X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;f=lib%2Flibrte_eal%2Flinuxapp%2Feal%2Feal_pci.c;h=434697323c00a85d3e7b82fe3327f398681a4a4d;hb=756ce64b1ecdf107acfa45fc3f31359ca338649e;hp=20a4342b0de962d49f84bec72ec59b9e10683db8;hpb=eee16c964cd6b90622c36a7976ad951313b2ccd3;p=dpdk.git diff --git a/lib/librte_eal/linuxapp/eal/eal_pci.c b/lib/librte_eal/linuxapp/eal/eal_pci.c index 20a4342b0d..434697323c 100644 --- a/lib/librte_eal/linuxapp/eal/eal_pci.c +++ b/lib/librte_eal/linuxapp/eal/eal_pci.c @@ -1,521 +1,196 @@ /*- * BSD LICENSE - * - * Copyright(c) 2010-2013 Intel Corporation. All rights reserved. + * + * 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 + * + * 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 + * + * * 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 - * notice, this list of conditions and the following disclaimer in - * the documentation and/or other materials provided with the + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the * distribution. - * * Neither the name of Intel Corporation nor the names of its - * contributors may be used to endorse or promote products derived + * * 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 - * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT - * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, - * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT - * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, - * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY - * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT - * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * + * 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 + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE * 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 "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. + * 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). */ -struct uio_map { - void *addr; - uint64_t offset; - uint64_t size; - uint64_t phaddr; -}; - -#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 */ - -/* - * 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); - -/* - * 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. - */ -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[]) +/* unbind kernel driver for this device */ +int +pci_unbind_kernel_driver(struct rte_pci_device *dev) { - FILE *f; int n; + FILE *f; + char filename[PATH_MAX]; char buf[BUFSIZ]; - char dev_bind[PATH_MAX]; struct rte_pci_addr *loc = &dev->addr; - RTE_LOG(DEBUG, EAL, "bind PCI device "PCI_PRI_FMT"\n", - loc->domain, loc->bus, loc->devid, loc->function); + /* open /sys/bus/pci/devices/AAAA:BB:CC.D/driver */ + snprintf(filename, sizeof(filename), + SYSFS_PCI_DEVICES "/" PCI_PRI_FMT "/driver/unbind", + loc->domain, loc->bus, loc->devid, loc->function); - 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(filename, "w"); + if (f == NULL) /* device was not bound */ + return 0; - f = fopen(dev_bind, "w"); - if (f == NULL) { - RTE_LOG(ERR, EAL, "Cannot open %s\n", dev->previous_dr); - return -1; - } - n = rte_snprintf(buf, sizeof(buf), PCI_PRI_FMT "\n", - loc->domain, loc->bus, loc->devid, loc->function); + 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, "Cannot rte_snprintf PCI infos\n"); - fclose(f); - return -1; + RTE_LOG(ERR, EAL, "%s(): snprintf failed\n", __func__); + goto error; } if (fwrite(buf, n, 1, f) == 0) { - fclose(f); - return -1; + RTE_LOG(ERR, EAL, "%s(): could not write to %s\n", __func__, + filename); + goto error; } - RTE_LOG(DEBUG, EAL, "Device bound\n"); - fclose(f); return 0; + +error: + fclose(f); + return -1; } static int -pci_uio_bind_device(struct rte_pci_device *dev, const char *module_name) +pci_get_kernel_driver_by_path(const char *filename, char *dri_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; - } + int count; + char path[PATH_MAX]; + char *name; - 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"); + if (!filename || !dri_name) return -1; - } - f = fopen(uio_newid, "w"); - if (f == NULL) { - RTE_LOG(ERR, EAL, "Cannot open %s\n", uio_newid); + count = readlink(filename, path, PATH_MAX); + if (count >= PATH_MAX) return -1; - } - if (fwrite(buf, n, 1, f) == 0) { - fclose(f); - return -1; - } - fclose(f); - pci_bind_device(dev, uio_bind); - return 0; -} + /* For device does not have a driver */ + if (count < 0) + return 1; + path[count] = '\0'; -/* - * open devname: it can take some time to - * appear, so we wait some time before returning an error - */ -static int uio_open(const char *devname) -{ - int n, fd; - - for (n=0; n < UIO_DEV_WAIT_TIMEOUT*10; n++) { - fd = open(devname, O_RDWR); - if (fd >= 0) - return fd; - - if (errno != ENOENT) - break; - usleep(100000); - } - return -1; -} - -/* map a particular resource from a file */ -static void * -pci_mmap(int fd, void *addr, off_t offset, size_t size) -{ - void *mapaddr; - - /* Map the PCI memory resource of device */ - mapaddr = mmap(addr, size, PROT_READ | PROT_WRITE, MAP_SHARED, - fd, offset); - if (mapaddr == MAP_FAILED || (addr != NULL && mapaddr != addr)) { - RTE_LOG(ERR, EAL, "%s(): cannot mmap %zd@0x%lx: %s\n", - __func__, size, offset, strerror(errno)); - return NULL; + 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 dirname2[PATH_MAX]; - char devname[PATH_MAX]; /* contains the /dev/uioX */ - void *mapaddr; - unsigned uio_num; - 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; - - RTE_LOG(DEBUG, EAL, "map PCI resource for device "PCI_PRI_FMT"\n", - loc->domain, loc->bus, loc->devid, loc->function); - - /* secondary processes - use already recorded details */ - if (rte_eal_process_type() != RTE_PROC_PRIMARY) - return pci_uio_map_restore(dev); + const struct rte_memseg *seg = rte_eal_get_physmem_layout(); + const struct rte_memseg *last = seg; + unsigned i = 0; - /* depending on kernel version, uio can be located in uio/uioX - * or uio:uioX */ - - rte_snprintf(dirname, sizeof(dirname), - "/sys/bus/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), - "/sys/bus/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) { - int shortprefix_len = sizeof("uio") - 1; /* format could be uio%d ...*/ - int longprefix_len = sizeof("uio:uio") - 1; /* ... or uio:uio%d */ - 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) { - rte_snprintf(dirname2, sizeof(dirname2), - "%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) { - rte_snprintf(dirname2, sizeof(dirname2), - "%s/uio:uio%u", dirname, uio_num); + for (i = 0; i < RTE_MAX_MEMSEG; i++, seg++) { + if (seg->addr == NULL) break; - } - } - closedir(dir); - /* 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; - } + if (seg->addr > last->addr) + last = seg; - /* 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(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; - } - } - - TAILQ_INSERT_TAIL(uio_res_list, uio_res, next); - - return (0); + return RTE_PTR_ADD(last->addr, last->len); } /* parse the "resource" sysfs file */ -#define IORESOURCE_MEM 0x00000200 - static int pci_parse_sysfs_resource(const char *filename, struct rte_pci_device *dev) { @@ -541,12 +216,14 @@ 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); @@ -573,55 +251,6 @@ 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, @@ -630,11 +259,12 @@ 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; @@ -643,7 +273,7 @@ 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; @@ -651,7 +281,7 @@ 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; @@ -659,7 +289,7 @@ 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); @@ -668,7 +298,7 @@ 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); @@ -676,32 +306,90 @@ pci_scan_one(const char *dirname, uint16_t domain, uint8_t bus, } dev->id.subsystem_device_id = (uint16_t)tmp; + /* get max_vfs */ + dev->max_vfs = 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 */ + snprintf(filename, sizeof(filename), "%s/numa_node", + dirname); + if (access(filename, R_OK) != 0) { + /* if no NUMA support, set default to 0 */ + dev->numa_node = 0; + } else { + if (eal_parse_sysfs_value(filename, &tmp) < 0) { + free(dev); + return -1; + } + dev->numa_node = tmp; + } + /* 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; } - /* 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; + /* 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; + } - TAILQ_FOREACH(dev2, &device_list, next) { - if (pci_addr_comparison(&dev->addr, &dev2->addr)) + 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_UNKNOWN; + + /* device is valid, add in list (sorted) */ + 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, &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; } @@ -756,8 +444,8 @@ 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; @@ -780,7 +468,7 @@ pci_scan(void) &bus, &devid, &function) != 0) continue; - rte_snprintf(dirname, sizeof(dirname), "%s/%s", SYSFS_PCI_DEVICES, + snprintf(dirname, sizeof(dirname), "%s/%s", SYSFS_PCI_DEVICES, e->d_name); if (pci_scan_one(dirname, domain, bus, devid, function) < 0) goto error; @@ -793,217 +481,346 @@ error: return -1; } -/* unbind kernel driver for this device */ +#ifdef RTE_PCI_CONFIG static int -pci_unbind_kernel_driver(struct rte_pci_device *dev) +pci_config_extended_tag(struct rte_pci_device *dev) { - int n; - FILE *f; + struct rte_pci_addr *loc = &dev->addr; char filename[PATH_MAX]; char buf[BUFSIZ]; - 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); - - RTE_LOG(DEBUG, EAL, "unbind kernel driver %s\n", filename); + FILE *f; - f = fopen(filename, "w"); - if (f == NULL) /* device was not bound */ + /* not configured, let it as is */ + if (strncmp(RTE_PCI_EXTENDED_TAG, "on", 2) != 0 && + strncmp(RTE_PCI_EXTENDED_TAG, "off", 3) != 0) return 0; - 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, "%s(): rte_snprintf failed\n", __func__); - goto error; - } - if (fwrite(buf, n, 1, f) == 0) { - RTE_LOG(ERR, EAL, "%s(): could not write to %s\n", __func__, - filename); - goto error; - } + snprintf(filename, sizeof(filename), + SYSFS_PCI_DEVICES "/" PCI_PRI_FMT "/" "extended_tag", + loc->domain, loc->bus, loc->devid, loc->function); + f = fopen(filename, "rw+"); + if (!f) + return -1; + fgets(buf, sizeof(buf), f); + if (strncmp(RTE_PCI_EXTENDED_TAG, "on", 2) == 0) { + /* enable Extended Tag*/ + if (strncmp(buf, "on", 2) != 0) { + fseek(f, 0, SEEK_SET); + fputs("on", f); + } + } else { + /* disable Extended Tag */ + if (strncmp(buf, "off", 3) != 0) { + fseek(f, 0, SEEK_SET); + fputs("off", f); + } + } fclose(f); - return 0; -error: - fclose(f); - return -1; + return 0; } static int -pci_exit_process(struct rte_pci_device *dev) +pci_config_max_read_request_size(struct rte_pci_device *dev) { - if (munmap(dev->mem_resource.addr, dev->mem_resource.len) == -1){ - RTE_LOG(ERR, EAL, "Error with munmap\n"); - return -1; - } - if (close(dev->intr_handle.fd) == -1){ - RTE_LOG(ERR, EAL, "Error closing interrupt handle\n"); + struct rte_pci_addr *loc = &dev->addr; + char filename[PATH_MAX]; + char buf[BUFSIZ], param[BUFSIZ]; + FILE *f; + /* size can be 128, 256, 512, 1024, 2048, 4096 */ + uint32_t max_size = RTE_PCI_MAX_READ_REQUEST_SIZE; + + /* not configured, let it as is */ + if (!max_size) + return 0; + + snprintf(filename, sizeof(filename), + SYSFS_PCI_DEVICES "/" PCI_PRI_FMT "/" "max_read_request_size", + loc->domain, loc->bus, loc->devid, loc->function); + f = fopen(filename, "rw+"); + if (!f) return -1; + + fgets(buf, sizeof(buf), f); + snprintf(param, sizeof(param), "%d", max_size); + + /* check if the size to be set is the same as current */ + if (strcmp(buf, param) == 0) { + fclose(f); + return 0; } - if (rte_eal_process_type() == RTE_PROC_PRIMARY) { - if (pci_unbind_kernel_driver(dev) < 0){ - RTE_LOG(ERR, EAL, "Error unbinding\n"); - return -1; - } - if (pci_bind_device(dev, dev->previous_dr) < 0){ - RTE_LOG(ERR, EAL, "Error binding\n"); - return -1; - } - } + fseek(f, 0, SEEK_SET); + fputs(param, f); + fclose(f); return 0; } -static void -pci_get_previous_driver_path(struct rte_pci_device *dev) +void +pci_config_space_set(struct rte_pci_device *dev) { - int n; - char dev_path[PATH_MAX]; - char dr_path[PATH_MAX]; - struct rte_pci_addr *loc = &dev->addr; + if (rte_eal_process_type() != RTE_PROC_PRIMARY) + return; - n = rte_snprintf(dev_path, sizeof(dev_path), SYSFS_PCI_DEVICES "/" - PCI_PRI_FMT "/driver", loc->domain, loc->bus, loc->devid, loc->function ); - if ((n < 0) || (n >= (int)sizeof(dev_path))) - RTE_LOG(ERR, EAL, "Cannot rte_snprintf device filepath\n"); + /* configure extended tag */ + pci_config_extended_tag(dev); - n = readlink(dev_path, dr_path, sizeof(dr_path)); - if ((n < 0) || (n >= (int)sizeof(dr_path))){ - RTE_LOG(ERR, EAL, "Cannot readlink driver filepath\n"); - dev->previous_dr[0] = '\0'; - return; + /* configure max read request size */ + pci_config_max_read_request_size(dev); +} +#endif + +/* Read PCI config space. */ +int rte_eal_pci_read_config(const struct rte_pci_device *device, + 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_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; } - dr_path[n] = '\0'; - - if (dr_path[0] != '/') - n = rte_snprintf(dev->previous_dr, sizeof(dev->previous_dr), - SYSFS_PCI_DEVICES "/" PCI_PRI_FMT "/%s", - loc->domain, loc->bus, loc->devid, loc->function, dr_path); - else - n = rte_snprintf(dev->previous_dr, sizeof(dev->previous_dr), - "%s", dr_path); - if ((n < 0) || (n >= (int)sizeof(dev_path))) - RTE_LOG(ERR, EAL, "Cannot rte_snprintf driver filepath\n"); } -/* - * 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) +/* 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; + } +} + +#if defined(RTE_ARCH_X86_64) || defined(RTE_ARCH_I686) +static int +pci_ioport_map(struct rte_pci_device *dev, int bar __rte_unused, + struct rte_pci_ioport *p) { - struct rte_pci_id *id_table; - const char *module_name = NULL; - int uio_status = -1; + 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; + } - if (dr->drv_flags & RTE_PCI_DRV_NEED_IGB_UIO) - module_name = IGB_UIO_NAME; + while (getdelim(&line, &linesz, '\n', fp) > 0) { + char *ptr = line; + char *left; + int n; - for (id_table = dr->id_table ; id_table->vendor_id != 0; id_table++) { + n = strcspn(ptr, ":"); + ptr[n] = 0; + left = &ptr[n + 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 (*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; - /* reference driver structure */ - dev->driver = dr; + while (*ptr && isspace(*ptr)) + ptr++; - /* no initialization when blacklisted, return without error */ - if (dev->blacklisted) - return 0; + sscanf(ptr, "%04hx-%04hx", &start, &end); - /* Unbind PCI devices if needed */ - if (module_name != NULL) { - - 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 */ - pci_get_previous_driver_path(dev); - 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) - return -1; + break; + } + } + + 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; + + switch (dev->kdrv) { +#ifdef VFIO_PRESENT + case RTE_KDRV_VFIO: + ret = -1; + if (pci_vfio_is_enabled()) + ret = pci_vfio_ioport_map(dev, bar, p); + break; +#endif + case RTE_KDRV_IGB_UIO: + case RTE_KDRV_UIO_GENERIC: + ret = pci_uio_ioport_map(dev, bar, p); + break; + default: +#if defined(RTE_ARCH_X86_64) || defined(RTE_ARCH_I686) + /* special case for x86 ... */ + ret = pci_ioport_map(dev, bar, p); #else - /* just map the NIC resources */ - if (pci_uio_map_resource(dev) < 0) - return -1; + ret = -1; #endif + 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; - /* call the driver devinit() function */ - return dr->devinit(dr, 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: + case RTE_KDRV_UIO_GENERIC: + pci_uio_ioport_read(p, data, len, offset); + break; + default: +#if defined(RTE_ARCH_X86_64) || defined(RTE_ARCH_I686) + /* special case for x86 ... */ + pci_uio_ioport_read(p, data, len, offset); +#endif + break; } - return -1; } -/*Start the exit process for each dev in use*/ void -rte_eal_pci_exit(void) +rte_eal_pci_ioport_write(struct rte_pci_ioport *p, + const void *data, size_t len, off_t offset) { - struct rte_pci_device *dev = NULL; + 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: + case RTE_KDRV_UIO_GENERIC: + pci_uio_ioport_write(p, data, len, offset); + break; + default: +#if defined(RTE_ARCH_X86_64) || defined(RTE_ARCH_I686) + /* special case for x86 ... */ + pci_uio_ioport_write(p, data, len, offset); +#endif + break; + } +} - TAILQ_FOREACH(dev, &device_list, next){ - if (dev->previous_dr[0] == '\0') - continue; - if(pci_exit_process(dev) == 1) - RTE_LOG(ERR, EAL, "Exit process failure\n"); +int +rte_eal_pci_ioport_unmap(struct rte_pci_ioport *p) +{ + int ret; + + switch (p->dev->kdrv) { +#ifdef VFIO_PRESENT + case RTE_KDRV_VFIO: + ret = -1; + if (pci_vfio_is_enabled()) + ret = pci_vfio_ioport_unmap(p); + break; +#endif + case RTE_KDRV_IGB_UIO: + case RTE_KDRV_UIO_GENERIC: + ret = pci_uio_ioport_unmap(p); + break; + default: +#if defined(RTE_ARCH_X86_64) || defined(RTE_ARCH_I686) + /* special case for x86 ... nothing to do */ + ret = 0; +#else + ret = -1; +#endif + 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); + TAILQ_INIT(&pci_driver_list); + TAILQ_INIT(&pci_device_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; } +#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; }