Merge remote-tracking branch 'remotes/stefanha/tags/block-pull-reques…

…t' into staging # gpg: Signature made Fri Jun 12 15:57:47 2015 BST using RSA key ID 81AB73C8 # gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>" # gpg: aka "Stefan Hajnoczi <stefanha@gmail.com>" * remotes/stefanha/tags/block-pull-request: qemu-iotests: expand test 093 to support group throttling throttle: Update throttle infrastructure copyright throttle: add the name of the ThrottleGroup to BlockDeviceInfo throttle: acquire the ThrottleGroup lock in bdrv_swap() throttle: Add throttle group support throttle: Add throttle group infrastructure tests throttle: Add throttle group infrastructure throttle: Extract timers from ThrottleState into a separate structure raw-posix: Fix .bdrv_co_get_block_status() for unaligned image size Revert "iothread: release iothread around aio_poll" Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
mujehoxe · Jun 12, 2015 · 8aeaa05 · 8aeaa05
2 parents 0a2df85 + 2db33f8
commit 8aeaa05
Show file tree

Hide file tree

Showing 21 changed files with 956 additions and 221 deletions.
diff --git a/async.c b/async.c
@@ -280,6 +280,12 @@ static void aio_timerlist_notify(void *opaque)
     aio_notify(opaque);
 }
 
+static void aio_rfifolock_cb(void *opaque)
+{
+    /* Kick owner thread in case they are blocked in aio_poll() */
+    aio_notify(opaque);
+}
+
 AioContext *aio_context_new(Error **errp)
 {
     int ret;
@@ -297,7 +303,7 @@ AioContext *aio_context_new(Error **errp)
                            event_notifier_test_and_clear);
     ctx->thread_pool = NULL;
     qemu_mutex_init(&ctx->bh_lock);
-    rfifolock_init(&ctx->lock, NULL, NULL);
+    rfifolock_init(&ctx->lock, aio_rfifolock_cb, ctx);
     timerlistgroup_init(&ctx->tlg, aio_timerlist_notify, ctx);
 
     return ctx;

diff --git a/block.c b/block.c
@@ -36,6 +36,7 @@
 #include "qmp-commands.h"
 #include "qemu/timer.h"
 #include "qapi-event.h"
+#include "block/throttle-groups.h"
 
 #ifdef CONFIG_BSD
 #include <sys/types.h>
@@ -1822,12 +1823,18 @@ static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
     bs_dest->enable_write_cache = bs_src->enable_write_cache;
 
     /* i/o throttled req */
-    memcpy(&bs_dest->throttle_state,
-           &bs_src->throttle_state,
-           sizeof(ThrottleState));
+    bs_dest->throttle_state     = bs_src->throttle_state,
+    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
+    bs_dest->pending_reqs[0]    = bs_src->pending_reqs[0];
+    bs_dest->pending_reqs[1]    = bs_src->pending_reqs[1];
     bs_dest->throttled_reqs[0]  = bs_src->throttled_reqs[0];
     bs_dest->throttled_reqs[1]  = bs_src->throttled_reqs[1];
-    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
+    memcpy(&bs_dest->round_robin,
+           &bs_src->round_robin,
+           sizeof(bs_dest->round_robin));
+    memcpy(&bs_dest->throttle_timers,
+           &bs_src->throttle_timers,
+           sizeof(ThrottleTimers));
 
     /* r/w error */
     bs_dest->on_read_error      = bs_src->on_read_error;
@@ -1881,12 +1888,21 @@ void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old)
         QTAILQ_REMOVE(&graph_bdrv_states, bs_old, node_list);
     }
 
+    /* If the BlockDriverState is part of a throttling group acquire
+     * its lock since we're going to mess with the protected fields.
+     * Otherwise there's no need to worry since no one else can touch
+     * them. */
+    if (bs_old->throttle_state) {
+        throttle_group_lock(bs_old);
+    }
+
     /* bs_new must be unattached and shouldn't have anything fancy enabled */
     assert(!bs_new->blk);
     assert(QLIST_EMPTY(&bs_new->dirty_bitmaps));
     assert(bs_new->job == NULL);
     assert(bs_new->io_limits_enabled == false);
-    assert(!throttle_have_timer(&bs_new->throttle_state));
+    assert(bs_new->throttle_state == NULL);
+    assert(!throttle_timers_are_initialized(&bs_new->throttle_timers));
 
     tmp = *bs_new;
     *bs_new = *bs_old;
@@ -1903,7 +1919,13 @@ void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old)
     /* Check a few fields that should remain attached to the device */
     assert(bs_new->job == NULL);
     assert(bs_new->io_limits_enabled == false);
-    assert(!throttle_have_timer(&bs_new->throttle_state));
+    assert(bs_new->throttle_state == NULL);
+    assert(!throttle_timers_are_initialized(&bs_new->throttle_timers));
+
+    /* Release the ThrottleGroup lock */
+    if (bs_old->throttle_state) {
+        throttle_group_unlock(bs_old);
+    }
 
     /* insert the nodes back into the graph node list if needed */
     if (bs_new->node_name[0] != '\0') {
@@ -3691,7 +3713,7 @@ void bdrv_detach_aio_context(BlockDriverState *bs)
     }
 
     if (bs->io_limits_enabled) {
-        throttle_detach_aio_context(&bs->throttle_state);
+        throttle_timers_detach_aio_context(&bs->throttle_timers);
     }
     if (bs->drv->bdrv_detach_aio_context) {
         bs->drv->bdrv_detach_aio_context(bs);
@@ -3727,7 +3749,7 @@ void bdrv_attach_aio_context(BlockDriverState *bs,
         bs->drv->bdrv_attach_aio_context(bs, new_context);
     }
     if (bs->io_limits_enabled) {
-        throttle_attach_aio_context(&bs->throttle_state, new_context);
+        throttle_timers_attach_aio_context(&bs->throttle_timers, new_context);
     }
 
     QLIST_FOREACH(ban, &bs->aio_notifiers, list) {

diff --git a/block/Makefile.objs b/block/Makefile.objs
@@ -10,6 +10,7 @@ block-obj-$(CONFIG_WIN32) += raw-win32.o win32-aio.o
 block-obj-$(CONFIG_POSIX) += raw-posix.o
 block-obj-$(CONFIG_LINUX_AIO) += linux-aio.o
 block-obj-y += null.o mirror.o io.o
+block-obj-y += throttle-groups.o
 
 block-obj-y += nbd.o nbd-client.o sheepdog.o
 block-obj-$(CONFIG_LIBISCSI) += iscsi.o

diff --git a/block/io.c b/block/io.c
@@ -23,9 +23,9 @@
  */
 
 #include "trace.h"
-#include "sysemu/qtest.h"
 #include "block/blockjob.h"
 #include "block/block_int.h"
+#include "block/throttle-groups.h"
 
 #define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */
 
@@ -65,7 +65,7 @@ void bdrv_set_io_limits(BlockDriverState *bs,
 {
     int i;
 
-    throttle_config(&bs->throttle_state, cfg);
+    throttle_group_config(bs, cfg);
 
     for (i = 0; i < 2; i++) {
         qemu_co_enter_next(&bs->throttled_reqs[i]);
@@ -95,72 +95,33 @@ static bool bdrv_start_throttled_reqs(BlockDriverState *bs)
 void bdrv_io_limits_disable(BlockDriverState *bs)
 {
     bs->io_limits_enabled = false;
-
     bdrv_start_throttled_reqs(bs);
-
-    throttle_destroy(&bs->throttle_state);
-}
-
-static void bdrv_throttle_read_timer_cb(void *opaque)
-{
-    BlockDriverState *bs = opaque;
-    qemu_co_enter_next(&bs->throttled_reqs[0]);
-}
-
-static void bdrv_throttle_write_timer_cb(void *opaque)
-{
-    BlockDriverState *bs = opaque;
-    qemu_co_enter_next(&bs->throttled_reqs[1]);
+    throttle_group_unregister_bs(bs);
 }
 
 /* should be called before bdrv_set_io_limits if a limit is set */
-void bdrv_io_limits_enable(BlockDriverState *bs)
+void bdrv_io_limits_enable(BlockDriverState *bs, const char *group)
 {
-    int clock_type = QEMU_CLOCK_REALTIME;
-
-    if (qtest_enabled()) {
-        /* For testing block IO throttling only */
-        clock_type = QEMU_CLOCK_VIRTUAL;
-    }
     assert(!bs->io_limits_enabled);
-    throttle_init(&bs->throttle_state,
-                  bdrv_get_aio_context(bs),
-                  clock_type,
-                  bdrv_throttle_read_timer_cb,
-                  bdrv_throttle_write_timer_cb,
-                  bs);
+    throttle_group_register_bs(bs, group);
     bs->io_limits_enabled = true;
 }
 
-/* This function makes an IO wait if needed
- *
- * @nb_sectors: the number of sectors of the IO
- * @is_write:   is the IO a write
- */
-static void bdrv_io_limits_intercept(BlockDriverState *bs,
-                                     unsigned int bytes,
-                                     bool is_write)
+void bdrv_io_limits_update_group(BlockDriverState *bs, const char *group)
 {
-    /* does this io must wait */
-    bool must_wait = throttle_schedule_timer(&bs->throttle_state, is_write);
-
-    /* if must wait or any request of this type throttled queue the IO */
-    if (must_wait ||
-        !qemu_co_queue_empty(&bs->throttled_reqs[is_write])) {
-        qemu_co_queue_wait(&bs->throttled_reqs[is_write]);
+    /* this bs is not part of any group */
+    if (!bs->throttle_state) {
+        return;
     }
 
-    /* the IO will be executed, do the accounting */
-    throttle_account(&bs->throttle_state, is_write, bytes);
-
-
-    /* if the next request must wait -> do nothing */
-    if (throttle_schedule_timer(&bs->throttle_state, is_write)) {
+    /* this bs is a part of the same group than the one we want */
+    if (!g_strcmp0(throttle_group_get_name(bs), group)) {
         return;
     }
 
-    /* else queue next request for execution */
-    qemu_co_queue_next(&bs->throttled_reqs[is_write]);
+    /* need to change the group this bs belong to */
+    bdrv_io_limits_disable(bs);
+    bdrv_io_limits_enable(bs, group);
 }
 
 void bdrv_setup_io_funcs(BlockDriver *bdrv)
@@ -967,7 +928,7 @@ static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
 
     /* throttling disk I/O */
     if (bs->io_limits_enabled) {
-        bdrv_io_limits_intercept(bs, bytes, false);
+        throttle_group_co_io_limits_intercept(bs, bytes, false);
     }
 
     /* Align read if necessary by padding qiov */
@@ -1297,7 +1258,7 @@ static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
 
     /* throttling disk I/O */
     if (bs->io_limits_enabled) {
-        bdrv_io_limits_intercept(bs, bytes, true);
+        throttle_group_co_io_limits_intercept(bs, bytes, true);
     }
 
     /*

diff --git a/block/qapi.c b/block/qapi.c
@@ -24,6 +24,7 @@
 
 #include "block/qapi.h"
 #include "block/block_int.h"
+#include "block/throttle-groups.h"
 #include "block/write-threshold.h"
 #include "qmp-commands.h"
 #include "qapi-visit.h"
@@ -65,7 +66,9 @@ BlockDeviceInfo *bdrv_block_device_info(BlockDriverState *bs, Error **errp)
 
     if (bs->io_limits_enabled) {
         ThrottleConfig cfg;
-        throttle_get_config(&bs->throttle_state, &cfg);
+
+        throttle_group_get_config(bs, &cfg);
+
         info->bps     = cfg.buckets[THROTTLE_BPS_TOTAL].avg;
         info->bps_rd  = cfg.buckets[THROTTLE_BPS_READ].avg;
         info->bps_wr  = cfg.buckets[THROTTLE_BPS_WRITE].avg;
@@ -90,6 +93,9 @@ BlockDeviceInfo *bdrv_block_device_info(BlockDriverState *bs, Error **errp)
 
         info->has_iops_size = cfg.op_size;
         info->iops_size = cfg.op_size;
+
+        info->has_group = true;
+        info->group = g_strdup(throttle_group_get_name(bs));
     }
 
     info->write_threshold = bdrv_write_threshold_get(bs);

diff --git a/block/raw-posix.c b/block/raw-posix.c
@@ -1848,8 +1848,9 @@ static int64_t coroutine_fn raw_co_get_block_status(BlockDriverState *bs,
         *pnum = nb_sectors;
         ret = BDRV_BLOCK_DATA;
     } else if (data == start) {
-        /* On a data extent, compute sectors to the end of the extent.  */
-        *pnum = MIN(nb_sectors, (hole - start) / BDRV_SECTOR_SIZE);
+        /* On a data extent, compute sectors to the end of the extent,
+         * possibly including a partial sector at EOF. */
+        *pnum = MIN(nb_sectors, DIV_ROUND_UP(hole - start, BDRV_SECTOR_SIZE));
         ret = BDRV_BLOCK_DATA;
     } else {
         /* On a hole, compute sectors to the beginning of the next extent.  */