From patchwork Fri Sep 6 15:38:34 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Stefan Hajnoczi X-Patchwork-Id: 273236 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from lists.gnu.org (lists.gnu.org [IPv6:2001:4830:134:3::11]) (using TLSv1 with cipher AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTPS id B273C2C010A for ; Sat, 7 Sep 2013 01:40:19 +1000 (EST) Received: from localhost ([::1]:38405 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VHy8v-0004P6-NE for incoming@patchwork.ozlabs.org; Fri, 06 Sep 2013 11:40:17 -0400 Received: from eggs.gnu.org ([2001:4830:134:3::10]:40146) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VHy8E-0004EN-Pd for qemu-devel@nongnu.org; Fri, 06 Sep 2013 11:39:39 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VHy89-00055h-K8 for qemu-devel@nongnu.org; Fri, 06 Sep 2013 11:39:34 -0400 Received: from mx1.redhat.com ([209.132.183.28]:60817) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VHy89-00055S-9U for qemu-devel@nongnu.org; Fri, 06 Sep 2013 11:39:29 -0400 Received: from int-mx10.intmail.prod.int.phx2.redhat.com (int-mx10.intmail.prod.int.phx2.redhat.com [10.5.11.23]) by mx1.redhat.com (8.14.4/8.14.4) with ESMTP id r86FdQh7004265 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Fri, 6 Sep 2013 11:39:26 -0400 Received: from localhost (ovpn-112-53.ams2.redhat.com [10.36.112.53]) by int-mx10.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id r86FdOeZ022872; Fri, 6 Sep 2013 11:39:24 -0400 From: Stefan Hajnoczi To: Date: Fri, 6 Sep 2013 17:38:34 +0200 Message-Id: <1378481953-23099-4-git-send-email-stefanha@redhat.com> In-Reply-To: <1378481953-23099-1-git-send-email-stefanha@redhat.com> References: <1378481953-23099-1-git-send-email-stefanha@redhat.com> MIME-Version: 1.0 X-Scanned-By: MIMEDefang 2.68 on 10.5.11.23 X-MIME-Autoconverted: from 8bit to quoted-printable by mx1.redhat.com id r86FdQh7004265 X-detected-operating-system: by eggs.gnu.org: GNU/Linux 3.x X-Received-From: 209.132.183.28 Cc: Stefan Hajnoczi , =?UTF-8?q?Beno=C3=AEt=20Canet?= , Anthony Liguori Subject: [Qemu-devel] [PULL 03/42] block: Enable the new throttling code in the block layer. X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Sender: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org From: Benoît Canet Signed-off-by: Benoit Canet Signed-off-by: Stefan Hajnoczi --- block.c | 338 +++++++++++++--------------------------------- block/qapi.c | 21 ++- blockdev.c | 102 +++++++------- include/block/block.h | 1 - include/block/block_int.h | 32 +---- 5 files changed, 165 insertions(+), 329 deletions(-) diff --git a/block.c b/block.c index 26639e8..0292d1d 100644 --- a/block.c +++ b/block.c @@ -86,13 +86,6 @@ static void coroutine_fn bdrv_co_do_rw(void *opaque); static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs, int64_t sector_num, int nb_sectors); -static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors, - bool is_write, double elapsed_time, uint64_t *wait); -static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write, - double elapsed_time, uint64_t *wait); -static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors, - bool is_write, int64_t *wait); - static QTAILQ_HEAD(, BlockDriverState) bdrv_states = QTAILQ_HEAD_INITIALIZER(bdrv_states); @@ -123,69 +116,101 @@ int is_windows_drive(const char *filename) #endif /* throttling disk I/O limits */ -void bdrv_io_limits_disable(BlockDriverState *bs) +void bdrv_set_io_limits(BlockDriverState *bs, + ThrottleConfig *cfg) { - bs->io_limits_enabled = false; + int i; - do {} while (qemu_co_enter_next(&bs->throttled_reqs)); + throttle_config(&bs->throttle_state, cfg); - if (bs->block_timer) { - timer_del(bs->block_timer); - timer_free(bs->block_timer); - bs->block_timer = NULL; + for (i = 0; i < 2; i++) { + qemu_co_enter_next(&bs->throttled_reqs[i]); } +} + +/* this function drain all the throttled IOs */ +static bool bdrv_start_throttled_reqs(BlockDriverState *bs) +{ + bool drained = false; + bool enabled = bs->io_limits_enabled; + int i; + + bs->io_limits_enabled = false; + + for (i = 0; i < 2; i++) { + while (qemu_co_enter_next(&bs->throttled_reqs[i])) { + drained = true; + } + } + + bs->io_limits_enabled = enabled; - bs->slice_start = 0; - bs->slice_end = 0; + return drained; } -static void bdrv_block_timer(void *opaque) +void bdrv_io_limits_disable(BlockDriverState *bs) { - BlockDriverState *bs = opaque; + bs->io_limits_enabled = false; - qemu_co_enter_next(&bs->throttled_reqs); + bdrv_start_throttled_reqs(bs); + + throttle_destroy(&bs->throttle_state); } -void bdrv_io_limits_enable(BlockDriverState *bs) +static void bdrv_throttle_read_timer_cb(void *opaque) { - bs->block_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, bdrv_block_timer, bs); - bs->io_limits_enabled = true; + BlockDriverState *bs = opaque; + qemu_co_enter_next(&bs->throttled_reqs[0]); } -bool bdrv_io_limits_enabled(BlockDriverState *bs) +static void bdrv_throttle_write_timer_cb(void *opaque) { - BlockIOLimit *io_limits = &bs->io_limits; - return io_limits->bps[BLOCK_IO_LIMIT_READ] - || io_limits->bps[BLOCK_IO_LIMIT_WRITE] - || io_limits->bps[BLOCK_IO_LIMIT_TOTAL] - || io_limits->iops[BLOCK_IO_LIMIT_READ] - || io_limits->iops[BLOCK_IO_LIMIT_WRITE] - || io_limits->iops[BLOCK_IO_LIMIT_TOTAL]; + BlockDriverState *bs = opaque; + qemu_co_enter_next(&bs->throttled_reqs[1]); } +/* should be called before bdrv_set_io_limits if a limit is set */ +void bdrv_io_limits_enable(BlockDriverState *bs) +{ + assert(!bs->io_limits_enabled); + throttle_init(&bs->throttle_state, + QEMU_CLOCK_VIRTUAL, + bdrv_throttle_read_timer_cb, + bdrv_throttle_write_timer_cb, + bs); + 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, - bool is_write, int nb_sectors) + int nb_sectors, + bool is_write) { - int64_t wait_time = -1; + /* does this io must wait */ + bool must_wait = throttle_schedule_timer(&bs->throttle_state, is_write); - if (!qemu_co_queue_empty(&bs->throttled_reqs)) { - qemu_co_queue_wait(&bs->throttled_reqs); + /* 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]); } - /* In fact, we hope to keep each request's timing, in FIFO mode. The next - * throttled requests will not be dequeued until the current request is - * allowed to be serviced. So if the current request still exceeds the - * limits, it will be inserted to the head. All requests followed it will - * be still in throttled_reqs queue. - */ + /* the IO will be executed, do the accounting */ + throttle_account(&bs->throttle_state, + is_write, + nb_sectors * BDRV_SECTOR_SIZE); - while (bdrv_exceed_io_limits(bs, nb_sectors, is_write, &wait_time)) { - timer_mod(bs->block_timer, - wait_time + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL)); - qemu_co_queue_wait_insert_head(&bs->throttled_reqs); + /* if the next request must wait -> do nothing */ + if (throttle_schedule_timer(&bs->throttle_state, is_write)) { + return; } - qemu_co_queue_next(&bs->throttled_reqs); + /* else queue next request for execution */ + qemu_co_queue_next(&bs->throttled_reqs[is_write]); } /* check if the path starts with ":" */ @@ -305,7 +330,8 @@ BlockDriverState *bdrv_new(const char *device_name) bdrv_iostatus_disable(bs); notifier_list_init(&bs->close_notifiers); notifier_with_return_list_init(&bs->before_write_notifiers); - qemu_co_queue_init(&bs->throttled_reqs); + qemu_co_queue_init(&bs->throttled_reqs[0]); + qemu_co_queue_init(&bs->throttled_reqs[1]); return bs; } @@ -1112,11 +1138,6 @@ int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options, bdrv_dev_change_media_cb(bs, true); } - /* throttling disk I/O limits */ - if (bs->io_limits_enabled) { - bdrv_io_limits_enable(bs); - } - return 0; unlink_and_fail: @@ -1435,7 +1456,10 @@ static bool bdrv_requests_pending(BlockDriverState *bs) if (!QLIST_EMPTY(&bs->tracked_requests)) { return true; } - if (!qemu_co_queue_empty(&bs->throttled_reqs)) { + if (!qemu_co_queue_empty(&bs->throttled_reqs[0])) { + return true; + } + if (!qemu_co_queue_empty(&bs->throttled_reqs[1])) { return true; } if (bs->file && bdrv_requests_pending(bs->file)) { @@ -1481,7 +1505,7 @@ void bdrv_drain_all(void) * a busy wait. */ QTAILQ_FOREACH(bs, &bdrv_states, list) { - while (qemu_co_enter_next(&bs->throttled_reqs)) { + if (bdrv_start_throttled_reqs(bs)) { busy = true; } } @@ -1523,13 +1547,12 @@ static void bdrv_move_feature_fields(BlockDriverState *bs_dest, bs_dest->enable_write_cache = bs_src->enable_write_cache; - /* i/o timing parameters */ - bs_dest->slice_start = bs_src->slice_start; - bs_dest->slice_end = bs_src->slice_end; - bs_dest->slice_submitted = bs_src->slice_submitted; - bs_dest->io_limits = bs_src->io_limits; - bs_dest->throttled_reqs = bs_src->throttled_reqs; - bs_dest->block_timer = bs_src->block_timer; + /* i/o throttled req */ + memcpy(&bs_dest->throttle_state, + &bs_src->throttle_state, + sizeof(ThrottleState)); + 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; /* r/w error */ @@ -1576,7 +1599,7 @@ void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old) assert(bs_new->dev == NULL); assert(bs_new->in_use == 0); assert(bs_new->io_limits_enabled == false); - assert(bs_new->block_timer == NULL); + assert(!throttle_have_timer(&bs_new->throttle_state)); tmp = *bs_new; *bs_new = *bs_old; @@ -1595,7 +1618,7 @@ void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old) assert(bs_new->job == NULL); assert(bs_new->in_use == 0); assert(bs_new->io_limits_enabled == false); - assert(bs_new->block_timer == NULL); + assert(!throttle_have_timer(&bs_new->throttle_state)); bdrv_rebind(bs_new); bdrv_rebind(bs_old); @@ -2538,11 +2561,6 @@ static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs, return -EIO; } - /* throttling disk read I/O */ - if (bs->io_limits_enabled) { - bdrv_io_limits_intercept(bs, false, nb_sectors); - } - if (bs->copy_on_read) { flags |= BDRV_REQ_COPY_ON_READ; } @@ -2554,6 +2572,11 @@ static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs, wait_for_overlapping_requests(bs, sector_num, nb_sectors); } + /* throttling disk I/O */ + if (bs->io_limits_enabled) { + bdrv_io_limits_intercept(bs, nb_sectors, false); + } + tracked_request_begin(&req, bs, sector_num, nb_sectors, false); if (flags & BDRV_REQ_COPY_ON_READ) { @@ -2679,15 +2702,15 @@ static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs, return -EIO; } - /* throttling disk write I/O */ - if (bs->io_limits_enabled) { - bdrv_io_limits_intercept(bs, true, nb_sectors); - } - if (bs->copy_on_read_in_flight) { wait_for_overlapping_requests(bs, sector_num, nb_sectors); } + /* throttling disk I/O */ + if (bs->io_limits_enabled) { + bdrv_io_limits_intercept(bs, nb_sectors, true); + } + tracked_request_begin(&req, bs, sector_num, nb_sectors, true); ret = notifier_with_return_list_notify(&bs->before_write_notifiers, &req); @@ -2805,14 +2828,6 @@ void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr) *nb_sectors_ptr = length; } -/* throttling disk io limits */ -void bdrv_set_io_limits(BlockDriverState *bs, - BlockIOLimit *io_limits) -{ - bs->io_limits = *io_limits; - bs->io_limits_enabled = bdrv_io_limits_enabled(bs); -} - void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error, BlockdevOnError on_write_error) { @@ -3622,169 +3637,6 @@ void bdrv_aio_cancel(BlockDriverAIOCB *acb) acb->aiocb_info->cancel(acb); } -/* block I/O throttling */ -static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors, - bool is_write, double elapsed_time, uint64_t *wait) -{ - uint64_t bps_limit = 0; - uint64_t extension; - double bytes_limit, bytes_base, bytes_res; - double slice_time, wait_time; - - if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) { - bps_limit = bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]; - } else if (bs->io_limits.bps[is_write]) { - bps_limit = bs->io_limits.bps[is_write]; - } else { - if (wait) { - *wait = 0; - } - - return false; - } - - slice_time = bs->slice_end - bs->slice_start; - slice_time /= (NANOSECONDS_PER_SECOND); - bytes_limit = bps_limit * slice_time; - bytes_base = bs->slice_submitted.bytes[is_write]; - if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) { - bytes_base += bs->slice_submitted.bytes[!is_write]; - } - - /* bytes_base: the bytes of data which have been read/written; and - * it is obtained from the history statistic info. - * bytes_res: the remaining bytes of data which need to be read/written. - * (bytes_base + bytes_res) / bps_limit: used to calcuate - * the total time for completing reading/writting all data. - */ - bytes_res = (unsigned) nb_sectors * BDRV_SECTOR_SIZE; - - if (bytes_base + bytes_res <= bytes_limit) { - if (wait) { - *wait = 0; - } - - return false; - } - - /* Calc approx time to dispatch */ - wait_time = (bytes_base + bytes_res) / bps_limit - elapsed_time; - - /* When the I/O rate at runtime exceeds the limits, - * bs->slice_end need to be extended in order that the current statistic - * info can be kept until the timer fire, so it is increased and tuned - * based on the result of experiment. - */ - extension = wait_time * NANOSECONDS_PER_SECOND; - extension = DIV_ROUND_UP(extension, BLOCK_IO_SLICE_TIME) * - BLOCK_IO_SLICE_TIME; - bs->slice_end += extension; - if (wait) { - *wait = wait_time * NANOSECONDS_PER_SECOND; - } - - return true; -} - -static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write, - double elapsed_time, uint64_t *wait) -{ - uint64_t iops_limit = 0; - double ios_limit, ios_base; - double slice_time, wait_time; - - if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) { - iops_limit = bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]; - } else if (bs->io_limits.iops[is_write]) { - iops_limit = bs->io_limits.iops[is_write]; - } else { - if (wait) { - *wait = 0; - } - - return false; - } - - slice_time = bs->slice_end - bs->slice_start; - slice_time /= (NANOSECONDS_PER_SECOND); - ios_limit = iops_limit * slice_time; - ios_base = bs->slice_submitted.ios[is_write]; - if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) { - ios_base += bs->slice_submitted.ios[!is_write]; - } - - if (ios_base + 1 <= ios_limit) { - if (wait) { - *wait = 0; - } - - return false; - } - - /* Calc approx time to dispatch, in seconds */ - wait_time = (ios_base + 1) / iops_limit; - if (wait_time > elapsed_time) { - wait_time = wait_time - elapsed_time; - } else { - wait_time = 0; - } - - /* Exceeded current slice, extend it by another slice time */ - bs->slice_end += BLOCK_IO_SLICE_TIME; - if (wait) { - *wait = wait_time * NANOSECONDS_PER_SECOND; - } - - return true; -} - -static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors, - bool is_write, int64_t *wait) -{ - int64_t now, max_wait; - uint64_t bps_wait = 0, iops_wait = 0; - double elapsed_time; - int bps_ret, iops_ret; - - now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL); - if (now > bs->slice_end) { - bs->slice_start = now; - bs->slice_end = now + BLOCK_IO_SLICE_TIME; - memset(&bs->slice_submitted, 0, sizeof(bs->slice_submitted)); - } - - elapsed_time = now - bs->slice_start; - elapsed_time /= (NANOSECONDS_PER_SECOND); - - bps_ret = bdrv_exceed_bps_limits(bs, nb_sectors, - is_write, elapsed_time, &bps_wait); - iops_ret = bdrv_exceed_iops_limits(bs, is_write, - elapsed_time, &iops_wait); - if (bps_ret || iops_ret) { - max_wait = bps_wait > iops_wait ? bps_wait : iops_wait; - if (wait) { - *wait = max_wait; - } - - now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL); - if (bs->slice_end < now + max_wait) { - bs->slice_end = now + max_wait; - } - - return true; - } - - if (wait) { - *wait = 0; - } - - bs->slice_submitted.bytes[is_write] += (int64_t)nb_sectors * - BDRV_SECTOR_SIZE; - bs->slice_submitted.ios[is_write]++; - - return false; -} - /**************************************************************/ /* async block device emulation */ diff --git a/block/qapi.c b/block/qapi.c index a4bc411..cac3919 100644 --- a/block/qapi.c +++ b/block/qapi.c @@ -223,18 +223,15 @@ void bdrv_query_info(BlockDriverState *bs, info->inserted->backing_file_depth = bdrv_get_backing_file_depth(bs); if (bs->io_limits_enabled) { - info->inserted->bps = - bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]; - info->inserted->bps_rd = - bs->io_limits.bps[BLOCK_IO_LIMIT_READ]; - info->inserted->bps_wr = - bs->io_limits.bps[BLOCK_IO_LIMIT_WRITE]; - info->inserted->iops = - bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]; - info->inserted->iops_rd = - bs->io_limits.iops[BLOCK_IO_LIMIT_READ]; - info->inserted->iops_wr = - bs->io_limits.iops[BLOCK_IO_LIMIT_WRITE]; + ThrottleConfig cfg; + throttle_get_config(&bs->throttle_state, &cfg); + info->inserted->bps = cfg.buckets[THROTTLE_BPS_TOTAL].avg; + info->inserted->bps_rd = cfg.buckets[THROTTLE_BPS_READ].avg; + info->inserted->bps_wr = cfg.buckets[THROTTLE_BPS_WRITE].avg; + + info->inserted->iops = cfg.buckets[THROTTLE_OPS_TOTAL].avg; + info->inserted->iops_rd = cfg.buckets[THROTTLE_OPS_READ].avg; + info->inserted->iops_wr = cfg.buckets[THROTTLE_OPS_WRITE].avg; } bs0 = bs; diff --git a/blockdev.c b/blockdev.c index e70e16e..5f5ba96 100644 --- a/blockdev.c +++ b/blockdev.c @@ -280,32 +280,16 @@ static int parse_block_error_action(const char *buf, bool is_read) } } -static bool do_check_io_limits(BlockIOLimit *io_limits, Error **errp) +static bool check_throttle_config(ThrottleConfig *cfg, Error **errp) { - bool bps_flag; - bool iops_flag; - - assert(io_limits); - - bps_flag = (io_limits->bps[BLOCK_IO_LIMIT_TOTAL] != 0) - && ((io_limits->bps[BLOCK_IO_LIMIT_READ] != 0) - || (io_limits->bps[BLOCK_IO_LIMIT_WRITE] != 0)); - iops_flag = (io_limits->iops[BLOCK_IO_LIMIT_TOTAL] != 0) - && ((io_limits->iops[BLOCK_IO_LIMIT_READ] != 0) - || (io_limits->iops[BLOCK_IO_LIMIT_WRITE] != 0)); - if (bps_flag || iops_flag) { - error_setg(errp, "bps(iops) and bps_rd/bps_wr(iops_rd/iops_wr) " - "cannot be used at the same time"); + if (throttle_conflicting(cfg)) { + error_setg(errp, "bps/iops/max total values and read/write values" + " cannot be used at the same time"); return false; } - if (io_limits->bps[BLOCK_IO_LIMIT_TOTAL] < 0 || - io_limits->bps[BLOCK_IO_LIMIT_WRITE] < 0 || - io_limits->bps[BLOCK_IO_LIMIT_READ] < 0 || - io_limits->iops[BLOCK_IO_LIMIT_TOTAL] < 0 || - io_limits->iops[BLOCK_IO_LIMIT_WRITE] < 0 || - io_limits->iops[BLOCK_IO_LIMIT_READ] < 0) { - error_setg(errp, "bps and iops values must be 0 or greater"); + if (!throttle_is_valid(cfg)) { + error_setg(errp, "bps/iops/maxs values must be 0 or greater"); return false; } @@ -330,7 +314,7 @@ static DriveInfo *blockdev_init(QemuOpts *all_opts, int on_read_error, on_write_error; const char *devaddr; DriveInfo *dinfo; - BlockIOLimit io_limits; + ThrottleConfig cfg; int snapshot = 0; bool copy_on_read; int ret; @@ -496,20 +480,31 @@ static DriveInfo *blockdev_init(QemuOpts *all_opts, } /* disk I/O throttling */ - io_limits.bps[BLOCK_IO_LIMIT_TOTAL] = + memset(&cfg, 0, sizeof(cfg)); + cfg.buckets[THROTTLE_BPS_TOTAL].avg = qemu_opt_get_number(opts, "throttling.bps-total", 0); - io_limits.bps[BLOCK_IO_LIMIT_READ] = + cfg.buckets[THROTTLE_BPS_READ].avg = qemu_opt_get_number(opts, "throttling.bps-read", 0); - io_limits.bps[BLOCK_IO_LIMIT_WRITE] = + cfg.buckets[THROTTLE_BPS_WRITE].avg = qemu_opt_get_number(opts, "throttling.bps-write", 0); - io_limits.iops[BLOCK_IO_LIMIT_TOTAL] = + cfg.buckets[THROTTLE_OPS_TOTAL].avg = qemu_opt_get_number(opts, "throttling.iops-total", 0); - io_limits.iops[BLOCK_IO_LIMIT_READ] = + cfg.buckets[THROTTLE_OPS_READ].avg = qemu_opt_get_number(opts, "throttling.iops-read", 0); - io_limits.iops[BLOCK_IO_LIMIT_WRITE] = + cfg.buckets[THROTTLE_OPS_WRITE].avg = qemu_opt_get_number(opts, "throttling.iops-write", 0); - if (!do_check_io_limits(&io_limits, &error)) { + cfg.buckets[THROTTLE_BPS_TOTAL].max = 0; + cfg.buckets[THROTTLE_BPS_READ].max = 0; + cfg.buckets[THROTTLE_BPS_WRITE].max = 0; + + cfg.buckets[THROTTLE_OPS_TOTAL].max = 0; + cfg.buckets[THROTTLE_OPS_READ].max = 0; + cfg.buckets[THROTTLE_OPS_WRITE].max = 0; + + cfg.op_size = 0; + + if (!check_throttle_config(&cfg, &error)) { error_report("%s", error_get_pretty(error)); error_free(error); return NULL; @@ -636,7 +631,10 @@ static DriveInfo *blockdev_init(QemuOpts *all_opts, bdrv_set_on_error(dinfo->bdrv, on_read_error, on_write_error); /* disk I/O throttling */ - bdrv_set_io_limits(dinfo->bdrv, &io_limits); + if (throttle_enabled(&cfg)) { + bdrv_io_limits_enable(dinfo->bdrv); + bdrv_set_io_limits(dinfo->bdrv, &cfg); + } switch(type) { case IF_IDE: @@ -1250,7 +1248,7 @@ void qmp_block_set_io_throttle(const char *device, int64_t bps, int64_t bps_rd, int64_t bps_wr, int64_t iops, int64_t iops_rd, int64_t iops_wr, Error **errp) { - BlockIOLimit io_limits; + ThrottleConfig cfg; BlockDriverState *bs; bs = bdrv_find(device); @@ -1259,27 +1257,37 @@ void qmp_block_set_io_throttle(const char *device, int64_t bps, int64_t bps_rd, return; } - io_limits.bps[BLOCK_IO_LIMIT_TOTAL] = bps; - io_limits.bps[BLOCK_IO_LIMIT_READ] = bps_rd; - io_limits.bps[BLOCK_IO_LIMIT_WRITE] = bps_wr; - io_limits.iops[BLOCK_IO_LIMIT_TOTAL]= iops; - io_limits.iops[BLOCK_IO_LIMIT_READ] = iops_rd; - io_limits.iops[BLOCK_IO_LIMIT_WRITE]= iops_wr; + memset(&cfg, 0, sizeof(cfg)); + cfg.buckets[THROTTLE_BPS_TOTAL].avg = bps; + cfg.buckets[THROTTLE_BPS_READ].avg = bps_rd; + cfg.buckets[THROTTLE_BPS_WRITE].avg = bps_wr; + + cfg.buckets[THROTTLE_OPS_TOTAL].avg = iops; + cfg.buckets[THROTTLE_OPS_READ].avg = iops_rd; + cfg.buckets[THROTTLE_OPS_WRITE].avg = iops_wr; + + cfg.buckets[THROTTLE_BPS_TOTAL].max = 0; + cfg.buckets[THROTTLE_BPS_READ].max = 0; + cfg.buckets[THROTTLE_BPS_WRITE].max = 0; + + cfg.buckets[THROTTLE_OPS_TOTAL].max = 0; + cfg.buckets[THROTTLE_OPS_READ].max = 0; + cfg.buckets[THROTTLE_OPS_WRITE].max = 0; - if (!do_check_io_limits(&io_limits, errp)) { + cfg.op_size = 0; + + if (!check_throttle_config(&cfg, errp)) { return; } - bs->io_limits = io_limits; - - if (!bs->io_limits_enabled && bdrv_io_limits_enabled(bs)) { + if (!bs->io_limits_enabled && throttle_enabled(&cfg)) { bdrv_io_limits_enable(bs); - } else if (bs->io_limits_enabled && !bdrv_io_limits_enabled(bs)) { + } else if (bs->io_limits_enabled && !throttle_enabled(&cfg)) { bdrv_io_limits_disable(bs); - } else { - if (bs->block_timer) { - timer_mod(bs->block_timer, qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL)); - } + } + + if (bs->io_limits_enabled) { + bdrv_set_io_limits(bs, &cfg); } } diff --git a/include/block/block.h b/include/block/block.h index e6b391c..6207ff2 100644 --- a/include/block/block.h +++ b/include/block/block.h @@ -107,7 +107,6 @@ void bdrv_info_stats(Monitor *mon, QObject **ret_data); /* disk I/O throttling */ void bdrv_io_limits_enable(BlockDriverState *bs); void bdrv_io_limits_disable(BlockDriverState *bs); -bool bdrv_io_limits_enabled(BlockDriverState *bs); void bdrv_init(void); void bdrv_init_with_whitelist(void); diff --git a/include/block/block_int.h b/include/block/block_int.h index 8012e25..c3c9c61 100644 --- a/include/block/block_int.h +++ b/include/block/block_int.h @@ -35,18 +35,12 @@ #include "qemu/hbitmap.h" #include "block/snapshot.h" #include "qemu/main-loop.h" +#include "qemu/throttle.h" #define BLOCK_FLAG_ENCRYPT 1 #define BLOCK_FLAG_COMPAT6 4 #define BLOCK_FLAG_LAZY_REFCOUNTS 8 -#define BLOCK_IO_LIMIT_READ 0 -#define BLOCK_IO_LIMIT_WRITE 1 -#define BLOCK_IO_LIMIT_TOTAL 2 - -#define BLOCK_IO_SLICE_TIME 100000000 -#define NANOSECONDS_PER_SECOND 1000000000.0 - #define BLOCK_OPT_SIZE "size" #define BLOCK_OPT_ENCRYPT "encryption" #define BLOCK_OPT_COMPAT6 "compat6" @@ -70,17 +64,6 @@ typedef struct BdrvTrackedRequest { CoQueue wait_queue; /* coroutines blocked on this request */ } BdrvTrackedRequest; - -typedef struct BlockIOLimit { - int64_t bps[3]; - int64_t iops[3]; -} BlockIOLimit; - -typedef struct BlockIOBaseValue { - uint64_t bytes[2]; - uint64_t ios[2]; -} BlockIOBaseValue; - struct BlockDriver { const char *format_name; int instance_size; @@ -264,13 +247,9 @@ struct BlockDriverState { /* number of in-flight copy-on-read requests */ unsigned int copy_on_read_in_flight; - /* the time for latest disk I/O */ - int64_t slice_start; - int64_t slice_end; - BlockIOLimit io_limits; - BlockIOBaseValue slice_submitted; - CoQueue throttled_reqs; - QEMUTimer *block_timer; + /* I/O throttling */ + ThrottleState throttle_state; + CoQueue throttled_reqs[2]; bool io_limits_enabled; /* I/O stats (display with "info blockstats"). */ @@ -312,7 +291,8 @@ struct BlockDriverState { int get_tmp_filename(char *filename, int size); void bdrv_set_io_limits(BlockDriverState *bs, - BlockIOLimit *io_limits); + ThrottleConfig *cfg); + /** * bdrv_add_before_write_notifier: