From: Kevin Wolf <kwolf@redhat.com>
To: anthony@codemonkey.ws
Cc: kwolf@redhat.com, qemu-devel@nongnu.org
Subject: [Qemu-devel] [PATCH 3/8] block: fix I/O throttling accounting blind spot
Date: Fri, 5 Apr 2013 15:28:05 +0200 [thread overview]
Message-ID: <1365168490-29616-4-git-send-email-kwolf@redhat.com> (raw)
In-Reply-To: <1365168490-29616-1-git-send-email-kwolf@redhat.com>
From: Stefan Hajnoczi <stefanha@redhat.com>
I/O throttling relies on bdrv_acct_done() which is called when a request
completes. This leaves a blind spot since we only charge for completed
requests, not submitted requests.
For example, if there is 1 operation remaining in this time slice the
guest could submit 3 operations and they will all be submitted
successfully since they don't actually get accounted for until they
complete.
Originally we probably thought this is okay since the requests will be
accounted when the time slice is extended. In practice it causes
fluctuations since the guest can exceed its I/O limit and it will be
punished for this later on.
Account for I/O upon submission so that I/O limits are enforced
properly.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-By: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
---
block.c | 21 ++++++++++-----------
include/block/block_int.h | 2 +-
2 files changed, 11 insertions(+), 12 deletions(-)
diff --git a/block.c b/block.c
index 0ae2e93..25976b5 100644
--- a/block.c
+++ b/block.c
@@ -141,7 +141,6 @@ void bdrv_io_limits_disable(BlockDriverState *bs)
bs->slice_start = 0;
bs->slice_end = 0;
bs->slice_time = 0;
- memset(&bs->io_base, 0, sizeof(bs->io_base));
}
static void bdrv_block_timer(void *opaque)
@@ -1436,8 +1435,8 @@ static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
bs_dest->slice_time = bs_src->slice_time;
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->io_base = bs_src->io_base;
bs_dest->throttled_reqs = bs_src->throttled_reqs;
bs_dest->block_timer = bs_src->block_timer;
bs_dest->io_limits_enabled = bs_src->io_limits_enabled;
@@ -3768,9 +3767,9 @@ static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors,
slice_time = bs->slice_end - bs->slice_start;
slice_time /= (NANOSECONDS_PER_SECOND);
bytes_limit = bps_limit * slice_time;
- bytes_base = bs->nr_bytes[is_write] - bs->io_base.bytes[is_write];
+ bytes_base = bs->slice_submitted.bytes[is_write];
if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
- bytes_base += bs->nr_bytes[!is_write] - bs->io_base.bytes[!is_write];
+ bytes_base += bs->slice_submitted.bytes[!is_write];
}
/* bytes_base: the bytes of data which have been read/written; and
@@ -3828,9 +3827,9 @@ static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write,
slice_time = bs->slice_end - bs->slice_start;
slice_time /= (NANOSECONDS_PER_SECOND);
ios_limit = iops_limit * slice_time;
- ios_base = bs->nr_ops[is_write] - bs->io_base.ios[is_write];
+ ios_base = bs->slice_submitted.ios[is_write];
if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
- ios_base += bs->nr_ops[!is_write] - bs->io_base.ios[!is_write];
+ ios_base += bs->slice_submitted.ios[!is_write];
}
if (ios_base + 1 <= ios_limit) {
@@ -3875,11 +3874,7 @@ static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
bs->slice_start = now;
bs->slice_end = now + bs->slice_time;
- bs->io_base.bytes[is_write] = bs->nr_bytes[is_write];
- bs->io_base.bytes[!is_write] = bs->nr_bytes[!is_write];
-
- bs->io_base.ios[is_write] = bs->nr_ops[is_write];
- bs->io_base.ios[!is_write] = bs->nr_ops[!is_write];
+ memset(&bs->slice_submitted, 0, sizeof(bs->slice_submitted));
}
elapsed_time = now - bs->slice_start;
@@ -3907,6 +3902,10 @@ static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
*wait = 0;
}
+ bs->slice_submitted.bytes[is_write] += (int64_t)nb_sectors *
+ BDRV_SECTOR_SIZE;
+ bs->slice_submitted.ios[is_write]++;
+
return false;
}
diff --git a/include/block/block_int.h b/include/block/block_int.h
index 0986a2d..83941d8 100644
--- a/include/block/block_int.h
+++ b/include/block/block_int.h
@@ -256,7 +256,7 @@ struct BlockDriverState {
int64_t slice_start;
int64_t slice_end;
BlockIOLimit io_limits;
- BlockIOBaseValue io_base;
+ BlockIOBaseValue slice_submitted;
CoQueue throttled_reqs;
QEMUTimer *block_timer;
bool io_limits_enabled;
--
1.8.1.4
next prev parent reply other threads:[~2013-04-05 13:28 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-04-05 13:28 [Qemu-devel] [PULL 0/8] Block patches Kevin Wolf
2013-04-05 13:28 ` [Qemu-devel] [PATCH 1/8] virtio-blk-x: fix configuration synchronization Kevin Wolf
2013-04-05 16:47 ` Anthony Liguori
2013-04-05 17:56 ` Peter Maydell
2013-04-05 19:24 ` Anthony Liguori
2013-04-06 19:31 ` KONRAD Frédéric
2013-04-05 13:28 ` [Qemu-devel] [PATCH 2/8] usb-storage: Forward serial number to scsi-disk Kevin Wolf
2013-04-05 13:28 ` Kevin Wolf [this message]
2013-04-05 13:28 ` [Qemu-devel] [PATCH 4/8] block: keep I/O throttling slice time constant Kevin Wolf
2013-04-05 13:28 ` [Qemu-devel] [PATCH 5/8] block: drop duplicated slice extension code Kevin Wolf
2013-04-05 13:28 ` [Qemu-devel] [PATCH 6/8] block: clean up I/O throttling wait_time code Kevin Wolf
2013-04-05 13:28 ` [Qemu-devel] [PATCH 7/8] qcow2: Return real error in qcow2_update_snapshot_refcount Kevin Wolf
2013-04-05 13:28 ` [Qemu-devel] [PATCH 8/8] qcow2: Fix L1 write error handling " Kevin Wolf
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1365168490-29616-4-git-send-email-kwolf@redhat.com \
--to=kwolf@redhat.com \
--cc=anthony@codemonkey.ws \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).