && isspace(*left)) + left++; - RTE_LOG(DEBUG, EAL, " probe driver: %x:%x %s\n", dev->id.vendor_id, - dev->id.device_id, dr->name); + if (!strncmp(left, pci_id, strlen(pci_id))) { + found = 1; - /* no initialization when blacklisted, return without error */ - if (dev->blacklisted) { - RTE_LOG(DEBUG, EAL, " Device is blacklisted, not initializing\n"); - return 0; + while (*ptr && isspace(*ptr)) + ptr++; + + sscanf(ptr, "%04hx-%04hx", &start, &end); + + break; } + } -#ifdef RTE_EAL_UNBIND_PORTS - /* Unbind PCI devices if needed */ - if (module_name != NULL) - if (pci_switch_module(dr, dev, uio_status, module_name) < 0) - return -1; + free(line); + fclose(fp); + + if (!found) + return -1; + + dev->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN; + p->base = start; + RTE_LOG(DEBUG, EAL, "PCI Port IO found start=0x%x\n", start); + + return 0; +} +#endif + +int +rte_eal_pci_ioport_map(struct rte_pci_device *dev, int bar, + struct rte_pci_ioport *p) +{ + int ret = -1; + + switch (dev->kdrv) { +#ifdef VFIO_PRESENT + case RTE_KDRV_VFIO: + if (pci_vfio_is_enabled()) + ret = pci_vfio_ioport_map(dev, bar, p); + break; +#endif + case RTE_KDRV_IGB_UIO: + ret = pci_uio_ioport_map(dev, bar, p); + break; + case RTE_KDRV_UIO_GENERIC: +#if defined(RTE_ARCH_X86) + ret = pci_ioport_map(dev, bar, p); #else - /* just map the NIC resources */ - if (pci_uio_map_resource(dev) < 0) - return -1; + ret = pci_uio_ioport_map(dev, bar, p); +#endif + break; + case RTE_KDRV_NONE: +#if defined(RTE_ARCH_X86) + ret = pci_ioport_map(dev, bar, p); #endif + break; + default: + break; + } - /* We always should have BAR0 mapped */ - if (rte_eal_process_type() == RTE_PROC_PRIMARY && - dev->mem_resource[0].addr == NULL) { - RTE_LOG(ERR, EAL, - "%s(): BAR0 is not mapped\n", - __func__); - return (-1); - } - - /* reference driver structure */ - dev->driver = dr; + if (!ret) + p->dev = dev; + + return ret; +} + +void +rte_eal_pci_ioport_read(struct rte_pci_ioport *p, + void *data, size_t len, off_t offset) +{ + switch (p->dev->kdrv) { +#ifdef VFIO_PRESENT + case RTE_KDRV_VFIO: + pci_vfio_ioport_read(p, data, len, offset); + break; +#endif + case RTE_KDRV_IGB_UIO: + pci_uio_ioport_read(p, data, len, offset); + break; + case RTE_KDRV_UIO_GENERIC: + pci_uio_ioport_read(p, data, len, offset); + break; + case RTE_KDRV_NONE: +#if defined(RTE_ARCH_X86) + pci_uio_ioport_read(p, data, len, offset); +#endif + break; + default: + break; + } +} - /* call the driver devinit() function */ - return dr->devinit(dr, dev); +void +rte_eal_pci_ioport_write(struct rte_pci_ioport *p, + const void *data, size_t len, off_t offset) +{ + switch (p->dev->kdrv) { +#ifdef VFIO_PRESENT + case RTE_KDRV_VFIO: + pci_vfio_ioport_write(p, data, len, offset); + break; +#endif + case RTE_KDRV_IGB_UIO: + pci_uio_ioport_write(p, data, len, offset); + break; + case RTE_KDRV_UIO_GENERIC: + pci_uio_ioport_write(p, data, len, offset); + break; + case RTE_KDRV_NONE: +#if defined(RTE_ARCH_X86) + pci_uio_ioport_write(p, data, len, offset); +#endif + break; + default: + break; } - /* return positive value if driver is not found */ - return 1; +} + +int +rte_eal_pci_ioport_unmap(struct rte_pci_ioport *p) +{ + int ret = -1; + + switch (p->dev->kdrv) { +#ifdef VFIO_PRESENT + case RTE_KDRV_VFIO: + if (pci_vfio_is_enabled()) + ret = pci_vfio_ioport_unmap(p); + break; +#endif + case RTE_KDRV_IGB_UIO: + ret = pci_uio_ioport_unmap(p); + break; + case RTE_KDRV_UIO_GENERIC: +#if defined(RTE_ARCH_X86) + ret = 0; +#else + ret = pci_uio_ioport_unmap(p); +#endif + break; + case RTE_KDRV_NONE: +#if defined(RTE_ARCH_X86) + ret = 0; +#endif + break; + default: + break; + } + + return ret; } /* Init the PCI EAL subsystem */ int rte_eal_pci_init(void) { - TAILQ_INIT(&driver_list); - TAILQ_INIT(&device_list); - uio_res_list = RTE_TAILQ_RESERVE_BY_IDX(RTE_TAILQ_PCI, uio_res_list); - /* for debug purposes, PCI can be disabled */ if (internal_config.no_pci) return 0; - if (pci_scan() < 0) { + if (rte_eal_pci_scan() < 0) { RTE_LOG(ERR, EAL, "%s(): Cannot scan PCI bus\n", __func__); return -1; } + return 0; }