Support adding individual regions in libvhost-user

When the VHOST_USER_PROTOCOL_F_CONFIGURE_MEM_SLOTS is enabled, qemu will
transmit memory regions to a backend individually using the new message
VHOST_USER_ADD_MEM_REG. With this change vhost-user backends built with
libvhost-user can now map in new memory regions when VHOST_USER_ADD_MEM_REG
messages are received.

Qemu only sends VHOST_USER_ADD_MEM_REG messages when the
VHOST_USER_PROTOCOL_F_CONFIGURE_MEM_SLOTS feature is negotiated, and
since it is not yet supported in libvhost-user, this new functionality
is not yet used.

Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <1588533678-23450-9-git-send-email-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

contrib/libvhost-user/libvhost-user.c

@@ -138,6 +138,7 @@ vu_request_to_string(unsigned int req)
         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_MAX),
     };
 #undef REQ
@@ -662,6 +663,106 @@ generate_faults(VuDev *dev) {
     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 bool
 vu_set_mem_table_exec_postcopy(VuDev *dev, VhostUserMsg *vmsg)
 {
@@ -1668,6 +1769,8 @@ vu_process_message(VuDev *dev, VhostUserMsg *vmsg)
         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);
     default:
         vmsg_close_fds(vmsg);
         vu_panic(dev, "Unhandled request: %d", vmsg->request);

contrib/libvhost-user/libvhost-user.h

@@ -98,6 +98,7 @@ typedef enum VhostUserRequest {
     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_MAX
 } VhostUserRequest;
 
@@ -124,6 +125,11 @@ typedef struct VhostUserMemory {
     VhostUserMemoryRegion regions[VHOST_MEMORY_MAX_NREGIONS];
 } VhostUserMemory;
 
+typedef struct VhostUserMemRegMsg {
+    uint32_t padding;
+    VhostUserMemoryRegion region;
+} VhostUserMemRegMsg;
+
 typedef struct VhostUserLog {
     uint64_t mmap_size;
     uint64_t mmap_offset;
@@ -176,6 +182,7 @@ typedef struct VhostUserMsg {
         struct vhost_vring_state state;
         struct vhost_vring_addr addr;
         VhostUserMemory memory;
+        VhostUserMemRegMsg memreg;
         VhostUserLog log;
         VhostUserConfig config;
         VhostUserVringArea area;