diff options
Diffstat (limited to 'contrib/libvhost-user')
| -rw-r--r-- | contrib/libvhost-user/libvhost-user.c | 351 | ||||
| -rw-r--r-- | contrib/libvhost-user/libvhost-user.h | 24 |
2 files changed, 299 insertions, 76 deletions
diff --git a/contrib/libvhost-user/libvhost-user.c b/contrib/libvhost-user/libvhost-user.c index 3bca996c62..d315db1396 100644 --- a/contrib/libvhost-user/libvhost-user.c +++ b/contrib/libvhost-user/libvhost-user.c @@ -137,6 +137,9 @@ vu_request_to_string(unsigned int req) REQ(VHOST_USER_SET_INFLIGHT_FD), REQ(VHOST_USER_GPU_SET_SOCKET), REQ(VHOST_USER_VRING_KICK), + REQ(VHOST_USER_GET_MAX_MEM_SLOTS), + REQ(VHOST_USER_ADD_MEM_REG), + REQ(VHOST_USER_REM_MEM_REG), REQ(VHOST_USER_MAX), }; #undef REQ @@ -266,7 +269,7 @@ have_userfault(void) static bool vu_message_read(VuDev *dev, int conn_fd, VhostUserMsg *vmsg) { - char control[CMSG_SPACE(VHOST_MEMORY_MAX_NREGIONS * sizeof(int))] = { }; + char control[CMSG_SPACE(VHOST_MEMORY_BASELINE_NREGIONS * sizeof(int))] = {}; struct iovec iov = { .iov_base = (char *)vmsg, .iov_len = VHOST_USER_HDR_SIZE, @@ -337,7 +340,7 @@ vu_message_write(VuDev *dev, int conn_fd, VhostUserMsg *vmsg) { int rc; uint8_t *p = (uint8_t *)vmsg; - char control[CMSG_SPACE(VHOST_MEMORY_MAX_NREGIONS * sizeof(int))] = { }; + char control[CMSG_SPACE(VHOST_MEMORY_BASELINE_NREGIONS * sizeof(int))] = {}; struct iovec iov = { .iov_base = (char *)vmsg, .iov_len = VHOST_USER_HDR_SIZE, @@ -350,7 +353,7 @@ vu_message_write(VuDev *dev, int conn_fd, VhostUserMsg *vmsg) struct cmsghdr *cmsg; memset(control, 0, sizeof(control)); - assert(vmsg->fd_num <= VHOST_MEMORY_MAX_NREGIONS); + assert(vmsg->fd_num <= VHOST_MEMORY_BASELINE_NREGIONS); if (vmsg->fd_num > 0) { size_t fdsize = vmsg->fd_num * sizeof(int); msg.msg_controllen = CMSG_SPACE(fdsize); @@ -495,6 +498,16 @@ static bool vu_get_features_exec(VuDev *dev, VhostUserMsg *vmsg) { vmsg->payload.u64 = + /* + * The following VIRTIO feature bits are supported by our virtqueue + * implementation: + */ + 1ULL << VIRTIO_F_NOTIFY_ON_EMPTY | + 1ULL << VIRTIO_RING_F_INDIRECT_DESC | + 1ULL << VIRTIO_RING_F_EVENT_IDX | + 1ULL << VIRTIO_F_VERSION_1 | + + /* vhost-user feature bits */ 1ULL << VHOST_F_LOG_ALL | 1ULL << VHOST_USER_F_PROTOCOL_FEATURES; @@ -584,6 +597,244 @@ map_ring(VuDev *dev, VuVirtq *vq) } static bool +generate_faults(VuDev *dev) { + int i; + for (i = 0; i < dev->nregions; i++) { + VuDevRegion *dev_region = &dev->regions[i]; + int ret; +#ifdef UFFDIO_REGISTER + /* + * We should already have an open ufd. Mark each memory + * range as ufd. + * Discard any mapping we have here; note I can't use MADV_REMOVE + * or fallocate to make the hole since I don't want to lose + * data that's already arrived in the shared process. + * TODO: How to do hugepage + */ + ret = madvise((void *)(uintptr_t)dev_region->mmap_addr, + dev_region->size + dev_region->mmap_offset, + MADV_DONTNEED); + if (ret) { + fprintf(stderr, + "%s: Failed to madvise(DONTNEED) region %d: %s\n", + __func__, i, strerror(errno)); + } + /* + * Turn off transparent hugepages so we dont get lose wakeups + * in neighbouring pages. + * TODO: Turn this backon later. + */ + ret = madvise((void *)(uintptr_t)dev_region->mmap_addr, + dev_region->size + dev_region->mmap_offset, + MADV_NOHUGEPAGE); + if (ret) { + /* + * Note: This can happen legally on kernels that are configured + * without madvise'able hugepages + */ + fprintf(stderr, + "%s: Failed to madvise(NOHUGEPAGE) region %d: %s\n", + __func__, i, strerror(errno)); + } + struct uffdio_register reg_struct; + reg_struct.range.start = (uintptr_t)dev_region->mmap_addr; + reg_struct.range.len = dev_region->size + dev_region->mmap_offset; + reg_struct.mode = UFFDIO_REGISTER_MODE_MISSING; + + if (ioctl(dev->postcopy_ufd, UFFDIO_REGISTER, ®_struct)) { + vu_panic(dev, "%s: Failed to userfault region %d " + "@%p + size:%zx offset: %zx: (ufd=%d)%s\n", + __func__, i, + dev_region->mmap_addr, + dev_region->size, dev_region->mmap_offset, + dev->postcopy_ufd, strerror(errno)); + return false; + } + if (!(reg_struct.ioctls & ((__u64)1 << _UFFDIO_COPY))) { + vu_panic(dev, "%s Region (%d) doesn't support COPY", + __func__, i); + return false; + } + DPRINT("%s: region %d: Registered userfault for %" + PRIx64 " + %" PRIx64 "\n", __func__, i, + (uint64_t)reg_struct.range.start, + (uint64_t)reg_struct.range.len); + /* Now it's registered we can let the client at it */ + if (mprotect((void *)(uintptr_t)dev_region->mmap_addr, + dev_region->size + dev_region->mmap_offset, + PROT_READ | PROT_WRITE)) { + vu_panic(dev, "failed to mprotect region %d for postcopy (%s)", + i, strerror(errno)); + return false; + } + /* TODO: Stash 'zero' support flags somewhere */ +#endif + } + + return true; +} + +static bool +vu_add_mem_reg(VuDev *dev, VhostUserMsg *vmsg) { + int i; + bool track_ramblocks = dev->postcopy_listening; + VhostUserMemoryRegion m = vmsg->payload.memreg.region, *msg_region = &m; + VuDevRegion *dev_region = &dev->regions[dev->nregions]; + void *mmap_addr; + + /* + * If we are in postcopy mode and we receive a u64 payload with a 0 value + * we know all the postcopy client bases have been recieved, and we + * should start generating faults. + */ + if (track_ramblocks && + vmsg->size == sizeof(vmsg->payload.u64) && + vmsg->payload.u64 == 0) { + (void)generate_faults(dev); + return false; + } + + DPRINT("Adding region: %d\n", dev->nregions); + DPRINT(" guest_phys_addr: 0x%016"PRIx64"\n", + msg_region->guest_phys_addr); + DPRINT(" memory_size: 0x%016"PRIx64"\n", + msg_region->memory_size); + DPRINT(" userspace_addr 0x%016"PRIx64"\n", + msg_region->userspace_addr); + DPRINT(" mmap_offset 0x%016"PRIx64"\n", + msg_region->mmap_offset); + + dev_region->gpa = msg_region->guest_phys_addr; + dev_region->size = msg_region->memory_size; + dev_region->qva = msg_region->userspace_addr; + dev_region->mmap_offset = msg_region->mmap_offset; + + /* + * We don't use offset argument of mmap() since the + * mapped address has to be page aligned, and we use huge + * pages. + */ + if (track_ramblocks) { + /* + * In postcopy we're using PROT_NONE here to catch anyone + * accessing it before we userfault. + */ + mmap_addr = mmap(0, dev_region->size + dev_region->mmap_offset, + PROT_NONE, MAP_SHARED, + vmsg->fds[0], 0); + } else { + mmap_addr = mmap(0, dev_region->size + dev_region->mmap_offset, + PROT_READ | PROT_WRITE, MAP_SHARED, vmsg->fds[0], + 0); + } + + if (mmap_addr == MAP_FAILED) { + vu_panic(dev, "region mmap error: %s", strerror(errno)); + } else { + dev_region->mmap_addr = (uint64_t)(uintptr_t)mmap_addr; + DPRINT(" mmap_addr: 0x%016"PRIx64"\n", + dev_region->mmap_addr); + } + + close(vmsg->fds[0]); + + if (track_ramblocks) { + /* + * Return the address to QEMU so that it can translate the ufd + * fault addresses back. + */ + msg_region->userspace_addr = (uintptr_t)(mmap_addr + + dev_region->mmap_offset); + + /* Send the message back to qemu with the addresses filled in. */ + vmsg->fd_num = 0; + if (!vu_send_reply(dev, dev->sock, vmsg)) { + vu_panic(dev, "failed to respond to add-mem-region for postcopy"); + return false; + } + + DPRINT("Successfully added new region in postcopy\n"); + dev->nregions++; + return false; + + } else { + for (i = 0; i < dev->max_queues; i++) { + if (dev->vq[i].vring.desc) { + if (map_ring(dev, &dev->vq[i])) { + vu_panic(dev, "remapping queue %d for new memory region", + i); + } + } + } + + DPRINT("Successfully added new region\n"); + dev->nregions++; + vmsg_set_reply_u64(vmsg, 0); + return true; + } +} + +static inline bool reg_equal(VuDevRegion *vudev_reg, + VhostUserMemoryRegion *msg_reg) +{ + if (vudev_reg->gpa == msg_reg->guest_phys_addr && + vudev_reg->qva == msg_reg->userspace_addr && + vudev_reg->size == msg_reg->memory_size) { + return true; + } + + return false; +} + +static bool +vu_rem_mem_reg(VuDev *dev, VhostUserMsg *vmsg) { + int i, j; + bool found = false; + VuDevRegion shadow_regions[VHOST_USER_MAX_RAM_SLOTS] = {}; + VhostUserMemoryRegion m = vmsg->payload.memreg.region, *msg_region = &m; + + DPRINT("Removing region:\n"); + DPRINT(" guest_phys_addr: 0x%016"PRIx64"\n", + msg_region->guest_phys_addr); + DPRINT(" memory_size: 0x%016"PRIx64"\n", + msg_region->memory_size); + DPRINT(" userspace_addr 0x%016"PRIx64"\n", + msg_region->userspace_addr); + DPRINT(" mmap_offset 0x%016"PRIx64"\n", + msg_region->mmap_offset); + + for (i = 0, j = 0; i < dev->nregions; i++) { + if (!reg_equal(&dev->regions[i], msg_region)) { + shadow_regions[j].gpa = dev->regions[i].gpa; + shadow_regions[j].size = dev->regions[i].size; + shadow_regions[j].qva = dev->regions[i].qva; + shadow_regions[j].mmap_offset = dev->regions[i].mmap_offset; + j++; + } else { + found = true; + VuDevRegion *r = &dev->regions[i]; + void *m = (void *) (uintptr_t) r->mmap_addr; + + if (m) { + munmap(m, r->size + r->mmap_offset); + } + } + } + + if (found) { + memcpy(dev->regions, shadow_regions, + sizeof(VuDevRegion) * VHOST_USER_MAX_RAM_SLOTS); + DPRINT("Successfully removed a region\n"); + dev->nregions--; + vmsg_set_reply_u64(vmsg, 0); + } else { + vu_panic(dev, "Specified region not found\n"); + } + + return true; +} + +static bool vu_set_mem_table_exec_postcopy(VuDev *dev, VhostUserMsg *vmsg) { int i; @@ -655,74 +906,7 @@ vu_set_mem_table_exec_postcopy(VuDev *dev, VhostUserMsg *vmsg) } /* OK, now we can go and register the memory and generate faults */ - for (i = 0; i < dev->nregions; i++) { - VuDevRegion *dev_region = &dev->regions[i]; - int ret; -#ifdef UFFDIO_REGISTER - /* We should already have an open ufd. Mark each memory - * range as ufd. - * Discard any mapping we have here; note I can't use MADV_REMOVE - * or fallocate to make the hole since I don't want to lose - * data that's already arrived in the shared process. - * TODO: How to do hugepage - */ - ret = madvise((void *)(uintptr_t)dev_region->mmap_addr, - dev_region->size + dev_region->mmap_offset, - MADV_DONTNEED); - if (ret) { - fprintf(stderr, - "%s: Failed to madvise(DONTNEED) region %d: %s\n", - __func__, i, strerror(errno)); - } - /* Turn off transparent hugepages so we dont get lose wakeups - * in neighbouring pages. - * TODO: Turn this backon later. - */ - ret = madvise((void *)(uintptr_t)dev_region->mmap_addr, - dev_region->size + dev_region->mmap_offset, - MADV_NOHUGEPAGE); - if (ret) { - /* Note: This can happen legally on kernels that are configured - * without madvise'able hugepages - */ - fprintf(stderr, - "%s: Failed to madvise(NOHUGEPAGE) region %d: %s\n", - __func__, i, strerror(errno)); - } - struct uffdio_register reg_struct; - reg_struct.range.start = (uintptr_t)dev_region->mmap_addr; - reg_struct.range.len = dev_region->size + dev_region->mmap_offset; - reg_struct.mode = UFFDIO_REGISTER_MODE_MISSING; - - if (ioctl(dev->postcopy_ufd, UFFDIO_REGISTER, ®_struct)) { - vu_panic(dev, "%s: Failed to userfault region %d " - "@%p + size:%zx offset: %zx: (ufd=%d)%s\n", - __func__, i, - dev_region->mmap_addr, - dev_region->size, dev_region->mmap_offset, - dev->postcopy_ufd, strerror(errno)); - return false; - } - if (!(reg_struct.ioctls & ((__u64)1 << _UFFDIO_COPY))) { - vu_panic(dev, "%s Region (%d) doesn't support COPY", - __func__, i); - return false; - } - DPRINT("%s: region %d: Registered userfault for %" - PRIx64 " + %" PRIx64 "\n", __func__, i, - (uint64_t)reg_struct.range.start, - (uint64_t)reg_struct.range.len); - /* Now it's registered we can let the client at it */ - if (mprotect((void *)(uintptr_t)dev_region->mmap_addr, - dev_region->size + dev_region->mmap_offset, - PROT_READ | PROT_WRITE)) { - vu_panic(dev, "failed to mprotect region %d for postcopy (%s)", - i, strerror(errno)); - return false; - } - /* TODO: Stash 'zero' support flags somewhere */ -#endif - } + (void)generate_faults(dev); return false; } @@ -1220,7 +1404,8 @@ vu_get_protocol_features_exec(VuDev *dev, VhostUserMsg *vmsg) 1ULL << VHOST_USER_PROTOCOL_F_SLAVE_REQ | 1ULL << VHOST_USER_PROTOCOL_F_HOST_NOTIFIER | 1ULL << VHOST_USER_PROTOCOL_F_SLAVE_SEND_FD | - 1ULL << VHOST_USER_PROTOCOL_F_REPLY_ACK; + 1ULL << VHOST_USER_PROTOCOL_F_REPLY_ACK | + 1ULL << VHOST_USER_PROTOCOL_F_CONFIGURE_MEM_SLOTS; if (have_userfault()) { features |= 1ULL << VHOST_USER_PROTOCOL_F_PAGEFAULT; @@ -1554,6 +1739,22 @@ vu_handle_vring_kick(VuDev *dev, VhostUserMsg *vmsg) return false; } +static bool vu_handle_get_max_memslots(VuDev *dev, VhostUserMsg *vmsg) +{ + vmsg->flags = VHOST_USER_REPLY_MASK | VHOST_USER_VERSION; + vmsg->size = sizeof(vmsg->payload.u64); + vmsg->payload.u64 = VHOST_USER_MAX_RAM_SLOTS; + vmsg->fd_num = 0; + + if (!vu_message_write(dev, dev->sock, vmsg)) { + vu_panic(dev, "Failed to send max ram slots: %s\n", strerror(errno)); + } + + DPRINT("u64: 0x%016"PRIx64"\n", (uint64_t) VHOST_USER_MAX_RAM_SLOTS); + + return false; +} + static bool vu_process_message(VuDev *dev, VhostUserMsg *vmsg) { @@ -1638,6 +1839,12 @@ vu_process_message(VuDev *dev, VhostUserMsg *vmsg) return vu_set_inflight_fd(dev, vmsg); case VHOST_USER_VRING_KICK: return vu_handle_vring_kick(dev, vmsg); + case VHOST_USER_GET_MAX_MEM_SLOTS: + return vu_handle_get_max_memslots(dev, vmsg); + case VHOST_USER_ADD_MEM_REG: + return vu_add_mem_reg(dev, vmsg); + case VHOST_USER_REM_MEM_REG: + return vu_rem_mem_reg(dev, vmsg); default: vmsg_close_fds(vmsg); vu_panic(dev, "Unhandled request: %d", vmsg->request); diff --git a/contrib/libvhost-user/libvhost-user.h b/contrib/libvhost-user/libvhost-user.h index f30394fab6..844c37c648 100644 --- a/contrib/libvhost-user/libvhost-user.h +++ b/contrib/libvhost-user/libvhost-user.h @@ -28,7 +28,13 @@ #define VIRTQUEUE_MAX_SIZE 1024 -#define VHOST_MEMORY_MAX_NREGIONS 8 +#define VHOST_MEMORY_BASELINE_NREGIONS 8 + +/* + * Set a reasonable maximum number of ram slots, which will be supported by + * any architecture. + */ +#define VHOST_USER_MAX_RAM_SLOTS 32 typedef enum VhostSetConfigType { VHOST_SET_CONFIG_TYPE_MASTER = 0, @@ -55,6 +61,7 @@ enum VhostUserProtocolFeature { VHOST_USER_PROTOCOL_F_HOST_NOTIFIER = 11, VHOST_USER_PROTOCOL_F_INFLIGHT_SHMFD = 12, VHOST_USER_PROTOCOL_F_INBAND_NOTIFICATIONS = 14, + VHOST_USER_PROTOCOL_F_CONFIGURE_MEM_SLOTS = 15, VHOST_USER_PROTOCOL_F_MAX }; @@ -97,6 +104,9 @@ typedef enum VhostUserRequest { VHOST_USER_SET_INFLIGHT_FD = 32, VHOST_USER_GPU_SET_SOCKET = 33, VHOST_USER_VRING_KICK = 35, + VHOST_USER_GET_MAX_MEM_SLOTS = 36, + VHOST_USER_ADD_MEM_REG = 37, + VHOST_USER_REM_MEM_REG = 38, VHOST_USER_MAX } VhostUserRequest; @@ -120,9 +130,14 @@ typedef struct VhostUserMemoryRegion { typedef struct VhostUserMemory { uint32_t nregions; uint32_t padding; - VhostUserMemoryRegion regions[VHOST_MEMORY_MAX_NREGIONS]; + VhostUserMemoryRegion regions[VHOST_MEMORY_BASELINE_NREGIONS]; } VhostUserMemory; +typedef struct VhostUserMemRegMsg { + uint32_t padding; + VhostUserMemoryRegion region; +} VhostUserMemRegMsg; + typedef struct VhostUserLog { uint64_t mmap_size; uint64_t mmap_offset; @@ -175,13 +190,14 @@ typedef struct VhostUserMsg { struct vhost_vring_state state; struct vhost_vring_addr addr; VhostUserMemory memory; + VhostUserMemRegMsg memreg; VhostUserLog log; VhostUserConfig config; VhostUserVringArea area; VhostUserInflight inflight; } payload; - int fds[VHOST_MEMORY_MAX_NREGIONS]; + int fds[VHOST_MEMORY_BASELINE_NREGIONS]; int fd_num; uint8_t *data; } VU_PACKED VhostUserMsg; @@ -359,7 +375,7 @@ typedef struct VuDevInflightInfo { struct VuDev { int sock; uint32_t nregions; - VuDevRegion regions[VHOST_MEMORY_MAX_NREGIONS]; + VuDevRegion regions[VHOST_USER_MAX_RAM_SLOTS]; VuVirtq *vq; VuDevInflightInfo inflight_info; int log_call_fd; |