uint64_t num_packets;
/* pages sent through this channel */
uint64_t num_pages;
- /* syncs main thread and channels */
- QemuSemaphore sem_sync;
} MultiFDSendParams;
typedef struct {
struct {
MultiFDSendParams *params;
- /* number of created threads */
- int count;
/* array of pages to sent */
MultiFDPages_t *pages;
/* syncs main thread and channels */
* false.
*/
-static void multifd_send_pages(void)
+static int multifd_send_pages(void)
{
int i;
static int next_channel;
p = &multifd_send_state->params[i];
qemu_mutex_lock(&p->mutex);
+ if (p->quit) {
+ error_report("%s: channel %d has already quit!", __func__, i);
+ qemu_mutex_unlock(&p->mutex);
+ return -1;
+ }
if (!p->pending_job) {
p->pending_job++;
next_channel = (i + 1) % migrate_multifd_channels();
ram_counters.transferred += transferred;;
qemu_mutex_unlock(&p->mutex);
qemu_sem_post(&p->sem);
+
+ return 1;
}
-static void multifd_queue_page(RAMBlock *block, ram_addr_t offset)
+static int multifd_queue_page(RAMBlock *block, ram_addr_t offset)
{
MultiFDPages_t *pages = multifd_send_state->pages;
pages->used++;
if (pages->used < pages->allocated) {
- return;
+ return 1;
}
}
- multifd_send_pages();
+ if (multifd_send_pages() < 0) {
+ return -1;
+ }
if (pages->block != block) {
- multifd_queue_page(block, offset);
+ return multifd_queue_page(block, offset);
}
+
+ return 1;
}
static void multifd_send_terminate_threads(Error *err)
p->c = NULL;
qemu_mutex_destroy(&p->mutex);
qemu_sem_destroy(&p->sem);
- qemu_sem_destroy(&p->sem_sync);
g_free(p->name);
p->name = NULL;
multifd_pages_clear(p->pages);
return;
}
if (multifd_send_state->pages->used) {
- multifd_send_pages();
+ if (multifd_send_pages() < 0) {
+ error_report("%s: multifd_send_pages fail", __func__);
+ return;
+ }
}
for (i = 0; i < migrate_multifd_channels(); i++) {
MultiFDSendParams *p = &multifd_send_state->params[i];
qemu_mutex_lock(&p->mutex);
+ if (p->quit) {
+ error_report("%s: channel %d has already quit", __func__, i);
+ qemu_mutex_unlock(&p->mutex);
+ return;
+ }
+
p->packet_num = multifd_send_state->packet_num++;
p->flags |= MULTIFD_FLAG_SYNC;
p->pending_job++;
p->running = true;
qemu_thread_create(&p->thread, p->name, multifd_send_thread, p,
QEMU_THREAD_JOINABLE);
-
- atomic_inc(&multifd_send_state->count);
}
}
thread_count = migrate_multifd_channels();
multifd_send_state = g_malloc0(sizeof(*multifd_send_state));
multifd_send_state->params = g_new0(MultiFDSendParams, thread_count);
- atomic_set(&multifd_send_state->count, 0);
multifd_send_state->pages = multifd_pages_init(page_count);
qemu_sem_init(&multifd_send_state->sem_sync, 0);
qemu_sem_init(&multifd_send_state->channels_ready, 0);
qemu_mutex_init(&p->mutex);
qemu_sem_init(&p->sem, 0);
- qemu_sem_init(&p->sem_sync, 0);
p->quit = false;
p->pending_job = 0;
p->id = i;
trace_multifd_recv_sync_main_wait(p->id);
qemu_sem_wait(&multifd_recv_state->sem_sync);
+ }
+ for (i = 0; i < migrate_multifd_channels(); i++) {
+ MultiFDRecvParams *p = &multifd_recv_state->params[i];
+
qemu_mutex_lock(&p->mutex);
if (multifd_recv_state->packet_num < p->packet_num) {
multifd_recv_state->packet_num = p->packet_num;
}
qemu_mutex_unlock(&p->mutex);
- }
- for (i = 0; i < migrate_multifd_channels(); i++) {
- MultiFDRecvParams *p = &multifd_recv_state->params[i];
-
trace_multifd_recv_sync_main_signal(p->id);
qemu_sem_post(&p->sem_sync);
}
encoded_len = xbzrle_encode_buffer(prev_cached_page, XBZRLE.current_buf,
TARGET_PAGE_SIZE, XBZRLE.encoded_buf,
TARGET_PAGE_SIZE);
+
+ /*
+ * Update the cache contents, so that it corresponds to the data
+ * sent, in all cases except where we skip the page.
+ */
+ if (!last_stage && encoded_len != 0) {
+ memcpy(prev_cached_page, XBZRLE.current_buf, TARGET_PAGE_SIZE);
+ /*
+ * In the case where we couldn't compress, ensure that the caller
+ * sends the data from the cache, since the guest might have
+ * changed the RAM since we copied it.
+ */
+ *current_data = prev_cached_page;
+ }
+
if (encoded_len == 0) {
trace_save_xbzrle_page_skipping();
return 0;
} else if (encoded_len == -1) {
trace_save_xbzrle_page_overflow();
xbzrle_counters.overflow++;
- /* update data in the cache */
- if (!last_stage) {
- memcpy(prev_cached_page, *current_data, TARGET_PAGE_SIZE);
- *current_data = prev_cached_page;
- }
return -1;
}
- /* we need to update the data in the cache, in order to get the same data */
- if (!last_stage) {
- memcpy(prev_cached_page, XBZRLE.current_buf, TARGET_PAGE_SIZE);
- }
-
/* Send XBZRLE based compressed page */
bytes_xbzrle = save_page_header(rs, rs->f, block,
offset | RAM_SAVE_FLAG_XBZRLE);
bool ret;
qemu_mutex_lock(&rs->bitmap_mutex);
+
+ /*
+ * Clear dirty bitmap if needed. This _must_ be called before we
+ * send any of the page in the chunk because we need to make sure
+ * we can capture further page content changes when we sync dirty
+ * log the next time. So as long as we are going to send any of
+ * the page in the chunk we clear the remote dirty bitmap for all.
+ * Clearing it earlier won't be a problem, but too late will.
+ */
+ if (rb->clear_bmap && clear_bmap_test_and_clear(rb, page)) {
+ uint8_t shift = rb->clear_bmap_shift;
+ hwaddr size = 1ULL << (TARGET_PAGE_BITS + shift);
+ hwaddr start = (page << TARGET_PAGE_BITS) & (-size);
+
+ /*
+ * CLEAR_BITMAP_SHIFT_MIN should always guarantee this... this
+ * can make things easier sometimes since then start address
+ * of the small chunk will always be 64 pages aligned so the
+ * bitmap will always be aligned to unsigned long. We should
+ * even be able to remove this restriction but I'm simply
+ * keeping it.
+ */
+ assert(shift >= 6);
+ trace_migration_bitmap_clear_dirty(rb->idstr, start, size, page);
+ memory_region_clear_dirty_bitmap(rb->mr, start, size);
+ }
+
ret = test_and_clear_bit(page, rb->bmap);
if (ret) {
return ret;
}
+/* Called with RCU critical section */
static void migration_bitmap_sync_range(RAMState *rs, RAMBlock *rb,
ram_addr_t length)
{
static int ram_save_multifd_page(RAMState *rs, RAMBlock *block,
ram_addr_t offset)
{
- multifd_queue_page(block, offset);
+ if (multifd_queue_page(block, offset) < 0) {
+ return -1;
+ }
ram_counters.normal++;
return 1;
}
/**
- * get_queued_page: unqueue a page from the postocpy requests
+ * get_queued_page: unqueue a page from the postcopy requests
*
* Skips pages that are already sent (!dirty)
*
*/
pss->block = block;
pss->page = offset >> TARGET_PAGE_BITS;
+
+ /*
+ * This unqueued page would break the "one round" check, even is
+ * really rare.
+ */
+ pss->complete_round = false;
}
return !!block;
memory_global_dirty_log_stop();
RAMBLOCK_FOREACH_NOT_IGNORED(block) {
+ g_free(block->clear_bmap);
+ block->clear_bmap = NULL;
g_free(block->bmap);
block->bmap = NULL;
g_free(block->unsentmap);
*
* @ms: current migration state
* @pds: state for postcopy
- * @start: RAMBlock starting page
- * @length: RAMBlock size
+ * @block: RAMBlock to discard
*/
static int postcopy_send_discard_bm_ram(MigrationState *ms,
PostcopyDiscardState *pds,
}
/**
- * postcopy_chuck_hostpages: discrad any partially sent host page
+ * postcopy_chunk_hostpages: discard any partially sent host page
*
* Utility for the outgoing postcopy code.
*
/*
* Count the total number of pages used by ram blocks not including any
* gaps due to alignment or unplugs.
+ * This must match with the initial values of dirty bitmap.
*/
(*rsp)->migration_dirty_pages = ram_bytes_total() >> TARGET_PAGE_BITS;
-
ram_state_reset(*rsp);
return 0;
static void ram_list_init_bitmaps(void)
{
+ MigrationState *ms = migrate_get_current();
RAMBlock *block;
unsigned long pages;
+ uint8_t shift;
/* Skip setting bitmap if there is no RAM */
if (ram_bytes_total()) {
+ shift = ms->clear_bitmap_shift;
+ if (shift > CLEAR_BITMAP_SHIFT_MAX) {
+ error_report("clear_bitmap_shift (%u) too big, using "
+ "max value (%u)", shift, CLEAR_BITMAP_SHIFT_MAX);
+ shift = CLEAR_BITMAP_SHIFT_MAX;
+ } else if (shift < CLEAR_BITMAP_SHIFT_MIN) {
+ error_report("clear_bitmap_shift (%u) too small, using "
+ "min value (%u)", shift, CLEAR_BITMAP_SHIFT_MIN);
+ shift = CLEAR_BITMAP_SHIFT_MIN;
+ }
+
RAMBLOCK_FOREACH_NOT_IGNORED(block) {
pages = block->max_length >> TARGET_PAGE_BITS;
+ /*
+ * The initial dirty bitmap for migration must be set with all
+ * ones to make sure we'll migrate every guest RAM page to
+ * destination.
+ * Here we set RAMBlock.bmap all to 1 because when rebegin a
+ * new migration after a failed migration, ram_list.
+ * dirty_memory[DIRTY_MEMORY_MIGRATION] don't include the whole
+ * guest memory.
+ */
block->bmap = bitmap_new(pages);
bitmap_set(block->bmap, 0, pages);
+ block->clear_bmap_shift = shift;
+ block->clear_bmap = bitmap_new(clear_bmap_size(pages, shift));
if (migrate_postcopy_ram()) {
block->unsentmap = bitmap_new(pages);
bitmap_set(block->unsentmap, 0, pages);
}
if (migrate_ignore_shared()) {
qemu_put_be64(f, block->mr->addr);
- qemu_put_byte(f, ramblock_is_ignored(block) ? 1 : 0);
}
}
*/
ram_control_after_iterate(f, RAM_CONTROL_ROUND);
- multifd_send_sync_main();
out:
+ multifd_send_sync_main();
qemu_put_be64(f, RAM_SAVE_FLAG_EOS);
qemu_fflush(f);
ram_counters.transferred += 8;
}
if (migrate_ignore_shared()) {
hwaddr addr = qemu_get_be64(f);
- bool ignored = qemu_get_byte(f);
- if (ignored != ramblock_is_ignored(block)) {
- error_report("RAM block %s should %s be migrated",
- id, ignored ? "" : "not");
- ret = -EINVAL;
- }
if (ramblock_is_ignored(block) &&
block->mr->addr != addr) {
error_report("Mismatched GPAs for block %s "
projects / qemu.git / blobdiff