From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pa0-f48.google.com (mail-pa0-f48.google.com [209.85.220.48]) by dpdk.org (Postfix) with ESMTP id E1D167F34 for ; Thu, 6 Nov 2014 12:05:43 +0100 (CET) Received: by mail-pa0-f48.google.com with SMTP id ey11so1051510pad.35 for ; Thu, 06 Nov 2014 03:15:10 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references; bh=COHvDcmJbsKIjeE5T7uJRO4F9BJ7MbSqiSsOOmhci4I=; b=F5fIv8Uq048mYN8KouY0s3FqMWhCnE0d+p3rLXgpQBRYqAH/Bc+1jJAd3JMAinVdrC sF0gO1OH8+N1AbEdIIhVgtljIWPMTodzh0zZpZUfoesPEUgKGnacLHuf7JLeOEikgUur 1Gv4iSn6U/QNpFbJDFjQZHzjaH8XKrIIhgCbQPnbrReJbsSwkXdC9niLuQufgh30GFnL A0Uali2FTvw//usjxn48wAiBKDG6RWqRLSGSEqNcmUS251Rt4e/NUrrqynFu1YO0Zl7O oDKVE/vJAuV2ivMmQGBGYr5KXj8GbwYrbb0ybP7Sp/FGK+37WhT7swnBWYytKnyFh3Xj 4F7Q== X-Gm-Message-State: ALoCoQmCArbbRk0gUy6FEpbGbUtjEFlX2OLr8OvntdOBwq1PA5sMdT5ebSw7ERFkvqSiT9lPmtlc X-Received: by 10.70.10.195 with SMTP id k3mr3599324pdb.41.1415272510350; Thu, 06 Nov 2014 03:15:10 -0800 (PST) Received: from localhost.localdomain (napt.igel.co.jp. [219.106.231.132]) by mx.google.com with ESMTPSA id jc3sm5652315pbb.49.2014.11.06.03.15.08 for (version=TLSv1.2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Thu, 06 Nov 2014 03:15:09 -0800 (PST) From: Tetsuya Mukawa To: dev@dpdk.org Date: Thu, 6 Nov 2014 20:14:31 +0900 Message-Id: <1415272471-3299-8-git-send-email-mukawa@igel.co.jp> X-Mailer: git-send-email 1.9.1 In-Reply-To: <1415272471-3299-1-git-send-email-mukawa@igel.co.jp> References: <1415272471-3299-1-git-send-email-mukawa@igel.co.jp> Cc: nakajima.yoshihiro@lab.ntt.co.jp, masutani.hitoshi@lab.ntt.co.jp Subject: [dpdk-dev] [RFC PATCH 7/7] lib/librte_vhost: Add vhost-user implementation X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 06 Nov 2014 11:05:45 -0000 This patch adds vhost-user implementation to librte_vhost. To communicate with vhost-user of QEMU, speficy VHOST_DRV_USER as a vhost_driver_type_t variable in rte_vhost_driver_register(). Signed-off-by: Tetsuya Mukawa --- lib/librte_vhost/rte_virtio_net.h | 19 +- lib/librte_vhost/vhost-net-user.c | 541 +++++++++++++++++++++++++++++++++++++ lib/librte_vhost/vhost-net.c | 39 ++- lib/librte_vhost/vhost-net.h | 7 + lib/librte_vhost/virtio-net-user.c | 410 ++++++++++++++++++++++++++++ lib/librte_vhost/virtio-net.c | 64 ++++- 6 files changed, 1073 insertions(+), 7 deletions(-) create mode 100644 lib/librte_vhost/vhost-net-user.c create mode 100644 lib/librte_vhost/virtio-net-user.c diff --git a/lib/librte_vhost/rte_virtio_net.h b/lib/librte_vhost/rte_virtio_net.h index a9e20ea..af07900 100644 --- a/lib/librte_vhost/rte_virtio_net.h +++ b/lib/librte_vhost/rte_virtio_net.h @@ -75,17 +75,32 @@ struct buf_vector { */ typedef enum { VHOST_DRV_CUSE, /* cuse driver */ + VHOST_DRV_USER, /* vhost-user driver */ VHOST_DRV_NUM /* the number of vhost driver types */ } vhost_driver_type_t; + +/** + * Structure contains vhost-user session specific information + */ +struct vhost_user_session { + int fh; /**< session identifier */ + pthread_t tid; /**< thread id of session handler */ + int socketfd; /**< fd of socket */ + int interval; /**< reconnection interval of session */ +}; + /** * Structure contains information relating vhost driver. */ struct vhost_driver { vhost_driver_type_t type; /**< driver type. */ const char *dev_name; /**< accessing device name. */ + void *priv; /**< private data. */ union { struct fuse_session *cuse_session; /**< fuse session. */ + struct vhost_user_session *user_session; + /**< vhost-user session. */ }; }; @@ -199,9 +214,11 @@ struct vhost_driver *rte_vhost_driver_register( const char *dev_name, vhost_driver_type_t type); /* Register callbacks. */ -int rte_vhost_driver_callback_register(struct virtio_net_device_ops const * const); +int rte_vhost_driver_callback_register(struct vhost_driver *drv, + struct virtio_net_device_ops const * const, void *priv); /* Start vhost driver session blocking loop. */ int rte_vhost_driver_session_start(struct vhost_driver *drv); +void rte_vhost_driver_session_stop(struct vhost_driver *drv); /** * This function adds buffers to the virtio devices RX virtqueue. Buffers can diff --git a/lib/librte_vhost/vhost-net-user.c b/lib/librte_vhost/vhost-net-user.c new file mode 100644 index 0000000..434f20f --- /dev/null +++ b/lib/librte_vhost/vhost-net-user.c @@ -0,0 +1,541 @@ +/*- + * BSD LICENSE + * + * Copyright (c) 2014 IGEL Co/.Ltd. + * 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 + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of IGEL 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 + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include +#include + +#define VHOST_USER_MAX_DEVICE (32) +#define VHOST_USER_MAX_FD_NUM (3) + +/* start id of vhost user device */ +rte_atomic16_t vhost_user_device_id; + +static struct vhost_net_device_ops const *ops; + +typedef enum VhostUserRequest { + VHOST_USER_NONE = 0, + VHOST_USER_GET_FEATURES = 1, + VHOST_USER_SET_FEATURES = 2, + VHOST_USER_SET_OWNER = 3, + VHOST_USER_RESET_OWNER = 4, + VHOST_USER_SET_MEM_TABLE = 5, + VHOST_USER_SET_LOG_BASE = 6, + VHOST_USER_SET_LOG_FD = 7, + VHOST_USER_SET_VRING_NUM = 8, + VHOST_USER_SET_VRING_ADDR = 9, + VHOST_USER_SET_VRING_BASE = 10, + VHOST_USER_GET_VRING_BASE = 11, + VHOST_USER_SET_VRING_KICK = 12, + VHOST_USER_SET_VRING_CALL = 13, + VHOST_USER_SET_VRING_ERR = 14, + VHOST_USER_MAX +} VhostUserRequest; + +#define VHOST_MEMORY_MAX_NREGIONS 8 + +typedef struct VhostUserMemoryRegion { + uint64_t guest_phys_addr; + uint64_t memory_size; + uint64_t userspace_addr; + uint64_t mmap_offset; +} VhostUserMemoryRegion; + +typedef struct VhostUserMemory { + uint32_t nregions; + uint32_t padding; + VhostUserMemoryRegion regions[VHOST_MEMORY_MAX_NREGIONS]; +} VhostUserMemory; + +typedef struct VhostUserMsg { + VhostUserRequest request; + +#define VHOST_USER_VERSION_MASK (0x3) +#define VHOST_USER_REPLY_MASK (0x1<<2) + uint32_t flags; + uint32_t size; /* the following payload size */ + union { +#define VHOST_USER_VRING_IDX_MASK (0xff) +#define VHOST_USER_VRING_NOFD_MASK (0x1<<8) + uint64_t u64; + struct vhost_vring_state state; + struct vhost_vring_addr addr; + VhostUserMemory memory; + }; +} __attribute__((packed)) VhostUserMsg; + +static VhostUserMsg m __attribute__ ((unused)); +#define VHOST_USER_HDR_SIZE (sizeof(m.request) \ + + sizeof(m.flags) + sizeof(m.size)) + +/* The version of the protocol we support */ +#define VHOST_USER_VERSION (0x1) + +static unsigned long int ioctl_to_vhost_user_request[VHOST_USER_MAX] = { + -1, /* VHOST_USER_NONE */ + VHOST_GET_FEATURES, /* VHOST_USER_GET_FEATURES */ + VHOST_SET_FEATURES, /* VHOST_USER_SET_FEATURES */ + VHOST_SET_OWNER, /* VHOST_USER_SET_OWNER */ + VHOST_RESET_OWNER, /* VHOST_USER_RESET_OWNER */ + VHOST_SET_MEM_TABLE, /* VHOST_USER_SET_MEM_TABLE */ + VHOST_SET_LOG_BASE, /* VHOST_USER_SET_LOG_BASE */ + VHOST_SET_LOG_FD, /* VHOST_USER_SET_LOG_FD */ + VHOST_SET_VRING_NUM, /* VHOST_USER_SET_VRING_NUM */ + VHOST_SET_VRING_ADDR, /* VHOST_USER_SET_VRING_ADDR */ + VHOST_SET_VRING_BASE, /* VHOST_USER_SET_VRING_BASE */ + VHOST_GET_VRING_BASE, /* VHOST_USER_GET_VRING_BASE */ + VHOST_SET_VRING_KICK, /* VHOST_USER_SET_VRING_KICK */ + VHOST_SET_VRING_CALL, /* VHOST_USER_SET_VRING_CALL */ + VHOST_SET_VRING_ERR /* VHOST_USER_SET_VRING_ERR */ +}; + +/** + * Returns vhost_device_ctx from given fuse_req_t. The index is populated later when + * the device is added to the device linked list. + */ +static struct vhost_device_ctx +vhost_driver_to_vhost_ctx(struct vhost_driver *drv) +{ + struct vhost_device_ctx ctx; + int device_id = drv->user_session->fh; + + ctx.type = VHOST_DRV_USER; + ctx.fh = device_id; + ctx.user.drv = drv; + + return ctx; +} + +/** + * When the device is created in QEMU it gets initialised here and added to the device linked list. + */ +static int +vhost_user_open(struct vhost_driver *drv) +{ + struct vhost_device_ctx ctx = vhost_driver_to_vhost_ctx(drv); + + int ret; + + ret = ops->new_device(ctx); + if (ret == -1) + return -1; + + RTE_LOG(INFO, VHOST_CONFIG, "(%"PRIu64") Device configuration started\n", ctx.fh); + + return 0; +} + +/** + * When QEMU is shutdown or killed the device gets released. + */ +static void +vhost_user_release(struct vhost_driver *drv) +{ + struct vhost_device_ctx ctx = vhost_driver_to_vhost_ctx(drv); + + ops->destroy_device(ctx); + RTE_LOG(INFO, VHOST_CONFIG, "(%"PRIu64") Device released\n", ctx.fh); +} + +/** + * Send data to vhost-user device on a QEMU. + */ +static int +vhost_user_write(struct vhost_driver *drv, VhostUserMsg *msg, + int *fds, size_t fd_num) +{ + int fd, len; + size_t fd_size = fd_num * sizeof(int); + char control[CMSG_SPACE(fd_size)]; + struct msghdr msg_header; + struct iovec iov[1]; + struct cmsghdr *cmsg_header; + struct vhost_device_ctx ctx = vhost_driver_to_vhost_ctx(drv); + + if ((drv == NULL) || (msg == NULL)) + return -EINVAL; + + fd = drv->user_session->socketfd; + + memset(&msg_header, 0, sizeof(msg_header)); + memset(control, 0, sizeof(control)); + + /* set the payload */ + iov[0].iov_base = (void *)msg; + iov[0].iov_len = VHOST_USER_HDR_SIZE + msg->size; + + msg_header.msg_iov = iov; + msg_header.msg_iovlen = 1; + + if (fd_num) { + msg_header.msg_control = control; + msg_header.msg_controllen = sizeof(control); + cmsg_header = CMSG_FIRSTHDR(&msg_header); + cmsg_header->cmsg_len = CMSG_LEN(fd_size); + cmsg_header->cmsg_level = SOL_SOCKET; + cmsg_header->cmsg_type = SCM_RIGHTS; + memcpy(CMSG_DATA(cmsg_header), fds, fd_size); + } else { + msg_header.msg_control = 0; + msg_header.msg_controllen = 0; + } + + do { + len = sendmsg(fd, &msg_header, 0); + } while (len < 0 && errno == EINTR); + + if (len < 0) + goto error; + + return 0; + +error: + RTE_LOG(INFO, VHOST_CONFIG, "(%"PRIu64") Device cannot send message\n", ctx.fh); + return -EFAULT; +} + +/** + * Receive data from vhost-user device on a QEMU. + */ +static int +vhost_user_read(struct vhost_driver *drv, VhostUserMsg *msg, + int *fds, size_t *fd_num) +{ + int fd, len; + size_t fd_size = (*fd_num) * sizeof(int); + char control[CMSG_SPACE(fd_size)]; + struct msghdr msg_header; + struct iovec iov[1]; + struct cmsghdr *cmsg_header; + struct vhost_device_ctx ctx = vhost_driver_to_vhost_ctx(drv); + + if ((drv == NULL) || (msg == NULL)) + return -EINVAL; + + fd = drv->user_session->socketfd; + + memset(&msg_header, 0, sizeof(msg_header)); + memset(control, 0, sizeof(control)); + *fd_num = 0; + + /* set the payload */ + iov[0].iov_base = (void *)msg; + iov[0].iov_len = VHOST_USER_HDR_SIZE; + + msg_header.msg_iov = iov; + msg_header.msg_iovlen = 1; + msg_header.msg_control = control; + msg_header.msg_controllen = sizeof(control); + + if ((len = recvmsg(fd, &msg_header, 0)) <= 0) + goto error; + + if (msg_header.msg_flags & (MSG_TRUNC | MSG_CTRUNC)) + goto error; + + cmsg_header = CMSG_FIRSTHDR(&msg_header); + if (cmsg_header && cmsg_header->cmsg_len > 0 && + cmsg_header->cmsg_level == SOL_SOCKET && + cmsg_header->cmsg_type == SCM_RIGHTS) { + if (fd_size >= cmsg_header->cmsg_len - CMSG_LEN(0)) { + fd_size = cmsg_header->cmsg_len - CMSG_LEN(0); + memcpy(fds, CMSG_DATA(cmsg_header), fd_size); + *fd_num = fd_size / sizeof(int); + } + } + + if (read(fd, ((char *)msg) + len, msg->size) < 0) + goto error; + + return 0; + +error: + RTE_LOG(INFO, VHOST_CONFIG, "(%"PRIu64") Device cannot receive message\n", ctx.fh); + return -EFAULT; +} + +/* + * Boilerplate code for vhost-user IOCTL + * Implicit arguments: ctx, req, result. + */ +#define VHOST_USER_IOCTL(func) do { \ + result = (func)(ctx); \ +} while (0) + +/* + * Boilerplate code for vhost-user Read IOCTL + * Implicit arguments: ctx, req, result, in_bufsz, in_buf. + */ +#define VHOST_USER_IOCTL_R(type, var, func) do {\ + result = func(ctx, &(var)); \ +} while (0) + +/* + * Boilerplate code for vhost-user Write IOCTL + * Implicit arguments: ctx, req, result, out_bufsz. + */ +#define VHOST_USER_IOCTL_W(type, var, func) do {\ + result = (func)(ctx, &(var)); \ + msg->flags |= VHOST_USER_REPLY_MASK; \ + msg->size = sizeof(type); \ + vhost_user_write(drv, msg, NULL, 0); \ +} while (0) + +/* + * Boilerplate code for vhost-user Read/Write IOCTL + * Implicit arguments: ctx, req, result, in_bufsz, in_buf. + */ +#define VHOST_USER_IOCTL_RW(type1, var1, type2, var2, func) do {\ + result = (func)(ctx, (var1), &(var2)); \ + msg->flags |= VHOST_USER_REPLY_MASK; \ + msg->size = sizeof(type2); \ + vhost_user_write(drv, msg, NULL, 0); \ +} while (0) + +/** + * The IOCTLs are handled using unix domain socket in userspace. + */ + static int +vhost_user_ioctl(struct vhost_driver *drv, VhostUserMsg *msg, + int *fds, int fd_num) +{ + struct vhost_device_ctx ctx = vhost_driver_to_vhost_ctx(drv); + struct vhost_vring_file file; + int result = 0; + + switch (ioctl_to_vhost_user_request[msg->request]) { + case VHOST_GET_FEATURES: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_GET_FEATURES\n", ctx.fh); + VHOST_USER_IOCTL_W(uint64_t, msg->u64, ops->get_features); + break; + + case VHOST_SET_FEATURES: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_FEATURES\n", ctx.fh); + VHOST_USER_IOCTL_R(uint64_t, msg->u64, ops->set_features); + break; + + case VHOST_RESET_OWNER: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_RESET_OWNER\n", ctx.fh); + VHOST_USER_IOCTL(ops->reset_owner); + break; + + case VHOST_SET_OWNER: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_OWNER\n", ctx.fh); + VHOST_USER_IOCTL(ops->set_owner); + break; + + case VHOST_SET_MEM_TABLE: + /*TODO fix race condition.*/ + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_MEM_TABLE\n", ctx.fh); + /* all fds should be same, because physical memory consist of an one file */ + ctx.user.fds = fds; + ctx.user.fd_num = fd_num; + result = ops->set_mem_table(ctx, &msg->memory, msg->memory.nregions); + break; + + case VHOST_SET_VRING_NUM: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_VRING_NUM\n", ctx.fh); + VHOST_USER_IOCTL_R(struct vhost_vring_state, msg->state, ops->set_vring_num); + break; + + case VHOST_SET_VRING_BASE: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_VRING_BASE\n", ctx.fh); + VHOST_USER_IOCTL_R(struct vhost_vring_state, msg->state, ops->set_vring_base); + break; + + case VHOST_GET_VRING_BASE: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_GET_VRING_BASE\n", ctx.fh); + VHOST_USER_IOCTL_RW(uint32_t, msg->addr.index, struct vhost_vring_state, msg->state, ops->get_vring_base); + break; + + case VHOST_SET_VRING_ADDR: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_VRING_ADDR\n", ctx.fh); + VHOST_USER_IOCTL_R(struct vhost_vring_addr, msg->addr, ops->set_vring_addr); + break; + + case VHOST_SET_VRING_KICK: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_VRING_KICK\n", ctx.fh); + ctx.user.fds = fds; + ctx.user.fd_num = fd_num; + file.index = msg->u64; + VHOST_USER_IOCTL_R(struct vhost_vring_file, file, ops->set_vring_kick); + break; + + case VHOST_SET_VRING_CALL: + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: VHOST_SET_VRING_CALL\n", ctx.fh); + ctx.user.fds = fds; + ctx.user.fd_num = fd_num; + file.index = msg->u64; + VHOST_USER_IOCTL_R(struct vhost_vring_file, file, ops->set_vring_call); + break; + + case VHOST_NET_SET_BACKEND: + default: + RTE_LOG(ERR, VHOST_CONFIG, "(%"PRIu64") IOCTL: DOESN NOT EXIST\n", ctx.fh); + result = -1; + } + + if (result < 0) + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: FAIL\n", ctx.fh); + else + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") IOCTL: SUCCESS\n", ctx.fh); + + return result; +} + +/** + * vhost-user specific registration. + */ +static int +vhost_user_driver_register(struct vhost_driver *drv) +{ + if ((drv == NULL) || (drv->dev_name == NULL) || + (strlen(drv->dev_name) > UNIX_PATH_MAX - 1)) + return -1; + + ops = get_virtio_net_callbacks(drv->type); + + drv->user_session = rte_malloc(NULL, sizeof(struct vhost_user_session), CACHE_LINE_SIZE); + if (drv->user_session == NULL) + return -1; + + drv->user_session->fh = + rte_atomic16_add_return(&vhost_user_device_id, 1) - 1; /* fh of first device is zero */ + drv->user_session->interval = 1; + + return 0; +} + +/** + * When vhost-user driver starts, the session handler communicates with vhost-user + * device on a QEMU using a unix domain sokcet. + */ +static void * +vhost_user_session_handler(void *data) +{ + struct vhost_driver *drv = data; + int ret; + struct sockaddr_un caddr; + VhostUserMsg msg; + int fds[VHOST_USER_MAX_FD_NUM]; + int socketfd; + int interval; + size_t fd_num; + + if ((drv == NULL) || (drv->dev_name == NULL)) + return NULL; + + bzero(&caddr, sizeof(caddr)); + caddr.sun_family = AF_LOCAL; + strncpy((char *)&caddr.sun_path, drv->dev_name, strlen(drv->dev_name)); + +reconnect: + drv->user_session->socketfd = socket(AF_UNIX, SOCK_STREAM, 0); + if (drv->user_session->socketfd < 0) + return NULL; + + socketfd = drv->user_session->socketfd; + interval = drv->user_session->interval; + while (1) { + ret = connect(socketfd, (struct sockaddr *)&caddr, sizeof(caddr)); + if (ret == 0) + break; /* success */ + sleep(interval); + } + + ret = vhost_user_open(drv); + if (ret != 0) { + RTE_LOG(ERR, VHOST_CONFIG, "(Socket %s) open failuer\n", drv->dev_name); + return NULL; + } + + for (;;) { + fd_num = VHOST_USER_MAX_FD_NUM; + bzero(&msg, sizeof(VhostUserMsg)); + ret = vhost_user_read(drv, &msg, fds, &fd_num); + if (ret != 0) { + RTE_LOG(ERR, VHOST_CONFIG, "(Socket %s) read failuer\n", drv->dev_name); + vhost_user_release(drv); + goto reconnect; + } + + ret = vhost_user_ioctl(drv, &msg, fds, fd_num); + if (ret != 0) { + RTE_LOG(ERR, VHOST_CONFIG, "(Socket %s) request failuer\n", drv->dev_name); + vhost_user_release(drv); + goto reconnect; + } + } + + return NULL; +} + +/** + * Create session handler + */ +static int +vhost_user_driver_start(struct vhost_driver *drv) +{ + if (pthread_create(&drv->user_session->tid, NULL, vhost_user_session_handler, drv)) { + RTE_LOG(ERR, VHOST_CONFIG, + "(Socket %s) starting event handler failuer\n", drv->dev_name); + return -1; + } + + /* TODO: The event handler thread may need to run on a core user speficied. */ + + return 0; +} + +/** + * Destroy session handler + */ +static void +vhost_user_driver_stop(struct vhost_driver *drv) +{ + pthread_t *tid = &drv->user_session->tid; + + if (pthread_create(tid, NULL, vhost_user_session_handler, drv)) { + RTE_LOG(ERR, VHOST_CONFIG, + "(Socket %s) starting event handler failuer\n", drv->dev_name); + return; + } + + /* stop event thread and wait until connection is closed */ + if (*tid) { + pthread_cancel(*tid); + pthread_join(*tid, NULL); + } + + vhost_user_release(drv); +} diff --git a/lib/librte_vhost/vhost-net.c b/lib/librte_vhost/vhost-net.c index b0de5fd..10f41e9 100644 --- a/lib/librte_vhost/vhost-net.c +++ b/lib/librte_vhost/vhost-net.c @@ -42,6 +42,11 @@ */ #include "vhost-net-cdev.c" +/* + * Include vhost-user depend functions and definitions + */ +#include "vhost-net-user.c" + /** * This function abstracts cuse and vhost-user driver registration. */ @@ -65,10 +70,17 @@ rte_vhost_driver_register(const char *dev_name, vhost_driver_type_t type) if (ret != 0) goto err; break; + case VHOST_DRV_USER: + ret = vhost_user_driver_register(drv); + break; default: + ret = -EINVAL; break; } + if (ret != 0) + goto err; + return drv; err: free(drv); @@ -81,17 +93,40 @@ err: int rte_vhost_driver_session_start(struct vhost_driver *drv) { + int ret; + if (drv == NULL) return -ENODEV; switch (drv->type) { case VHOST_DRV_CUSE: - vhost_cuse_driver_session_start(drv); + ret = vhost_cuse_driver_session_start(drv); + break; + case VHOST_DRV_USER: + ret = vhost_user_driver_start(drv); break; default: + ret = -EINVAL; break; } - return 0; + return ret; } +/** + * The vhost session is closed, only allow for vhost-user. + */ +void +rte_vhost_driver_session_stop(struct vhost_driver *drv) +{ + if (drv == NULL) + return; + + switch (drv->type) { + case VHOST_DRV_USER: + vhost_user_driver_stop(drv); + break; + default: + break; + } +} diff --git a/lib/librte_vhost/vhost-net.h b/lib/librte_vhost/vhost-net.h index ef04832..0e36ba0 100644 --- a/lib/librte_vhost/vhost-net.h +++ b/lib/librte_vhost/vhost-net.h @@ -76,6 +76,12 @@ struct vhost_device_cuse_ctx { pid_t pid; /* PID of process calling the IOCTL. */ }; +struct vhost_device_user_ctx { + int *fds; + int fd_num; + struct vhost_driver *drv; +}; + /* * Structure used to identify device context. */ @@ -83,6 +89,7 @@ struct vhost_device_ctx { vhost_driver_type_t type; /* driver type. */ uint64_t fh; /* Populated with fi->fh to track the device index. */ union { + struct vhost_device_user_ctx user; struct vhost_device_cuse_ctx cdev; }; }; diff --git a/lib/librte_vhost/virtio-net-user.c b/lib/librte_vhost/virtio-net-user.c new file mode 100644 index 0000000..1e78f98 --- /dev/null +++ b/lib/librte_vhost/virtio-net-user.c @@ -0,0 +1,410 @@ +/*- + * BSD LICENSE + * + * Copyright (c) 2014 IGEL Co.,Ltd. + * 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 + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of IGEL 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 + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include +#include +#include +#include +#include + +/* Functions defined in virtio_net.c */ +static void init_device(struct vhost_device_ctx ctx, struct virtio_net *dev); +static void cleanup_device(struct virtio_net *dev); +static void free_device(struct virtio_net_config_ll *ll_dev); +static int new_device(struct vhost_device_ctx ctx); +static void destroy_device(struct vhost_device_ctx ctx); +static int set_owner(struct vhost_device_ctx ctx); +static int reset_owner(struct vhost_device_ctx ctx); +static int get_features(struct vhost_device_ctx ctx, uint64_t *pu); +static int set_features(struct vhost_device_ctx ctx, uint64_t *pu); +static int set_vring_num(struct vhost_device_ctx ctx, struct vhost_vring_state *state); +static int set_vring_addr(struct vhost_device_ctx ctx, struct vhost_vring_addr *addr); +static int set_vring_base(struct vhost_device_ctx ctx, struct vhost_vring_state *state); +static int set_backend(struct vhost_device_ctx ctx, struct vhost_vring_file *file); + +/* Root address of the linked list in the configuration core. */ +static struct virtio_net_config_ll *user_ll_root; + +/** + * Retrieves an entry from the devices configuration linked list. + */ +static struct virtio_net_config_ll * +user_get_config_ll_entry(struct vhost_device_ctx ctx) +{ + struct virtio_net_config_ll *ll_dev = user_ll_root; + + /* Loop through linked list until the device_fh is found. */ + while (ll_dev != NULL) { + if (ll_dev->dev.device_fh == ctx.fh) + return ll_dev; + ll_dev = ll_dev->next; + } + + return NULL; +} + +/** + * Searches the configuration core linked list and retrieves the device if it exists. + */ +static struct virtio_net * +user_get_device(struct vhost_device_ctx ctx) +{ + struct virtio_net_config_ll *ll_dev; + + ll_dev = user_get_config_ll_entry(ctx); + + /* If a matching entry is found in the linked list, return the device in that entry. */ + if (ll_dev) + return &ll_dev->dev; + + RTE_LOG(ERR, VHOST_CONFIG, "(%"PRIu64") Device not found in linked list.\n", ctx.fh); + return NULL; +} + +/** + * Add entry containing a device to the device configuration linked list. + */ +static void +user_add_config_ll_entry(struct virtio_net_config_ll *new_ll_dev) +{ + struct virtio_net_config_ll *ll_dev = user_ll_root; + + /* If ll_dev == NULL then this is the first device so go to else */ + if (ll_dev) { + /* If the 1st device_fh != 0 then we insert our device here. */ + if (ll_dev->dev.device_fh != 0) { + new_ll_dev->dev.device_fh = 0; + new_ll_dev->next = ll_dev; + user_ll_root = new_ll_dev; + } else { + /* Increment through the ll until we find un unused device_fh. Insert the device at that entry*/ + while ((ll_dev->next != NULL) && (ll_dev->dev.device_fh == (ll_dev->next->dev.device_fh - 1))) + ll_dev = ll_dev->next; + + new_ll_dev->dev.device_fh = ll_dev->dev.device_fh + 1; + new_ll_dev->next = ll_dev->next; + ll_dev->next = new_ll_dev; + } + } else { + user_ll_root = new_ll_dev; + user_ll_root->dev.device_fh = 0; + } + +} + +/** + * Remove an entry from the device configuration linked list. + */ +static struct virtio_net_config_ll * +user_rm_config_ll_entry(struct virtio_net_config_ll *ll_dev, struct virtio_net_config_ll *ll_dev_last) +{ + /* First remove the device and then clean it up. */ + if (ll_dev == user_ll_root) { + user_ll_root = ll_dev->next; + cleanup_device(&ll_dev->dev); + free_device(ll_dev); + return user_ll_root; + } else { + if (likely(ll_dev_last != NULL)) { + ll_dev_last->next = ll_dev->next; + cleanup_device(&ll_dev->dev); + free_device(ll_dev); + return ll_dev_last->next; + } else { + cleanup_device(&ll_dev->dev); + free_device(ll_dev); + RTE_LOG(ERR, VHOST_CONFIG, "Remove entry from config_ll failed\n"); + return NULL; + } + } +} + +/** + * Returns the root entry of linked list + */ +static struct virtio_net_config_ll * +user_get_config_ll_root(void) +{ + return user_ll_root; +} + +/** + * vhost-user specific device initialization. + */ +static void +user_init_device(struct vhost_device_ctx ctx, struct virtio_net *dev) +{ + dev->priv = ctx.user.drv->priv; +} + +/** + * Locate the file containing QEMU's memory space and map it to our address space. + */ +static int +user_host_memory_map(struct virtio_net *dev, struct virtio_memory *mem, int fd, size_t size) +{ + void *map; + + map = mmap(0, size, PROT_READ|PROT_WRITE , MAP_POPULATE|MAP_SHARED, fd, 0); + close(fd); + + if (map == MAP_FAILED) { + RTE_LOG(ERR, VHOST_CONFIG, "(%"PRIu64") Error mapping the file fd %d\n", dev->device_fh, fd); + return -1; + } + + /* Store the memory address and size in the device data structure */ + mem->mapped_address = (uint64_t)(uintptr_t)map; + mem->mapped_size = size; + + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") Mem File: fd: %d - Size: %llu - VA: %p\n", dev->device_fh, + fd, (long long unsigned)mem->mapped_size, map); + + return 0; +} + +/* + * Called from IOCTL: VHOST_SET_MEM_TABLE + * This function creates and populates the memory structure for the device. This includes + * storing offsets used to translate buffer addresses. + */ +static int +user_set_mem_table(struct vhost_device_ctx ctx, const void *mem_regions_addr, uint32_t nregions) +{ + struct virtio_net *dev; + struct vhost_memory_region *mem_regions; + struct virtio_memory *mem; + uint64_t size = offsetof(struct vhost_memory, regions); + uint32_t regionidx, valid_regions; + size_t guest_memory_size = 0; + + dev = user_get_device(ctx); + if (dev == NULL) + return -1; + + if (dev->mem) { + munmap((void *)(uintptr_t)dev->mem->mapped_address, (size_t)dev->mem->mapped_size); + free(dev->mem); + } + + /* Malloc the memory structure depending on the number of regions. */ + mem = calloc(1, sizeof(struct virtio_memory) + (sizeof(struct virtio_memory_regions) * nregions)); + if (mem == NULL) { + RTE_LOG(ERR, VHOST_CONFIG, "(%"PRIu64") Failed to allocate memory for dev->mem.\n", dev->device_fh); + return -1; + } + + mem->nregions = nregions; + + mem_regions = (void*)(uintptr_t)((uint64_t)(uintptr_t)mem_regions_addr + size); + + for (regionidx = 0; regionidx < mem->nregions; regionidx++) { + /* Populate the region structure for each region. */ + mem->regions[regionidx].guest_phys_address = mem_regions[regionidx].guest_phys_addr; + mem->regions[regionidx].guest_phys_address_end = mem->regions[regionidx].guest_phys_address + + mem_regions[regionidx].memory_size; + mem->regions[regionidx].memory_size = mem_regions[regionidx].memory_size; + mem->regions[regionidx].userspace_address = mem_regions[regionidx].userspace_addr; + guest_memory_size += mem_regions[regionidx].memory_size; + + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") REGION: %u - GPA: %p - QEMU VA: %p - SIZE (%"PRIu64")\n", dev->device_fh, + regionidx, (void*)(uintptr_t)mem->regions[regionidx].guest_phys_address, + (void*)(uintptr_t)mem->regions[regionidx].userspace_address, + mem->regions[regionidx].memory_size); + } + + for (regionidx = 0; regionidx < mem->nregions; regionidx++) { + /*set the base address mapping*/ + if (mem->regions[regionidx].guest_phys_address == 0x0) { + mem->base_address = mem->regions[regionidx].userspace_address; + /* Map VM memory file */ + if (user_host_memory_map(dev, mem, ctx.user.fds[regionidx], guest_memory_size) != 0) { + free(mem); + return -1; + } + } else + close(ctx.user.fds[regionidx]); + } + + /* Check that we have a valid base address. */ + if (mem->base_address == 0) { + RTE_LOG(ERR, VHOST_CONFIG, "(%"PRIu64") Failed to find base address of qemu memory file.\n", dev->device_fh); + free(mem); + return -1; + } + + /* Check if all of our regions have valid mappings. Usually one does not exist in the QEMU memory file. */ + valid_regions = mem->nregions; + for (regionidx = 0; regionidx < mem->nregions; regionidx++) { + if ((mem->regions[regionidx].userspace_address < mem->base_address) || + (mem->regions[regionidx].userspace_address > (mem->base_address + mem->mapped_size))) + valid_regions--; + } + + /* If a region does not have a valid mapping we rebuild our memory struct to contain only valid entries. */ + if (valid_regions != mem->nregions) { + LOG_DEBUG(VHOST_CONFIG, "(%"PRIu64") Not all memory regions exist in the QEMU mem file. Re-populating mem structure\n", + dev->device_fh); + + /* Re-populate the memory structure with only valid regions. Invalid regions are over-written with memmove. */ + valid_regions = 0; + + for (regionidx = mem->nregions; 0 != regionidx--;) { + if ((mem->regions[regionidx].userspace_address < mem->base_address) || + (mem->regions[regionidx].userspace_address > (mem->base_address + mem->mapped_size))) { + memmove(&mem->regions[regionidx], &mem->regions[regionidx + 1], + sizeof(struct virtio_memory_regions) * valid_regions); + } else { + valid_regions++; + } + } + } + mem->nregions = valid_regions; + dev->mem = mem; + + /* + * Calculate the address offset for each region. This offset is used to identify the vhost virtual address + * corresponding to a QEMU guest physical address. + */ + for (regionidx = 0; regionidx < dev->mem->nregions; regionidx++) + dev->mem->regions[regionidx].address_offset = dev->mem->regions[regionidx].userspace_address - dev->mem->base_address + + dev->mem->mapped_address - dev->mem->regions[regionidx].guest_phys_address; + + return 0; +} + +/** + * Called from IOCTL: VHOST_GET_VRING_BASE + * We send the virtio device our available ring last used index. + */ +static int +user_get_vring_base(struct vhost_device_ctx ctx, uint32_t index, struct vhost_vring_state *state) +{ + struct virtio_net *dev; + + dev = user_get_device(ctx); + if (dev == NULL) + return -1; + + state->index = index; + /* State->index refers to the queue index. The TX queue is 1, RX queue is 0. */ + state->num = dev->virtqueue[state->index]->last_used_idx; + + return 0; +} + +/** + * Called from IOCTL: VHOST_SET_VRING_CALL + * The virtio device sends an eventfd to interrupt the guest. This fd gets copied in + * to our process space. + * Also this message is sent when virtio-net device is reset by device driver on QEMU. + */ +static int +user_set_vring_call(struct vhost_device_ctx ctx, struct vhost_vring_file *file) +{ + struct virtio_net *dev; + struct vhost_virtqueue *vq; + + dev = user_get_device(ctx); + if (dev == NULL) + return -1; + + /* file->index refers to the queue index. The TX queue is 1, RX queue is 0. */ + vq = dev->virtqueue[file->index]; + + if (vq->kickfd) + close((int)vq->kickfd); + + /* Populate the eventfd_copy structure and call eventfd_copy. */ + vq->kickfd = ctx.user.fds[0]; + + return 0; +} + +/** + * Called from IOCTL: VHOST_SET_VRING_KICK + * The virtio device sends an eventfd that it can use to notify us. This fd gets copied in + * to our process space. + */ +static int +user_set_vring_kick(struct vhost_device_ctx ctx, struct vhost_vring_file *file) +{ + struct virtio_net *dev; + struct vhost_virtqueue *vq; + + dev = user_get_device(ctx); + if (dev == NULL) + return -1; + + /* file->index refers to the queue index. The TX queue is 1, RX queue is 0. */ + vq = dev->virtqueue[file->index]; + + if (vq->callfd) + close((int)vq->callfd); + + /* Populate the eventfd_copy structure and call eventfd_copy. */ + vq->callfd = ctx.user.fds[0]; + + if ((dev->virtqueue[VIRTIO_RXQ] != NULL) && (dev->virtqueue[VIRTIO_TXQ]) != NULL) + return set_backend(ctx, file); + + return 0; +} + +/** + * Function pointers are set for the device operations to allow to call functions + * when an IOCTL, device_add or device_release is received. + */ +static const struct vhost_net_device_ops vhost_user_device_ops = { + .new_device = new_device, + .destroy_device = destroy_device, + + .get_features = get_features, + .set_features = set_features, + + .set_mem_table = user_set_mem_table, + + .set_vring_num = set_vring_num, + .set_vring_addr = set_vring_addr, + .set_vring_base = set_vring_base, + .get_vring_base = user_get_vring_base, + + .set_vring_kick = user_set_vring_kick, + .set_vring_call = user_set_vring_call, + + .set_backend = set_backend, + + .set_owner = set_owner, + .reset_owner = reset_owner, +}; diff --git a/lib/librte_vhost/virtio-net.c b/lib/librte_vhost/virtio-net.c index 13fbb6f..db810e7 100644 --- a/lib/librte_vhost/virtio-net.c +++ b/lib/librte_vhost/virtio-net.c @@ -96,6 +96,12 @@ qva_to_vva(struct virtio_net *dev, uint64_t qemu_va) */ #include "virtio-net-cdev.c" +/** + * Include vhost-user depend functions and definitions. + */ +#include "virtio-net-user.c" + + /* * Retrieves an entry from the devices configuration linked list. */ @@ -105,6 +111,8 @@ get_config_ll_entry(struct vhost_device_ctx ctx) switch (ctx.type) { case VHOST_DRV_CUSE: return cdev_get_config_ll_entry(ctx); + case VHOST_DRV_USER: + return user_get_config_ll_entry(ctx); default: break; } @@ -120,6 +128,8 @@ get_device(struct vhost_device_ctx ctx) switch (ctx.type) { case VHOST_DRV_CUSE: return cdev_get_device(ctx); + case VHOST_DRV_USER: + return user_get_device(ctx); default: break; } @@ -136,6 +146,8 @@ add_config_ll_entry(vhost_driver_type_t type, switch (type) { case VHOST_DRV_CUSE: return cdev_add_config_ll_entry(new_ll_dev); + case VHOST_DRV_USER: + return user_add_config_ll_entry(new_ll_dev); default: break; } @@ -149,8 +161,39 @@ cleanup_device(struct virtio_net *dev) { /* Unmap QEMU memory file if mapped. */ if (dev->mem) { - munmap((void *)(uintptr_t)dev->mem->mapped_address, - (size_t)dev->mem->mapped_size); + { + /* + * 'munmap()' will be failed when mapped_size isn't + * aligned with hugepage size. + * Usually a file size of QEMU physical memory is + * aligned by hugepage size. So In a case of CUSE, + * there is no problem. But with vhost-user, there is + * no way to get physical memory size. + * + * Let's assume hugepage size is 2MB or 1GB here. + * BTW, 'mmap()' automatically fixed size parameter + * to be aligned. Why does 'munmap()' do like so? + */ + int ret = 0; + size_t hugepagesize, size = dev->mem->mapped_size; + + /* assume hugepage size is 2MB */ + hugepagesize = 2 * 1024 * 1024; + size = (size + hugepagesize - 1) / + hugepagesize * hugepagesize; + ret = munmap((void *)(uintptr_t) + dev->mem->mapped_address, + size); + if (ret) { + /* assume hugepage size is 1GB, try again */ + hugepagesize = 1024 * 1024 * 1024; + size = (size + hugepagesize - 1) / + hugepagesize * hugepagesize; + munmap((void *)(uintptr_t) + dev->mem->mapped_address, + size); + } + } free(dev->mem); } @@ -187,6 +230,8 @@ rm_config_ll_entry(vhost_driver_type_t type, switch (type) { case VHOST_DRV_CUSE: return cdev_rm_config_ll_entry(ll_dev, ll_dev_last); + case VHOST_DRV_USER: + return user_rm_config_ll_entry(ll_dev, ll_dev_last); default: break; } @@ -201,7 +246,9 @@ get_config_ll_root(struct vhost_device_ctx ctx) { switch (ctx.type) { case VHOST_DRV_CUSE: - return cdev_get_config_ll_root(ctx); + return cdev_get_config_ll_root(); + case VHOST_DRV_USER: + return user_get_config_ll_root(); default: break; } @@ -232,6 +279,8 @@ init_device(struct vhost_device_ctx ctx, struct virtio_net *dev) switch (ctx.type) { case VHOST_DRV_CUSE: return cdev_init_device(ctx, dev); + case VHOST_DRV_USER: + return user_init_device(ctx, dev); default: break; } @@ -527,6 +576,8 @@ get_virtio_net_callbacks(vhost_driver_type_t type) switch (type) { case VHOST_DRV_CUSE: return &vhost_cuse_device_ops; + case VHOST_DRV_USER: + return &vhost_user_device_ops; default: break; } @@ -570,9 +621,14 @@ int rte_vhost_feature_enable(uint64_t feature_mask) * Register ops so that we can add/remove device to data core. */ int -rte_vhost_driver_callback_register(struct virtio_net_device_ops const * const ops) +rte_vhost_driver_callback_register(struct vhost_driver *drv, + struct virtio_net_device_ops const * const ops, void *priv) { + if (drv == NULL || ops == NULL) + return -1; + notify_ops = ops; + drv->priv = priv; return 0; } -- 1.9.1