diff options
author | Cindy Lu <lulu@redhat.com> | 2020-07-01 22:55:37 +0800 |
---|---|---|
committer | Michael S. Tsirkin <mst@redhat.com> | 2020-07-07 07:59:51 -0400 |
commit | 108a64818e69be0a97cde3838d768f2d9910c08b (patch) | |
tree | a60b587a2ff2b704dac731d9886bc8c30c7d0d0c /hw | |
parent | 38140cc4d9713dc9af78090503105bf9c82b6bff (diff) | |
download | qemu-108a64818e69be0a97cde3838d768f2d9910c08b.zip qemu-108a64818e69be0a97cde3838d768f2d9910c08b.tar.gz qemu-108a64818e69be0a97cde3838d768f2d9910c08b.tar.bz2 |
vhost-vdpa: introduce vhost-vdpa backend
Currently we have 2 types of vhost backends in QEMU: vhost kernel and
vhost-user. The above patch provides a generic device for vDPA purpose,
this vDPA device exposes to user space a non-vendor-specific configuration
interface for setting up a vhost HW accelerator, this patch set introduces
a third vhost backend called vhost-vdpa based on the vDPA interface.
Vhost-vdpa usage:
qemu-system-x86_64 -cpu host -enable-kvm \
......
-netdev type=vhost-vdpa,vhostdev=/dev/vhost-vdpa-id,id=vhost-vdpa0 \
-device virtio-net-pci,netdev=vhost-vdpa0,page-per-vq=on \
Signed-off-by: Lingshan zhu <lingshan.zhu@intel.com>
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Signed-off-by: Cindy Lu <lulu@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20200701145538.22333-14-lulu@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Diffstat (limited to 'hw')
-rw-r--r-- | hw/net/vhost_net.c | 18 | ||||
-rw-r--r-- | hw/net/virtio-net.c | 19 | ||||
-rw-r--r-- | hw/virtio/Makefile.objs | 1 | ||||
-rw-r--r-- | hw/virtio/vhost-backend.c | 6 | ||||
-rw-r--r-- | hw/virtio/vhost-vdpa.c | 475 |
5 files changed, 513 insertions, 6 deletions
diff --git a/hw/net/vhost_net.c b/hw/net/vhost_net.c index 4561665..24d555e 100644 --- a/hw/net/vhost_net.c +++ b/hw/net/vhost_net.c @@ -17,6 +17,7 @@ #include "net/net.h" #include "net/tap.h" #include "net/vhost-user.h" +#include "net/vhost-vdpa.h" #include "standard-headers/linux/vhost_types.h" #include "hw/virtio/virtio-net.h" @@ -33,12 +34,6 @@ #include "hw/virtio/vhost.h" #include "hw/virtio/virtio-bus.h" -struct vhost_net { - struct vhost_dev dev; - struct vhost_virtqueue vqs[2]; - int backend; - NetClientState *nc; -}; /* Features supported by host kernel. */ static const int kernel_feature_bits[] = { @@ -96,6 +91,11 @@ static const int *vhost_net_get_feature_bits(struct vhost_net *net) case NET_CLIENT_DRIVER_VHOST_USER: feature_bits = user_feature_bits; break; +#ifdef CONFIG_VHOST_NET_VDPA + case NET_CLIENT_DRIVER_VHOST_VDPA: + feature_bits = vdpa_feature_bits; + break; +#endif default: error_report("Feature bits not defined for this type: %d", net->nc->info->type); @@ -444,6 +444,12 @@ VHostNetState *get_vhost_net(NetClientState *nc) assert(vhost_net); break; #endif +#ifdef CONFIG_VHOST_NET_VDPA + case NET_CLIENT_DRIVER_VHOST_VDPA: + vhost_net = vhost_vdpa_get_vhost_net(nc); + assert(vhost_net); + break; +#endif default: break; } diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c index aff67a9..0a54b19 100644 --- a/hw/net/virtio-net.c +++ b/hw/net/virtio-net.c @@ -43,6 +43,7 @@ #include "monitor/qdev.h" #include "hw/pci/pci.h" #include "net_rx_pkt.h" +#include "hw/virtio/vhost.h" #define VIRTIO_NET_VM_VERSION 11 @@ -125,6 +126,8 @@ static void virtio_net_get_config(VirtIODevice *vdev, uint8_t *config) VirtIONet *n = VIRTIO_NET(vdev); struct virtio_net_config netcfg; + int ret = 0; + memset(&netcfg, 0 , sizeof(struct virtio_net_config)); virtio_stw_p(vdev, &netcfg.status, n->status); virtio_stw_p(vdev, &netcfg.max_virtqueue_pairs, n->max_queues); virtio_stw_p(vdev, &netcfg.mtu, n->net_conf.mtu); @@ -138,6 +141,15 @@ static void virtio_net_get_config(VirtIODevice *vdev, uint8_t *config) virtio_stl_p(vdev, &netcfg.supported_hash_types, VIRTIO_NET_RSS_SUPPORTED_HASHES); memcpy(config, &netcfg, n->config_size); + + NetClientState *nc = qemu_get_queue(n->nic); + if (nc->peer->info->type == NET_CLIENT_DRIVER_VHOST_VDPA) { + ret = vhost_net_get_config(get_vhost_net(nc->peer), (uint8_t *)&netcfg, + n->config_size); + if (ret != -1) { + memcpy(config, &netcfg, n->config_size); + } + } } static void virtio_net_set_config(VirtIODevice *vdev, const uint8_t *config) @@ -153,6 +165,13 @@ static void virtio_net_set_config(VirtIODevice *vdev, const uint8_t *config) memcpy(n->mac, netcfg.mac, ETH_ALEN); qemu_format_nic_info_str(qemu_get_queue(n->nic), n->mac); } + + NetClientState *nc = qemu_get_queue(n->nic); + if (nc->peer->info->type == NET_CLIENT_DRIVER_VHOST_VDPA) { + vhost_net_set_config(get_vhost_net(nc->peer), (uint8_t *)&netcfg, + 0, n->config_size, + VHOST_SET_CONFIG_TYPE_MASTER); + } } static bool virtio_net_started(VirtIONet *n, uint8_t status) diff --git a/hw/virtio/Makefile.objs b/hw/virtio/Makefile.objs index a986708..fc91719 100644 --- a/hw/virtio/Makefile.objs +++ b/hw/virtio/Makefile.objs @@ -5,6 +5,7 @@ obj-y += virtio.o obj-$(CONFIG_VHOST) += vhost.o vhost-backend.o common-obj-$(call lnot,$(CONFIG_VHOST)) += vhost-stub.o obj-$(CONFIG_VHOST_USER) += vhost-user.o +obj-$(CONFIG_VHOST_VDPA) += vhost-vdpa.o common-obj-$(CONFIG_VIRTIO_RNG) += virtio-rng.o common-obj-$(CONFIG_VIRTIO_PCI) += virtio-pci.o diff --git a/hw/virtio/vhost-backend.c b/hw/virtio/vhost-backend.c index 4890538..782b1d6 100644 --- a/hw/virtio/vhost-backend.c +++ b/hw/virtio/vhost-backend.c @@ -15,6 +15,7 @@ #include "qemu/main-loop.h" #include "standard-headers/linux/vhost_types.h" +#include "hw/virtio/vhost-vdpa.h" #ifdef CONFIG_VHOST_KERNEL #include <linux/vhost.h> #include <sys/ioctl.h> @@ -286,6 +287,11 @@ int vhost_set_backend_type(struct vhost_dev *dev, VhostBackendType backend_type) dev->vhost_ops = &user_ops; break; #endif +#ifdef CONFIG_VHOST_VDPA + case VHOST_BACKEND_TYPE_VDPA: + dev->vhost_ops = &vdpa_ops; + break; +#endif default: error_report("Unknown vhost backend type"); r = -1; diff --git a/hw/virtio/vhost-vdpa.c b/hw/virtio/vhost-vdpa.c new file mode 100644 index 0000000..a3d17fe --- /dev/null +++ b/hw/virtio/vhost-vdpa.c @@ -0,0 +1,475 @@ +/* + * vhost-vdpa + * + * Copyright(c) 2017-2018 Intel Corporation. + * Copyright(c) 2020 Red Hat, Inc. + * + * This work is licensed under the terms of the GNU GPL, version 2 or later. + * See the COPYING file in the top-level directory. + * + */ + +#include "qemu/osdep.h" +#include <linux/vhost.h> +#include <linux/vfio.h> +#include <sys/eventfd.h> +#include <sys/ioctl.h> +#include "hw/virtio/vhost.h" +#include "hw/virtio/vhost-backend.h" +#include "hw/virtio/virtio-net.h" +#include "hw/virtio/vhost-vdpa.h" +#include "qemu/main-loop.h" +#include <linux/kvm.h> +#include "sysemu/kvm.h" + +static bool vhost_vdpa_listener_skipped_section(MemoryRegionSection *section) +{ + return (!memory_region_is_ram(section->mr) && + !memory_region_is_iommu(section->mr)) || + /* + * Sizing an enabled 64-bit BAR can cause spurious mappings to + * addresses in the upper part of the 64-bit address space. These + * are never accessed by the CPU and beyond the address width of + * some IOMMU hardware. TODO: VDPA should tell us the IOMMU width. + */ + section->offset_within_address_space & (1ULL << 63); +} + +static int vhost_vdpa_dma_map(struct vhost_vdpa *v, hwaddr iova, hwaddr size, + void *vaddr, bool readonly) +{ + struct vhost_msg_v2 msg; + int fd = v->device_fd; + int ret = 0; + + msg.type = v->msg_type; + msg.iotlb.iova = iova; + msg.iotlb.size = size; + msg.iotlb.uaddr = (uint64_t)(uintptr_t)vaddr; + msg.iotlb.perm = readonly ? VHOST_ACCESS_RO : VHOST_ACCESS_RW; + msg.iotlb.type = VHOST_IOTLB_UPDATE; + + if (write(fd, &msg, sizeof(msg)) != sizeof(msg)) { + error_report("failed to write, fd=%d, errno=%d (%s)", + fd, errno, strerror(errno)); + return -EIO ; + } + + return ret; +} + +static int vhost_vdpa_dma_unmap(struct vhost_vdpa *v, hwaddr iova, + hwaddr size) +{ + struct vhost_msg_v2 msg; + int fd = v->device_fd; + int ret = 0; + + msg.type = v->msg_type; + msg.iotlb.iova = iova; + msg.iotlb.size = size; + msg.iotlb.type = VHOST_IOTLB_INVALIDATE; + + if (write(fd, &msg, sizeof(msg)) != sizeof(msg)) { + error_report("failed to write, fd=%d, errno=%d (%s)", + fd, errno, strerror(errno)); + return -EIO ; + } + + return ret; +} + +static void vhost_vdpa_listener_region_add(MemoryListener *listener, + MemoryRegionSection *section) +{ + struct vhost_vdpa *v = container_of(listener, struct vhost_vdpa, listener); + hwaddr iova; + Int128 llend, llsize; + void *vaddr; + int ret; + + if (vhost_vdpa_listener_skipped_section(section)) { + return; + } + + if (unlikely((section->offset_within_address_space & ~TARGET_PAGE_MASK) != + (section->offset_within_region & ~TARGET_PAGE_MASK))) { + error_report("%s received unaligned region", __func__); + return; + } + + iova = TARGET_PAGE_ALIGN(section->offset_within_address_space); + llend = int128_make64(section->offset_within_address_space); + llend = int128_add(llend, section->size); + llend = int128_and(llend, int128_exts64(TARGET_PAGE_MASK)); + + if (int128_ge(int128_make64(iova), llend)) { + return; + } + + memory_region_ref(section->mr); + + /* Here we assume that memory_region_is_ram(section->mr)==true */ + + vaddr = memory_region_get_ram_ptr(section->mr) + + section->offset_within_region + + (iova - section->offset_within_address_space); + + llsize = int128_sub(llend, int128_make64(iova)); + + ret = vhost_vdpa_dma_map(v, iova, int128_get64(llsize), + vaddr, section->readonly); + if (ret) { + error_report("vhost vdpa map fail!"); + if (memory_region_is_ram_device(section->mr)) { + /* Allow unexpected mappings not to be fatal for RAM devices */ + error_report("map ram fail!"); + return ; + } + goto fail; + } + + return; + +fail: + if (memory_region_is_ram_device(section->mr)) { + error_report("failed to vdpa_dma_map. pci p2p may not work"); + return; + + } + /* + * On the initfn path, store the first error in the container so we + * can gracefully fail. Runtime, there's not much we can do other + * than throw a hardware error. + */ + error_report("vhost-vdpa: DMA mapping failed, unable to continue"); + return; + +} + +static void vhost_vdpa_listener_region_del(MemoryListener *listener, + MemoryRegionSection *section) +{ + struct vhost_vdpa *v = container_of(listener, struct vhost_vdpa, listener); + hwaddr iova; + Int128 llend, llsize; + int ret; + bool try_unmap = true; + + if (vhost_vdpa_listener_skipped_section(section)) { + return; + } + + if (unlikely((section->offset_within_address_space & ~TARGET_PAGE_MASK) != + (section->offset_within_region & ~TARGET_PAGE_MASK))) { + error_report("%s received unaligned region", __func__); + return; + } + + iova = TARGET_PAGE_ALIGN(section->offset_within_address_space); + llend = int128_make64(section->offset_within_address_space); + llend = int128_add(llend, section->size); + llend = int128_and(llend, int128_exts64(TARGET_PAGE_MASK)); + + if (int128_ge(int128_make64(iova), llend)) { + return; + } + + llsize = int128_sub(llend, int128_make64(iova)); + + if (try_unmap) { + ret = vhost_vdpa_dma_unmap(v, iova, int128_get64(llsize)); + if (ret) { + error_report("vhost_vdpa dma unmap error!"); + } + } + + memory_region_unref(section->mr); +} +/* + * IOTLB API is used by vhost-vpda which requires incremental updating + * of the mapping. So we can not use generic vhost memory listener which + * depends on the addnop(). + */ +static const MemoryListener vhost_vdpa_memory_listener = { + .region_add = vhost_vdpa_listener_region_add, + .region_del = vhost_vdpa_listener_region_del, +}; + +static int vhost_vdpa_call(struct vhost_dev *dev, unsigned long int request, + void *arg) +{ + struct vhost_vdpa *v = dev->opaque; + int fd = v->device_fd; + + assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA); + + return ioctl(fd, request, arg); +} + +static void vhost_vdpa_add_status(struct vhost_dev *dev, uint8_t status) +{ + uint8_t s; + + if (vhost_vdpa_call(dev, VHOST_VDPA_GET_STATUS, &s)) { + return; + } + + s |= status; + + vhost_vdpa_call(dev, VHOST_VDPA_SET_STATUS, &s); +} + +static int vhost_vdpa_init(struct vhost_dev *dev, void *opaque) +{ + struct vhost_vdpa *v; + uint64_t features; + assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA); + + v = opaque; + dev->opaque = opaque ; + vhost_vdpa_call(dev, VHOST_GET_FEATURES, &features); + dev->backend_features = features; + v->listener = vhost_vdpa_memory_listener; + v->msg_type = VHOST_IOTLB_MSG_V2; + + vhost_vdpa_add_status(dev, VIRTIO_CONFIG_S_ACKNOWLEDGE | + VIRTIO_CONFIG_S_DRIVER); + + return 0; +} + +static int vhost_vdpa_cleanup(struct vhost_dev *dev) +{ + struct vhost_vdpa *v; + assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA); + v = dev->opaque; + memory_listener_unregister(&v->listener); + + dev->opaque = NULL; + return 0; +} + +static int vhost_vdpa_memslots_limit(struct vhost_dev *dev) +{ + return INT_MAX; +} + +static int vhost_vdpa_set_mem_table(struct vhost_dev *dev, + struct vhost_memory *mem) +{ + + if (mem->padding) { + return -1; + } + + return 0; +} + +static int vhost_vdpa_set_features(struct vhost_dev *dev, + uint64_t features) +{ + int ret; + ret = vhost_vdpa_call(dev, VHOST_SET_FEATURES, &features); + uint8_t status = 0; + if (ret) { + return ret; + } + vhost_vdpa_add_status(dev, VIRTIO_CONFIG_S_FEATURES_OK); + vhost_vdpa_call(dev, VHOST_VDPA_GET_STATUS, &status); + + return !(status & VIRTIO_CONFIG_S_FEATURES_OK); +} + +int vhost_vdpa_get_device_id(struct vhost_dev *dev, + uint32_t *device_id) +{ + return vhost_vdpa_call(dev, VHOST_VDPA_GET_DEVICE_ID, device_id); +} + +static int vhost_vdpa_reset_device(struct vhost_dev *dev) +{ + uint8_t status = 0; + + return vhost_vdpa_call(dev, VHOST_VDPA_SET_STATUS, &status); +} + +static int vhost_vdpa_get_vq_index(struct vhost_dev *dev, int idx) +{ + assert(idx >= dev->vq_index && idx < dev->vq_index + dev->nvqs); + + return idx - dev->vq_index; +} + +static int vhost_vdpa_set_vring_ready(struct vhost_dev *dev) +{ + int i; + for (i = 0; i < dev->nvqs; ++i) { + struct vhost_vring_state state = { + .index = dev->vq_index + i, + .num = 1, + }; + vhost_vdpa_call(dev, VHOST_VDPA_SET_VRING_ENABLE, &state); + } + return 0; +} + +static int vhost_vdpa_set_config(struct vhost_dev *dev, const uint8_t *data, + uint32_t offset, uint32_t size, + uint32_t flags) +{ + struct vhost_vdpa_config *config; + int ret; + unsigned long config_size = offsetof(struct vhost_vdpa_config, buf); + config = g_malloc(size + config_size); + if (config == NULL) { + return -1; + } + config->off = offset; + config->len = size; + memcpy(config->buf, data, size); + ret = vhost_vdpa_call(dev, VHOST_VDPA_SET_CONFIG, config); + g_free(config); + return ret; +} + +static int vhost_vdpa_get_config(struct vhost_dev *dev, uint8_t *config, + uint32_t config_len) +{ + struct vhost_vdpa_config *v_config; + unsigned long config_size = offsetof(struct vhost_vdpa_config, buf); + int ret; + + v_config = g_malloc(config_len + config_size); + if (v_config == NULL) { + return -1; + } + v_config->len = config_len; + v_config->off = 0; + ret = vhost_vdpa_call(dev, VHOST_VDPA_GET_CONFIG, v_config); + memcpy(config, v_config->buf, config_len); + g_free(v_config); + return ret; + } + +static int vhost_vdpa_dev_start(struct vhost_dev *dev, bool started) +{ + struct vhost_vdpa *v = dev->opaque; + if (started) { + uint8_t status = 0; + memory_listener_register(&v->listener, &address_space_memory); + vhost_vdpa_set_vring_ready(dev); + vhost_vdpa_add_status(dev, VIRTIO_CONFIG_S_DRIVER_OK); + vhost_vdpa_call(dev, VHOST_VDPA_GET_STATUS, &status); + + return !(status & VIRTIO_CONFIG_S_DRIVER_OK); + } else { + vhost_vdpa_reset_device(dev); + vhost_vdpa_add_status(dev, VIRTIO_CONFIG_S_ACKNOWLEDGE | + VIRTIO_CONFIG_S_DRIVER); + memory_listener_unregister(&v->listener); + + return 0; + } +} + +static int vhost_vdpa_set_log_base(struct vhost_dev *dev, uint64_t base, + struct vhost_log *log) +{ + return vhost_vdpa_call(dev, VHOST_SET_LOG_BASE, &base); +} + +static int vhost_vdpa_set_vring_addr(struct vhost_dev *dev, + struct vhost_vring_addr *addr) +{ + return vhost_vdpa_call(dev, VHOST_SET_VRING_ADDR, addr); +} + +static int vhost_vdpa_set_vring_num(struct vhost_dev *dev, + struct vhost_vring_state *ring) +{ + return vhost_vdpa_call(dev, VHOST_SET_VRING_NUM, ring); +} + +static int vhost_vdpa_set_vring_base(struct vhost_dev *dev, + struct vhost_vring_state *ring) +{ + return vhost_vdpa_call(dev, VHOST_SET_VRING_BASE, ring); +} + +static int vhost_vdpa_get_vring_base(struct vhost_dev *dev, + struct vhost_vring_state *ring) +{ + return vhost_vdpa_call(dev, VHOST_GET_VRING_BASE, ring); +} + +static int vhost_vdpa_set_vring_kick(struct vhost_dev *dev, + struct vhost_vring_file *file) +{ + return vhost_vdpa_call(dev, VHOST_SET_VRING_KICK, file); +} + +static int vhost_vdpa_set_vring_call(struct vhost_dev *dev, + struct vhost_vring_file *file) +{ + return vhost_vdpa_call(dev, VHOST_SET_VRING_CALL, file); +} + +static int vhost_vdpa_get_features(struct vhost_dev *dev, + uint64_t *features) +{ + return vhost_vdpa_call(dev, VHOST_GET_FEATURES, features); +} + +static int vhost_vdpa_set_owner(struct vhost_dev *dev) +{ + return vhost_vdpa_call(dev, VHOST_SET_OWNER, NULL); +} + +static int vhost_vdpa_vq_get_addr(struct vhost_dev *dev, + struct vhost_vring_addr *addr, struct vhost_virtqueue *vq) +{ + assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA); + addr->desc_user_addr = (uint64_t)(unsigned long)vq->desc_phys; + addr->avail_user_addr = (uint64_t)(unsigned long)vq->avail_phys; + addr->used_user_addr = (uint64_t)(unsigned long)vq->used_phys; + return 0; +} + +static bool vhost_vdpa_force_iommu(struct vhost_dev *dev) +{ + return true; +} + +const VhostOps vdpa_ops = { + .backend_type = VHOST_BACKEND_TYPE_VDPA, + .vhost_backend_init = vhost_vdpa_init, + .vhost_backend_cleanup = vhost_vdpa_cleanup, + .vhost_set_log_base = vhost_vdpa_set_log_base, + .vhost_set_vring_addr = vhost_vdpa_set_vring_addr, + .vhost_set_vring_num = vhost_vdpa_set_vring_num, + .vhost_set_vring_base = vhost_vdpa_set_vring_base, + .vhost_get_vring_base = vhost_vdpa_get_vring_base, + .vhost_set_vring_kick = vhost_vdpa_set_vring_kick, + .vhost_set_vring_call = vhost_vdpa_set_vring_call, + .vhost_get_features = vhost_vdpa_get_features, + .vhost_set_owner = vhost_vdpa_set_owner, + .vhost_set_vring_endian = NULL, + .vhost_backend_memslots_limit = vhost_vdpa_memslots_limit, + .vhost_set_mem_table = vhost_vdpa_set_mem_table, + .vhost_set_features = vhost_vdpa_set_features, + .vhost_reset_device = vhost_vdpa_reset_device, + .vhost_get_vq_index = vhost_vdpa_get_vq_index, + .vhost_get_config = vhost_vdpa_get_config, + .vhost_set_config = vhost_vdpa_set_config, + .vhost_requires_shm_log = NULL, + .vhost_migration_done = NULL, + .vhost_backend_can_merge = NULL, + .vhost_net_set_mtu = NULL, + .vhost_set_iotlb_callback = NULL, + .vhost_send_device_iotlb_msg = NULL, + .vhost_dev_start = vhost_vdpa_dev_start, + .vhost_get_device_id = vhost_vdpa_get_device_id, + .vhost_vq_get_addr = vhost_vdpa_vq_get_addr, + .vhost_force_iommu = vhost_vdpa_force_iommu, +}; |