* [Qemu-devel] [PATCH v10 1/3] block: add the command line support
2011-11-01 7:40 [Qemu-devel] [PATCH v10 0/3] The intro of QEMU block I/O throttling Zhi Yong Wu
@ 2011-11-01 7:40 ` Zhi Yong Wu
2011-11-01 22:43 ` Ryan Harper
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 2/3] block: add I/O throttling algorithm Zhi Yong Wu
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 3/3] hmp/qmp: add block_set_io_throttle Zhi Yong Wu
2 siblings, 1 reply; 9+ messages in thread
From: Zhi Yong Wu @ 2011-11-01 7:40 UTC (permalink / raw)
To: kwolf; +Cc: zwu.kernel, wuzhy, qemu-devel, stefanha
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
---
block.c | 40 ++++++++++++++++++++++++++++++++++++++++
block.h | 4 ++++
block_int.h | 29 +++++++++++++++++++++++++++++
blockdev.c | 32 ++++++++++++++++++++++++++++++++
qemu-config.c | 24 ++++++++++++++++++++++++
qemu-options.hx | 1 +
6 files changed, 130 insertions(+), 0 deletions(-)
diff --git a/block.c b/block.c
index 9bb236c..8f08dc5 100644
--- a/block.c
+++ b/block.c
@@ -30,6 +30,7 @@
#include "qjson.h"
#include "qemu-coroutine.h"
#include "qmp-commands.h"
+#include "qemu-timer.h"
#ifdef CONFIG_BSD
#include <sys/types.h>
@@ -105,6 +106,37 @@ int is_windows_drive(const char *filename)
}
#endif
+/* throttling disk I/O limits */
+static void bdrv_block_timer(void *opaque)
+{
+ BlockDriverState *bs = opaque;
+
+ qemu_co_queue_next(&bs->throttled_reqs);
+}
+
+void bdrv_io_limits_enable(BlockDriverState *bs)
+{
+ bs->io_limits_enabled = true;
+ qemu_co_queue_init(&bs->throttled_reqs);
+
+ bs->block_timer = qemu_new_timer_ns(vm_clock, bdrv_block_timer, bs);
+ bs->slice_time = 5 * BLOCK_IO_SLICE_TIME;
+ bs->slice_start = qemu_get_clock_ns(vm_clock);
+ bs->slice_end = bs->slice_start + bs->slice_time;
+ memset(&bs->io_disps, 0, sizeof(bs->io_disps));
+}
+
+bool bdrv_io_limits_enabled(BlockDriverState *bs)
+{
+ 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];
+}
+
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
@@ -1519,6 +1551,14 @@ void bdrv_get_geometry_hint(BlockDriverState *bs,
*psecs = bs->secs;
}
+/* 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);
+}
+
/* Recognize floppy formats */
typedef struct FDFormat {
FDriveType drive;
diff --git a/block.h b/block.h
index 38cd748..bc8315d 100644
--- a/block.h
+++ b/block.h
@@ -89,6 +89,10 @@ void bdrv_info(Monitor *mon, QObject **ret_data);
void bdrv_stats_print(Monitor *mon, const QObject *data);
void bdrv_info_stats(Monitor *mon, QObject **ret_data);
+/* disk I/O throttling */
+void bdrv_io_limits_enable(BlockDriverState *bs);
+bool bdrv_io_limits_enabled(BlockDriverState *bs);
+
void bdrv_init(void);
void bdrv_init_with_whitelist(void);
BlockDriver *bdrv_find_protocol(const char *filename);
diff --git a/block_int.h b/block_int.h
index f4547f6..b835ef6 100644
--- a/block_int.h
+++ b/block_int.h
@@ -34,6 +34,12 @@
#define BLOCK_FLAG_ENCRYPT 1
#define BLOCK_FLAG_COMPAT6 4
+#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 BLOCK_OPT_SIZE "size"
#define BLOCK_OPT_ENCRYPT "encryption"
#define BLOCK_OPT_COMPAT6 "compat6"
@@ -50,6 +56,16 @@ typedef struct AIOPool {
BlockDriverAIOCB *free_aiocb;
} AIOPool;
+typedef struct BlockIOLimit {
+ uint64_t bps[3];
+ uint64_t iops[3];
+} BlockIOLimit;
+
+typedef struct BlockIODisp {
+ uint64_t bytes[2];
+ uint64_t ios[2];
+} BlockIODisp;
+
struct BlockDriver {
const char *format_name;
int instance_size;
@@ -184,6 +200,16 @@ struct BlockDriverState {
void *sync_aiocb;
+ /* the time for latest disk I/O */
+ int64_t slice_time;
+ int64_t slice_start;
+ int64_t slice_end;
+ BlockIOLimit io_limits;
+ BlockIODisp io_disps;
+ CoQueue throttled_reqs;
+ QEMUTimer *block_timer;
+ bool io_limits_enabled;
+
/* I/O stats (display with "info blockstats"). */
uint64_t nr_bytes[BDRV_MAX_IOTYPE];
uint64_t nr_ops[BDRV_MAX_IOTYPE];
@@ -227,6 +253,9 @@ void *qemu_aio_get(AIOPool *pool, BlockDriverState *bs,
BlockDriverCompletionFunc *cb, void *opaque);
void qemu_aio_release(void *p);
+void bdrv_set_io_limits(BlockDriverState *bs,
+ BlockIOLimit *io_limits);
+
#ifdef _WIN32
int is_windows_drive(const char *filename);
#endif
diff --git a/blockdev.c b/blockdev.c
index 0827bf7..faf8c56 100644
--- a/blockdev.c
+++ b/blockdev.c
@@ -235,6 +235,9 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
int on_read_error, on_write_error;
const char *devaddr;
DriveInfo *dinfo;
+ BlockIOLimit io_limits;
+ bool bps_iol;
+ bool iops_iol;
int snapshot = 0;
int ret;
@@ -353,6 +356,32 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
}
}
+ /* disk I/O throttling */
+ io_limits.bps[BLOCK_IO_LIMIT_TOTAL] =
+ qemu_opt_get_number(opts, "bps", 0);
+ io_limits.bps[BLOCK_IO_LIMIT_READ] =
+ qemu_opt_get_number(opts, "bps_rd", 0);
+ io_limits.bps[BLOCK_IO_LIMIT_WRITE] =
+ qemu_opt_get_number(opts, "bps_wr", 0);
+ io_limits.iops[BLOCK_IO_LIMIT_TOTAL] =
+ qemu_opt_get_number(opts, "iops", 0);
+ io_limits.iops[BLOCK_IO_LIMIT_READ] =
+ qemu_opt_get_number(opts, "iops_rd", 0);
+ io_limits.iops[BLOCK_IO_LIMIT_WRITE] =
+ qemu_opt_get_number(opts, "iops_wr", 0);
+
+ bps_iol = (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_iol = (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_iol || iops_iol) {
+ error_report("bps(iops) and bps_rd/bps_wr(iops_rd/iops_wr)"
+ "cannot be used at the same time");
+ return NULL;
+ }
+
on_write_error = BLOCK_ERR_STOP_ENOSPC;
if ((buf = qemu_opt_get(opts, "werror")) != NULL) {
if (type != IF_IDE && type != IF_SCSI && type != IF_VIRTIO && type != IF_NONE) {
@@ -460,6 +489,9 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
bdrv_set_on_error(dinfo->bdrv, on_read_error, on_write_error);
+ /* disk I/O throttling */
+ bdrv_set_io_limits(dinfo->bdrv, &io_limits);
+
switch(type) {
case IF_IDE:
case IF_SCSI:
diff --git a/qemu-config.c b/qemu-config.c
index 597d7e1..1aa080f 100644
--- a/qemu-config.c
+++ b/qemu-config.c
@@ -85,6 +85,30 @@ static QemuOptsList qemu_drive_opts = {
.name = "readonly",
.type = QEMU_OPT_BOOL,
.help = "open drive file as read-only",
+ },{
+ .name = "iops",
+ .type = QEMU_OPT_NUMBER,
+ .help = "limit total I/O operations per second",
+ },{
+ .name = "iops_rd",
+ .type = QEMU_OPT_NUMBER,
+ .help = "limit read operations per second",
+ },{
+ .name = "iops_wr",
+ .type = QEMU_OPT_NUMBER,
+ .help = "limit write operations per second",
+ },{
+ .name = "bps",
+ .type = QEMU_OPT_NUMBER,
+ .help = "limit total bytes per second",
+ },{
+ .name = "bps_rd",
+ .type = QEMU_OPT_NUMBER,
+ .help = "limit read bytes per second",
+ },{
+ .name = "bps_wr",
+ .type = QEMU_OPT_NUMBER,
+ .help = "limit write bytes per second",
},
{ /* end of list */ }
},
diff --git a/qemu-options.hx b/qemu-options.hx
index 681eaf1..25a7be7 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -136,6 +136,7 @@ DEF("drive", HAS_ARG, QEMU_OPTION_drive,
" [,cache=writethrough|writeback|none|directsync|unsafe][,format=f]\n"
" [,serial=s][,addr=A][,id=name][,aio=threads|native]\n"
" [,readonly=on|off]\n"
+ " [[,bps=b]|[[,bps_rd=r][,bps_wr=w]]][[,iops=i]|[[,iops_rd=r][,iops_wr=w]]\n"
" use 'file' as a drive image\n", QEMU_ARCH_ALL)
STEXI
@item -drive @var{option}[,@var{option}[,@var{option}[,...]]]
--
1.7.6
^ permalink raw reply related [flat|nested] 9+ messages in thread* Re: [Qemu-devel] [PATCH v10 1/3] block: add the command line support
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 1/3] block: add the command line support Zhi Yong Wu
@ 2011-11-01 22:43 ` Ryan Harper
2011-11-02 3:44 ` Zhi Yong Wu
0 siblings, 1 reply; 9+ messages in thread
From: Ryan Harper @ 2011-11-01 22:43 UTC (permalink / raw)
To: Zhi Yong Wu; +Cc: kwolf, zwu.kernel, qemu-devel, stefanha
* Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> [2011-11-01 02:44]:
> Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
> ---
> block.c | 40 ++++++++++++++++++++++++++++++++++++++++
> block.h | 4 ++++
> block_int.h | 29 +++++++++++++++++++++++++++++
> blockdev.c | 32 ++++++++++++++++++++++++++++++++
> qemu-config.c | 24 ++++++++++++++++++++++++
> qemu-options.hx | 1 +
> 6 files changed, 130 insertions(+), 0 deletions(-)
I suggest changing the Subject of the patch to include something about
blkio limits. This will in searching git log later that this patch was
introducing blkio command line support.
>
> diff --git a/block.c b/block.c
> index 9bb236c..8f08dc5 100644
> --- a/block.c
> +++ b/block.c
> @@ -30,6 +30,7 @@
> #include "qjson.h"
> #include "qemu-coroutine.h"
> #include "qmp-commands.h"
> +#include "qemu-timer.h"
>
> #ifdef CONFIG_BSD
> #include <sys/types.h>
> @@ -105,6 +106,37 @@ int is_windows_drive(const char *filename)
> }
> #endif
>
> +/* throttling disk I/O limits */
> +static void bdrv_block_timer(void *opaque)
> +{
> + BlockDriverState *bs = opaque;
> +
> + qemu_co_queue_next(&bs->throttled_reqs);
> +}
> +
> +void bdrv_io_limits_enable(BlockDriverState *bs)
> +{
> + bs->io_limits_enabled = true;
We may want to move this to the end of the function so we don't raise
the flag that we're enabled until we're done setting everything up.
I don't think we'll race right now, but no reason not to move this to
the end.
> + qemu_co_queue_init(&bs->throttled_reqs);
> +
> + bs->block_timer = qemu_new_timer_ns(vm_clock, bdrv_block_timer, bs);
> + bs->slice_time = 5 * BLOCK_IO_SLICE_TIME;
> + bs->slice_start = qemu_get_clock_ns(vm_clock);
> + bs->slice_end = bs->slice_start + bs->slice_time;
> + memset(&bs->io_disps, 0, sizeof(bs->io_disps));
> +}
> +
> +bool bdrv_io_limits_enabled(BlockDriverState *bs)
> +{
> + 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];
> +}
> +
> /* check if the path starts with "<protocol>:" */
> static int path_has_protocol(const char *path)
> {
> @@ -1519,6 +1551,14 @@ void bdrv_get_geometry_hint(BlockDriverState *bs,
> *psecs = bs->secs;
> }
>
> +/* 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);
> +}
> +
> /* Recognize floppy formats */
> typedef struct FDFormat {
> FDriveType drive;
> diff --git a/block.h b/block.h
> index 38cd748..bc8315d 100644
> --- a/block.h
> +++ b/block.h
> @@ -89,6 +89,10 @@ void bdrv_info(Monitor *mon, QObject **ret_data);
> void bdrv_stats_print(Monitor *mon, const QObject *data);
> void bdrv_info_stats(Monitor *mon, QObject **ret_data);
>
> +/* disk I/O throttling */
> +void bdrv_io_limits_enable(BlockDriverState *bs);
> +bool bdrv_io_limits_enabled(BlockDriverState *bs);
> +
> void bdrv_init(void);
> void bdrv_init_with_whitelist(void);
> BlockDriver *bdrv_find_protocol(const char *filename);
> diff --git a/block_int.h b/block_int.h
> index f4547f6..b835ef6 100644
> --- a/block_int.h
> +++ b/block_int.h
> @@ -34,6 +34,12 @@
> #define BLOCK_FLAG_ENCRYPT 1
> #define BLOCK_FLAG_COMPAT6 4
>
> +#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 BLOCK_OPT_SIZE "size"
> #define BLOCK_OPT_ENCRYPT "encryption"
> #define BLOCK_OPT_COMPAT6 "compat6"
> @@ -50,6 +56,16 @@ typedef struct AIOPool {
> BlockDriverAIOCB *free_aiocb;
> } AIOPool;
>
> +typedef struct BlockIOLimit {
> + uint64_t bps[3];
> + uint64_t iops[3];
> +} BlockIOLimit;
> +
> +typedef struct BlockIODisp {
> + uint64_t bytes[2];
> + uint64_t ios[2];
> +} BlockIODisp;
> +
> struct BlockDriver {
> const char *format_name;
> int instance_size;
> @@ -184,6 +200,16 @@ struct BlockDriverState {
>
> void *sync_aiocb;
>
> + /* the time for latest disk I/O */
> + int64_t slice_time;
> + int64_t slice_start;
> + int64_t slice_end;
> + BlockIOLimit io_limits;
> + BlockIODisp io_disps;
> + CoQueue throttled_reqs;
> + QEMUTimer *block_timer;
> + bool io_limits_enabled;
> +
> /* I/O stats (display with "info blockstats"). */
> uint64_t nr_bytes[BDRV_MAX_IOTYPE];
> uint64_t nr_ops[BDRV_MAX_IOTYPE];
> @@ -227,6 +253,9 @@ void *qemu_aio_get(AIOPool *pool, BlockDriverState *bs,
> BlockDriverCompletionFunc *cb, void *opaque);
> void qemu_aio_release(void *p);
>
> +void bdrv_set_io_limits(BlockDriverState *bs,
> + BlockIOLimit *io_limits);
> +
> #ifdef _WIN32
> int is_windows_drive(const char *filename);
> #endif
> diff --git a/blockdev.c b/blockdev.c
> index 0827bf7..faf8c56 100644
> --- a/blockdev.c
> +++ b/blockdev.c
> @@ -235,6 +235,9 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
> int on_read_error, on_write_error;
> const char *devaddr;
> DriveInfo *dinfo;
> + BlockIOLimit io_limits;
> + bool bps_iol;
> + bool iops_iol;
> int snapshot = 0;
> int ret;
>
> @@ -353,6 +356,32 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
> }
> }
>
> + /* disk I/O throttling */
> + io_limits.bps[BLOCK_IO_LIMIT_TOTAL] =
> + qemu_opt_get_number(opts, "bps", 0);
> + io_limits.bps[BLOCK_IO_LIMIT_READ] =
> + qemu_opt_get_number(opts, "bps_rd", 0);
> + io_limits.bps[BLOCK_IO_LIMIT_WRITE] =
> + qemu_opt_get_number(opts, "bps_wr", 0);
> + io_limits.iops[BLOCK_IO_LIMIT_TOTAL] =
> + qemu_opt_get_number(opts, "iops", 0);
> + io_limits.iops[BLOCK_IO_LIMIT_READ] =
> + qemu_opt_get_number(opts, "iops_rd", 0);
> + io_limits.iops[BLOCK_IO_LIMIT_WRITE] =
> + qemu_opt_get_number(opts, "iops_wr", 0);
> +
> + bps_iol = (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_iol = (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_iol || iops_iol) {
> + error_report("bps(iops) and bps_rd/bps_wr(iops_rd/iops_wr)"
> + "cannot be used at the same time");
> + return NULL;
> + }
> +
How about extracting this section into the bdrv_set_io_limits() function or some
other common setting function. This should be the same code that's used when
modifying/setting these values via the monitor as well (minus how you obtain the
input values - option parsings vs. monitor values).
> on_write_error = BLOCK_ERR_STOP_ENOSPC;
> if ((buf = qemu_opt_get(opts, "werror")) != NULL) {
> if (type != IF_IDE && type != IF_SCSI && type != IF_VIRTIO && type != IF_NONE) {
> @@ -460,6 +489,9 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
>
> bdrv_set_on_error(dinfo->bdrv, on_read_error, on_write_error);
>
> + /* disk I/O throttling */
> + bdrv_set_io_limits(dinfo->bdrv, &io_limits);
> +
> switch(type) {
> case IF_IDE:
> case IF_SCSI:
> diff --git a/qemu-config.c b/qemu-config.c
> index 597d7e1..1aa080f 100644
> --- a/qemu-config.c
> +++ b/qemu-config.c
> @@ -85,6 +85,30 @@ static QemuOptsList qemu_drive_opts = {
> .name = "readonly",
> .type = QEMU_OPT_BOOL,
> .help = "open drive file as read-only",
> + },{
> + .name = "iops",
> + .type = QEMU_OPT_NUMBER,
> + .help = "limit total I/O operations per second",
> + },{
> + .name = "iops_rd",
> + .type = QEMU_OPT_NUMBER,
> + .help = "limit read operations per second",
> + },{
> + .name = "iops_wr",
> + .type = QEMU_OPT_NUMBER,
> + .help = "limit write operations per second",
> + },{
> + .name = "bps",
> + .type = QEMU_OPT_NUMBER,
> + .help = "limit total bytes per second",
> + },{
> + .name = "bps_rd",
> + .type = QEMU_OPT_NUMBER,
> + .help = "limit read bytes per second",
> + },{
> + .name = "bps_wr",
> + .type = QEMU_OPT_NUMBER,
> + .help = "limit write bytes per second",
> },
> { /* end of list */ }
> },
> diff --git a/qemu-options.hx b/qemu-options.hx
> index 681eaf1..25a7be7 100644
> --- a/qemu-options.hx
> +++ b/qemu-options.hx
> @@ -136,6 +136,7 @@ DEF("drive", HAS_ARG, QEMU_OPTION_drive,
> " [,cache=writethrough|writeback|none|directsync|unsafe][,format=f]\n"
> " [,serial=s][,addr=A][,id=name][,aio=threads|native]\n"
> " [,readonly=on|off]\n"
> + " [[,bps=b]|[[,bps_rd=r][,bps_wr=w]]][[,iops=i]|[[,iops_rd=r][,iops_wr=w]]\n"
> " use 'file' as a drive image\n", QEMU_ARCH_ALL)
> STEXI
> @item -drive @var{option}[,@var{option}[,@var{option}[,...]]]
> --
> 1.7.6
>
--
Ryan Harper
Software Engineer; Linux Technology Center
IBM Corp., Austin, Tx
ryanh@us.ibm.com
^ permalink raw reply [flat|nested] 9+ messages in thread* Re: [Qemu-devel] [PATCH v10 1/3] block: add the command line support
2011-11-01 22:43 ` Ryan Harper
@ 2011-11-02 3:44 ` Zhi Yong Wu
0 siblings, 0 replies; 9+ messages in thread
From: Zhi Yong Wu @ 2011-11-02 3:44 UTC (permalink / raw)
To: Ryan Harper; +Cc: kwolf, Zhi Yong Wu, qemu-devel, stefanha
On Wed, Nov 2, 2011 at 6:43 AM, Ryan Harper <ryanh@us.ibm.com> wrote:
> * Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> [2011-11-01 02:44]:
>> Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
>> ---
>> block.c | 40 ++++++++++++++++++++++++++++++++++++++++
>> block.h | 4 ++++
>> block_int.h | 29 +++++++++++++++++++++++++++++
>> blockdev.c | 32 ++++++++++++++++++++++++++++++++
>> qemu-config.c | 24 ++++++++++++++++++++++++
>> qemu-options.hx | 1 +
>> 6 files changed, 130 insertions(+), 0 deletions(-)
>
> I suggest changing the Subject of the patch to include something about
> blkio limits. This will in searching git log later that this patch was
> introducing blkio command line support.
Good suggestion
>
>>
>> diff --git a/block.c b/block.c
>> index 9bb236c..8f08dc5 100644
>> --- a/block.c
>> +++ b/block.c
>> @@ -30,6 +30,7 @@
>> #include "qjson.h"
>> #include "qemu-coroutine.h"
>> #include "qmp-commands.h"
>> +#include "qemu-timer.h"
>>
>> #ifdef CONFIG_BSD
>> #include <sys/types.h>
>> @@ -105,6 +106,37 @@ int is_windows_drive(const char *filename)
>> }
>> #endif
>>
>> +/* throttling disk I/O limits */
>> +static void bdrv_block_timer(void *opaque)
>> +{
>> + BlockDriverState *bs = opaque;
>> +
>> + qemu_co_queue_next(&bs->throttled_reqs);
>> +}
>> +
>> +void bdrv_io_limits_enable(BlockDriverState *bs)
>> +{
>> + bs->io_limits_enabled = true;
>
> We may want to move this to the end of the function so we don't raise
> the flag that we're enabled until we're done setting everything up.
> I don't think we'll race right now, but no reason not to move this to
> the end.
OK.
>
>> + qemu_co_queue_init(&bs->throttled_reqs);
>> +
>> + bs->block_timer = qemu_new_timer_ns(vm_clock, bdrv_block_timer, bs);
>> + bs->slice_time = 5 * BLOCK_IO_SLICE_TIME;
>> + bs->slice_start = qemu_get_clock_ns(vm_clock);
>> + bs->slice_end = bs->slice_start + bs->slice_time;
>> + memset(&bs->io_disps, 0, sizeof(bs->io_disps));
>> +}
>> +
>> +bool bdrv_io_limits_enabled(BlockDriverState *bs)
>> +{
>> + 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];
>> +}
>> +
>> /* check if the path starts with "<protocol>:" */
>> static int path_has_protocol(const char *path)
>> {
>> @@ -1519,6 +1551,14 @@ void bdrv_get_geometry_hint(BlockDriverState *bs,
>> *psecs = bs->secs;
>> }
>>
>> +/* 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);
>> +}
>> +
>> /* Recognize floppy formats */
>> typedef struct FDFormat {
>> FDriveType drive;
>> diff --git a/block.h b/block.h
>> index 38cd748..bc8315d 100644
>> --- a/block.h
>> +++ b/block.h
>> @@ -89,6 +89,10 @@ void bdrv_info(Monitor *mon, QObject **ret_data);
>> void bdrv_stats_print(Monitor *mon, const QObject *data);
>> void bdrv_info_stats(Monitor *mon, QObject **ret_data);
>>
>> +/* disk I/O throttling */
>> +void bdrv_io_limits_enable(BlockDriverState *bs);
>> +bool bdrv_io_limits_enabled(BlockDriverState *bs);
>> +
>> void bdrv_init(void);
>> void bdrv_init_with_whitelist(void);
>> BlockDriver *bdrv_find_protocol(const char *filename);
>> diff --git a/block_int.h b/block_int.h
>> index f4547f6..b835ef6 100644
>> --- a/block_int.h
>> +++ b/block_int.h
>> @@ -34,6 +34,12 @@
>> #define BLOCK_FLAG_ENCRYPT 1
>> #define BLOCK_FLAG_COMPAT6 4
>>
>> +#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 BLOCK_OPT_SIZE "size"
>> #define BLOCK_OPT_ENCRYPT "encryption"
>> #define BLOCK_OPT_COMPAT6 "compat6"
>> @@ -50,6 +56,16 @@ typedef struct AIOPool {
>> BlockDriverAIOCB *free_aiocb;
>> } AIOPool;
>>
>> +typedef struct BlockIOLimit {
>> + uint64_t bps[3];
>> + uint64_t iops[3];
>> +} BlockIOLimit;
>> +
>> +typedef struct BlockIODisp {
>> + uint64_t bytes[2];
>> + uint64_t ios[2];
>> +} BlockIODisp;
>> +
>> struct BlockDriver {
>> const char *format_name;
>> int instance_size;
>> @@ -184,6 +200,16 @@ struct BlockDriverState {
>>
>> void *sync_aiocb;
>>
>> + /* the time for latest disk I/O */
>> + int64_t slice_time;
>> + int64_t slice_start;
>> + int64_t slice_end;
>> + BlockIOLimit io_limits;
>> + BlockIODisp io_disps;
>> + CoQueue throttled_reqs;
>> + QEMUTimer *block_timer;
>> + bool io_limits_enabled;
>> +
>> /* I/O stats (display with "info blockstats"). */
>> uint64_t nr_bytes[BDRV_MAX_IOTYPE];
>> uint64_t nr_ops[BDRV_MAX_IOTYPE];
>> @@ -227,6 +253,9 @@ void *qemu_aio_get(AIOPool *pool, BlockDriverState *bs,
>> BlockDriverCompletionFunc *cb, void *opaque);
>> void qemu_aio_release(void *p);
>>
>> +void bdrv_set_io_limits(BlockDriverState *bs,
>> + BlockIOLimit *io_limits);
>> +
>> #ifdef _WIN32
>> int is_windows_drive(const char *filename);
>> #endif
>> diff --git a/blockdev.c b/blockdev.c
>> index 0827bf7..faf8c56 100644
>> --- a/blockdev.c
>> +++ b/blockdev.c
>> @@ -235,6 +235,9 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
>> int on_read_error, on_write_error;
>> const char *devaddr;
>> DriveInfo *dinfo;
>> + BlockIOLimit io_limits;
>> + bool bps_iol;
>> + bool iops_iol;
>> int snapshot = 0;
>> int ret;
>>
>> @@ -353,6 +356,32 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
>> }
>> }
>>
>> + /* disk I/O throttling */
>> + io_limits.bps[BLOCK_IO_LIMIT_TOTAL] =
>> + qemu_opt_get_number(opts, "bps", 0);
>> + io_limits.bps[BLOCK_IO_LIMIT_READ] =
>> + qemu_opt_get_number(opts, "bps_rd", 0);
>> + io_limits.bps[BLOCK_IO_LIMIT_WRITE] =
>> + qemu_opt_get_number(opts, "bps_wr", 0);
>> + io_limits.iops[BLOCK_IO_LIMIT_TOTAL] =
>> + qemu_opt_get_number(opts, "iops", 0);
>> + io_limits.iops[BLOCK_IO_LIMIT_READ] =
>> + qemu_opt_get_number(opts, "iops_rd", 0);
>> + io_limits.iops[BLOCK_IO_LIMIT_WRITE] =
>> + qemu_opt_get_number(opts, "iops_wr", 0);
>> +
>> + bps_iol = (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_iol = (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_iol || iops_iol) {
>> + error_report("bps(iops) and bps_rd/bps_wr(iops_rd/iops_wr)"
>> + "cannot be used at the same time");
>> + return NULL;
>> + }
>> +
>
>
> How about extracting this section into the bdrv_set_io_limits() function or some
> other common setting function. This should be the same code that's used when
Good idea.
> modifying/setting these values via the monitor as well (minus how you obtain the
> input values - option parsings vs. monitor values).
For monitor, they are obtained via the way below:
In blockdev.c:do_block_set_io_throttle(),
int64_t bps = qdict_get_try_int(qdict, "bps", -1);
int64_t bps_rd = qdict_get_try_int(qdict, "bps_rd", -1);
int64_t bps_wr = qdict_get_try_int(qdict, "bps_wr", -1);
int64_t iops = qdict_get_try_int(qdict, "iops", -1);
int64_t iops_rd = qdict_get_try_int(qdict, "iops_rd", -1);
int64_t iops_wr = qdict_get_try_int(qdict, "iops_wr", -1);
When they are not specified, the handling method is a bit different.
Actually they are also limited when these values are modified.
if ((bps != 0 && (bps_rd != 0 || bps_wr != 0))
|| (iops != 0 && (iops_rd != 0 || iops_wr != 0))) {
qerror_report(QERR_INVALID_PARAMETER_COMBINATION);
return -1;
}
But i also think that it is one good idea to put them in one common function.
>
>> on_write_error = BLOCK_ERR_STOP_ENOSPC;
>> if ((buf = qemu_opt_get(opts, "werror")) != NULL) {
>> if (type != IF_IDE && type != IF_SCSI && type != IF_VIRTIO && type != IF_NONE) {
>> @@ -460,6 +489,9 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
>>
>> bdrv_set_on_error(dinfo->bdrv, on_read_error, on_write_error);
>>
>> + /* disk I/O throttling */
>> + bdrv_set_io_limits(dinfo->bdrv, &io_limits);
>> +
>> switch(type) {
>> case IF_IDE:
>> case IF_SCSI:
>> diff --git a/qemu-config.c b/qemu-config.c
>> index 597d7e1..1aa080f 100644
>> --- a/qemu-config.c
>> +++ b/qemu-config.c
>> @@ -85,6 +85,30 @@ static QemuOptsList qemu_drive_opts = {
>> .name = "readonly",
>> .type = QEMU_OPT_BOOL,
>> .help = "open drive file as read-only",
>> + },{
>> + .name = "iops",
>> + .type = QEMU_OPT_NUMBER,
>> + .help = "limit total I/O operations per second",
>> + },{
>> + .name = "iops_rd",
>> + .type = QEMU_OPT_NUMBER,
>> + .help = "limit read operations per second",
>> + },{
>> + .name = "iops_wr",
>> + .type = QEMU_OPT_NUMBER,
>> + .help = "limit write operations per second",
>> + },{
>> + .name = "bps",
>> + .type = QEMU_OPT_NUMBER,
>> + .help = "limit total bytes per second",
>> + },{
>> + .name = "bps_rd",
>> + .type = QEMU_OPT_NUMBER,
>> + .help = "limit read bytes per second",
>> + },{
>> + .name = "bps_wr",
>> + .type = QEMU_OPT_NUMBER,
>> + .help = "limit write bytes per second",
>> },
>> { /* end of list */ }
>> },
>> diff --git a/qemu-options.hx b/qemu-options.hx
>> index 681eaf1..25a7be7 100644
>> --- a/qemu-options.hx
>> +++ b/qemu-options.hx
>> @@ -136,6 +136,7 @@ DEF("drive", HAS_ARG, QEMU_OPTION_drive,
>> " [,cache=writethrough|writeback|none|directsync|unsafe][,format=f]\n"
>> " [,serial=s][,addr=A][,id=name][,aio=threads|native]\n"
>> " [,readonly=on|off]\n"
>> + " [[,bps=b]|[[,bps_rd=r][,bps_wr=w]]][[,iops=i]|[[,iops_rd=r][,iops_wr=w]]\n"
>> " use 'file' as a drive image\n", QEMU_ARCH_ALL)
>> STEXI
>> @item -drive @var{option}[,@var{option}[,@var{option}[,...]]]
>> --
>> 1.7.6
>>
>
> --
> Ryan Harper
> Software Engineer; Linux Technology Center
> IBM Corp., Austin, Tx
> ryanh@us.ibm.com
>
>
>
--
Regards,
Zhi Yong Wu
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Qemu-devel] [PATCH v10 2/3] block: add I/O throttling algorithm
2011-11-01 7:40 [Qemu-devel] [PATCH v10 0/3] The intro of QEMU block I/O throttling Zhi Yong Wu
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 1/3] block: add the command line support Zhi Yong Wu
@ 2011-11-01 7:40 ` Zhi Yong Wu
2011-11-01 11:33 ` Stefan Hajnoczi
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 3/3] hmp/qmp: add block_set_io_throttle Zhi Yong Wu
2 siblings, 1 reply; 9+ messages in thread
From: Zhi Yong Wu @ 2011-11-01 7:40 UTC (permalink / raw)
To: kwolf; +Cc: zwu.kernel, wuzhy, qemu-devel, stefanha
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
---
block.c | 228 +++++++++++++++++++++++++++++++++++++++++++++++++
block.h | 1 +
block_int.h | 1 +
qemu-coroutine-lock.c | 8 ++
qemu-coroutine.h | 6 ++
5 files changed, 244 insertions(+), 0 deletions(-)
diff --git a/block.c b/block.c
index 8f08dc5..cb89372 100644
--- a/block.c
+++ b/block.c
@@ -74,6 +74,13 @@ static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
bool is_write);
static void coroutine_fn bdrv_co_do_rw(void *opaque);
+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);
@@ -107,6 +114,28 @@ int is_windows_drive(const char *filename)
#endif
/* throttling disk I/O limits */
+void bdrv_io_limits_disable(BlockDriverState *bs)
+{
+ bs->io_limits_enabled = false;
+
+ if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
+ while (qemu_co_queue_next(&bs->throttled_reqs));
+ }
+
+ qemu_co_queue_init(&bs->throttled_reqs);
+
+ if (bs->block_timer) {
+ qemu_del_timer(bs->block_timer);
+ qemu_free_timer(bs->block_timer);
+ bs->block_timer = NULL;
+ }
+
+ bs->slice_start = 0;
+ bs->slice_end = 0;
+ bs->slice_time = 0;
+ memset(&bs->io_disps, 0, sizeof(bs->io_disps));
+}
+
static void bdrv_block_timer(void *opaque)
{
BlockDriverState *bs = opaque;
@@ -137,6 +166,33 @@ bool bdrv_io_limits_enabled(BlockDriverState *bs)
|| io_limits->iops[BLOCK_IO_LIMIT_TOTAL];
}
+static void bdrv_io_limits_intercept(BlockDriverState *bs,
+ int nb_sectors)
+{
+ int64_t wait_time = -1;
+
+ if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
+ qemu_co_queue_wait(&bs->throttled_reqs);
+ goto resume;
+ } else if (bdrv_exceed_io_limits(bs, nb_sectors, false, &wait_time)) {
+ if (wait_time != -1) {
+ qemu_mod_timer(bs->block_timer,
+ wait_time + qemu_get_clock_ns(vm_clock));
+ }
+
+ qemu_co_queue_wait(&bs->throttled_reqs);
+
+resume:
+ while (bdrv_exceed_io_limits(bs, nb_sectors, false, &wait_time)) {
+ qemu_mod_timer(bs->block_timer,
+ wait_time + qemu_get_clock_ns(vm_clock));
+ qemu_co_queue_wait_insert_head(&bs->throttled_reqs);
+ }
+
+ qemu_co_queue_next(&bs->throttled_reqs);
+ }
+}
+
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
@@ -719,6 +775,11 @@ int bdrv_open(BlockDriverState *bs, const char *filename, int flags,
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:
@@ -754,6 +815,9 @@ void bdrv_close(BlockDriverState *bs)
bdrv_dev_change_media_cb(bs, false);
}
+
+ /*throttling disk I/O limits*/
+ bdrv_io_limits_disable(bs);
}
void bdrv_close_all(void)
@@ -1292,6 +1356,11 @@ 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, nb_sectors);
+ }
+
return drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
}
@@ -1322,6 +1391,11 @@ 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, nb_sectors);
+ }
+
ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
if (bs->dirty_bitmap) {
@@ -2513,6 +2587,160 @@ void bdrv_aio_cancel(BlockDriverAIOCB *acb)
acb->pool->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;
+ double bytes_limit, bytes_disp, 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_disp = bs->nr_bytes[is_write] - bs->io_disps.bytes[is_write];
+ if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
+ bytes_disp += bs->nr_bytes[!is_write] - bs->io_disps.bytes[!is_write];
+ }
+
+ bytes_res = (unsigned) nb_sectors * BDRV_SECTOR_SIZE;
+
+ if (bytes_disp + bytes_res <= bytes_limit) {
+ if (wait) {
+ *wait = 0;
+ }
+
+ return false;
+ }
+
+ /* Calc approx time to dispatch */
+ wait_time = (bytes_disp + bytes_res) / bps_limit - elapsed_time;
+
+ bs->slice_time = wait_time * BLOCK_IO_SLICE_TIME * 10;
+ bs->slice_end += bs->slice_time - 3 * BLOCK_IO_SLICE_TIME;
+ if (wait) {
+ *wait = wait_time * BLOCK_IO_SLICE_TIME * 10;
+ }
+
+ 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_disp;
+ 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_disp = bs->nr_ops[is_write] - bs->io_disps.ios[is_write];
+ if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
+ ios_disp += bs->nr_ops[!is_write] - bs->io_disps.ios[!is_write];
+ }
+
+ if (ios_disp + 1 <= ios_limit) {
+ if (wait) {
+ *wait = 0;
+ }
+
+ return false;
+ }
+
+ /* Calc approx time to dispatch */
+ wait_time = (ios_disp + 1) / iops_limit;
+ if (wait_time > elapsed_time) {
+ wait_time = wait_time - elapsed_time;
+ } else {
+ wait_time = 0;
+ }
+
+ bs->slice_time = wait_time * BLOCK_IO_SLICE_TIME * 10;
+ bs->slice_end += bs->slice_time - 3 * BLOCK_IO_SLICE_TIME;
+ if (wait) {
+ *wait = wait_time * BLOCK_IO_SLICE_TIME * 10;
+ }
+
+ 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_get_clock_ns(vm_clock);
+ if ((bs->slice_start < now)
+ && (bs->slice_end > now)) {
+ bs->slice_end = now + bs->slice_time;
+ } else {
+ bs->slice_time = 5 * BLOCK_IO_SLICE_TIME;
+ bs->slice_start = now;
+ bs->slice_end = now + bs->slice_time;
+
+ bs->io_disps.bytes[is_write] = bs->nr_bytes[is_write];
+ bs->io_disps.bytes[!is_write] = bs->nr_bytes[!is_write];
+
+ bs->io_disps.ios[is_write] = bs->nr_ops[is_write];
+ bs->io_disps.ios[!is_write] = bs->nr_ops[!is_write];
+ }
+
+ 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 (!qemu_co_queue_empty(&bs->throttled_reqs)) {
+ max_wait = -1;
+ }
+
+ if (wait) {
+ *wait = max_wait;
+ }
+
+ now = qemu_get_clock_ns(vm_clock);
+ if (bs->slice_end < now + max_wait) {
+ bs->slice_end = now + max_wait;
+ }
+
+ return true;
+ }
+
+ if (wait) {
+ *wait = 0;
+ }
+
+ return false;
+}
/**************************************************************/
/* async block device emulation */
diff --git a/block.h b/block.h
index bc8315d..9b5b35f 100644
--- a/block.h
+++ b/block.h
@@ -91,6 +91,7 @@ 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);
diff --git a/block_int.h b/block_int.h
index b835ef6..fc53f5b 100644
--- a/block_int.h
+++ b/block_int.h
@@ -39,6 +39,7 @@
#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"
diff --git a/qemu-coroutine-lock.c b/qemu-coroutine-lock.c
index 6b58160..9549c07 100644
--- a/qemu-coroutine-lock.c
+++ b/qemu-coroutine-lock.c
@@ -61,6 +61,14 @@ void coroutine_fn qemu_co_queue_wait(CoQueue *queue)
assert(qemu_in_coroutine());
}
+void coroutine_fn qemu_co_queue_wait_insert_head(CoQueue *queue)
+{
+ Coroutine *self = qemu_coroutine_self();
+ QTAILQ_INSERT_HEAD(&queue->entries, self, co_queue_next);
+ qemu_coroutine_yield();
+ assert(qemu_in_coroutine());
+}
+
bool qemu_co_queue_next(CoQueue *queue)
{
Coroutine *next;
diff --git a/qemu-coroutine.h b/qemu-coroutine.h
index b8fc4f4..8a2e5d2 100644
--- a/qemu-coroutine.h
+++ b/qemu-coroutine.h
@@ -118,6 +118,12 @@ void qemu_co_queue_init(CoQueue *queue);
void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
/**
+ * Adds the current coroutine to the head of the CoQueue and transfers control to the
+ * caller of the coroutine.
+ */
+void coroutine_fn qemu_co_queue_wait_insert_head(CoQueue *queue);
+
+/**
* Restarts the next coroutine in the CoQueue and removes it from the queue.
*
* Returns true if a coroutine was restarted, false if the queue is empty.
--
1.7.6
^ permalink raw reply related [flat|nested] 9+ messages in thread* Re: [Qemu-devel] [PATCH v10 2/3] block: add I/O throttling algorithm
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 2/3] block: add I/O throttling algorithm Zhi Yong Wu
@ 2011-11-01 11:33 ` Stefan Hajnoczi
2011-11-01 11:40 ` Zhi Yong Wu
0 siblings, 1 reply; 9+ messages in thread
From: Stefan Hajnoczi @ 2011-11-01 11:33 UTC (permalink / raw)
To: Zhi Yong Wu; +Cc: kwolf, zwu.kernel, qemu-devel, stefanha
On Tue, Nov 1, 2011 at 7:40 AM, Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> wrote:
> +static void bdrv_io_limits_intercept(BlockDriverState *bs,
> + int nb_sectors)
> +{
> + int64_t wait_time = -1;
> +
> + if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
> + qemu_co_queue_wait(&bs->throttled_reqs);
> + goto resume;
> + } else if (bdrv_exceed_io_limits(bs, nb_sectors, false, &wait_time)) {
> + if (wait_time != -1) {
> + qemu_mod_timer(bs->block_timer,
> + wait_time + qemu_get_clock_ns(vm_clock));
> + }
> +
> + qemu_co_queue_wait(&bs->throttled_reqs);
> +
> +resume:
> + while (bdrv_exceed_io_limits(bs, nb_sectors, false, &wait_time)) {
is_write needs to be passed in to bdrv_exceed_io_limits(). Currently
this accounts every I/O as a read.
> + qemu_mod_timer(bs->block_timer,
> + wait_time + qemu_get_clock_ns(vm_clock));
Do you need if (wait_time != -1) here?
Stefan
^ permalink raw reply [flat|nested] 9+ messages in thread* Re: [Qemu-devel] [PATCH v10 2/3] block: add I/O throttling algorithm
2011-11-01 11:33 ` Stefan Hajnoczi
@ 2011-11-01 11:40 ` Zhi Yong Wu
0 siblings, 0 replies; 9+ messages in thread
From: Zhi Yong Wu @ 2011-11-01 11:40 UTC (permalink / raw)
To: Stefan Hajnoczi; +Cc: kwolf, Zhi Yong Wu, qemu-devel, stefanha
On Tue, Nov 1, 2011 at 7:33 PM, Stefan Hajnoczi <stefanha@gmail.com> wrote:
> On Tue, Nov 1, 2011 at 7:40 AM, Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> wrote:
>> +static void bdrv_io_limits_intercept(BlockDriverState *bs,
>> + int nb_sectors)
>> +{
>> + int64_t wait_time = -1;
>> +
>> + if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
>> + qemu_co_queue_wait(&bs->throttled_reqs);
>> + goto resume;
>> + } else if (bdrv_exceed_io_limits(bs, nb_sectors, false, &wait_time)) {
>> + if (wait_time != -1) {
>> + qemu_mod_timer(bs->block_timer,
>> + wait_time + qemu_get_clock_ns(vm_clock));
>> + }
>> +
>> + qemu_co_queue_wait(&bs->throttled_reqs);
>> +
>> +resume:
>> + while (bdrv_exceed_io_limits(bs, nb_sectors, false, &wait_time)) {
>
> is_write needs to be passed in to bdrv_exceed_io_limits(). Currently
> this accounts every I/O as a read.
Sorry, It is one stupid error.
>
>> + qemu_mod_timer(bs->block_timer,
>> + wait_time + qemu_get_clock_ns(vm_clock));
>
> Do you need if (wait_time != -1) here?
Actually i think that we can drop the condition in our code.
>
> Stefan
>
--
Regards,
Zhi Yong Wu
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Qemu-devel] [PATCH v10 3/3] hmp/qmp: add block_set_io_throttle
2011-11-01 7:40 [Qemu-devel] [PATCH v10 0/3] The intro of QEMU block I/O throttling Zhi Yong Wu
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 1/3] block: add the command line support Zhi Yong Wu
2011-11-01 7:40 ` [Qemu-devel] [PATCH v10 2/3] block: add I/O throttling algorithm Zhi Yong Wu
@ 2011-11-01 7:40 ` Zhi Yong Wu
2 siblings, 0 replies; 9+ messages in thread
From: Zhi Yong Wu @ 2011-11-01 7:40 UTC (permalink / raw)
To: kwolf; +Cc: zwu.kernel, wuzhy, qemu-devel, stefanha
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
---
block.c | 15 +++++++++++++++
blockdev.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++++++
blockdev.h | 2 ++
hmp-commands.hx | 15 +++++++++++++++
hmp.c | 10 ++++++++++
qapi-schema.json | 16 +++++++++++++++-
qerror.c | 4 ++++
qerror.h | 3 +++
qmp-commands.hx | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
9 files changed, 167 insertions(+), 2 deletions(-)
diff --git a/block.c b/block.c
index cb89372..7d856c2 100644
--- a/block.c
+++ b/block.c
@@ -1976,6 +1976,21 @@ BlockInfoList *qmp_query_block(Error **errp)
info->value->inserted->has_backing_file = true;
info->value->inserted->backing_file = g_strdup(bs->backing_file);
}
+
+ if (bs->io_limits_enabled) {
+ info->value->inserted->bps =
+ bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL];
+ info->value->inserted->bps_rd =
+ bs->io_limits.bps[BLOCK_IO_LIMIT_READ];
+ info->value->inserted->bps_wr =
+ bs->io_limits.bps[BLOCK_IO_LIMIT_WRITE];
+ info->value->inserted->iops =
+ bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL];
+ info->value->inserted->iops_rd =
+ bs->io_limits.iops[BLOCK_IO_LIMIT_READ];
+ info->value->inserted->iops_wr =
+ bs->io_limits.iops[BLOCK_IO_LIMIT_WRITE];
+ }
}
/* XXX: waiting for the qapi to support GSList */
diff --git a/blockdev.c b/blockdev.c
index faf8c56..9eed973 100644
--- a/blockdev.c
+++ b/blockdev.c
@@ -745,6 +745,57 @@ int do_change_block(Monitor *mon, const char *device,
return monitor_read_bdrv_key_start(mon, bs, NULL, NULL);
}
+/* throttling disk I/O limits */
+int do_block_set_io_throttle(Monitor *mon,
+ const QDict *qdict, QObject **ret_data)
+{
+ const char *devname = qdict_get_str(qdict, "device");
+ int64_t bps = qdict_get_try_int(qdict, "bps", -1);
+ int64_t bps_rd = qdict_get_try_int(qdict, "bps_rd", -1);
+ int64_t bps_wr = qdict_get_try_int(qdict, "bps_wr", -1);
+ int64_t iops = qdict_get_try_int(qdict, "iops", -1);
+ int64_t iops_rd = qdict_get_try_int(qdict, "iops_rd", -1);
+ int64_t iops_wr = qdict_get_try_int(qdict, "iops_wr", -1);
+ BlockDriverState *bs;
+
+ bs = bdrv_find(devname);
+ if (!bs) {
+ qerror_report(QERR_DEVICE_NOT_FOUND, devname);
+ return -1;
+ }
+
+ if ((bps == -1) || (bps_rd == -1) || (bps_wr == -1)
+ || (iops == -1) || (iops_rd == -1) || (iops_wr == -1)) {
+ qerror_report(QERR_MISSING_PARAMETER,
+ "bps/bps_rd/bps_wr/iops/iops_rd/iops_wr");
+ return -1;
+ }
+
+ if ((bps != 0 && (bps_rd != 0 || bps_wr != 0))
+ || (iops != 0 && (iops_rd != 0 || iops_wr != 0))) {
+ qerror_report(QERR_INVALID_PARAMETER_COMBINATION);
+ return -1;
+ }
+
+ bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL] = bps;
+ bs->io_limits.bps[BLOCK_IO_LIMIT_READ] = bps_rd;
+ bs->io_limits.bps[BLOCK_IO_LIMIT_WRITE] = bps_wr;
+ bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL] = iops;
+ bs->io_limits.iops[BLOCK_IO_LIMIT_READ] = iops_rd;
+ bs->io_limits.iops[BLOCK_IO_LIMIT_WRITE] = iops_wr;
+ bs->slice_time = BLOCK_IO_SLICE_TIME;
+
+ if (!bs->io_limits_enabled && bdrv_io_limits_enabled(bs)) {
+ bdrv_io_limits_enable(bs);
+ } else if (bs->io_limits_enabled && !bdrv_io_limits_enabled(bs)) {
+ bdrv_io_limits_disable(bs);
+ } else {
+ qemu_mod_timer(bs->block_timer, qemu_get_clock_ns(vm_clock));
+ }
+
+ return 0;
+}
+
int do_drive_del(Monitor *mon, const QDict *qdict, QObject **ret_data)
{
const char *id = qdict_get_str(qdict, "id");
diff --git a/blockdev.h b/blockdev.h
index 3587786..1b48a75 100644
--- a/blockdev.h
+++ b/blockdev.h
@@ -63,6 +63,8 @@ int do_block_set_passwd(Monitor *mon, const QDict *qdict, QObject **ret_data);
int do_change_block(Monitor *mon, const char *device,
const char *filename, const char *fmt);
int do_drive_del(Monitor *mon, const QDict *qdict, QObject **ret_data);
+int do_block_set_io_throttle(Monitor *mon,
+ const QDict *qdict, QObject **ret_data);
int do_snapshot_blkdev(Monitor *mon, const QDict *qdict, QObject **ret_data);
int do_block_resize(Monitor *mon, const QDict *qdict, QObject **ret_data);
diff --git a/hmp-commands.hx b/hmp-commands.hx
index 089c1ac..48f3c21 100644
--- a/hmp-commands.hx
+++ b/hmp-commands.hx
@@ -1207,6 +1207,21 @@ ETEXI
},
STEXI
+@item block_set_io_throttle @var{device} @var{bps} @var{bps_rd} @var{bps_wr} @var{iops} @var{iops_rd} @var{iops_wr}
+@findex block_set_io_throttle
+Change I/O throttle limits for a block drive to @var{bps} @var{bps_rd} @var{bps_wr} @var{iops} @var{iops_rd} @var{iops_wr}
+ETEXI
+
+ {
+ .name = "block_set_io_throttle",
+ .args_type = "device:B,bps:i?,bps_rd:i?,bps_wr:i?,iops:i?,iops_rd:i?,iops_wr:i?",
+ .params = "device [bps] [bps_rd] [bps_wr] [iops] [iops_rd] [iops_wr]",
+ .help = "change I/O throttle limits for a block drive",
+ .user_print = monitor_user_noop,
+ .mhandler.cmd_new = do_block_set_io_throttle,
+ },
+
+STEXI
@item block_passwd @var{device} @var{password}
@findex block_passwd
Set the encrypted device @var{device} password to @var{password}
diff --git a/hmp.c b/hmp.c
index 443d3a7..dfab7ad 100644
--- a/hmp.c
+++ b/hmp.c
@@ -216,6 +216,16 @@ void hmp_info_block(Monitor *mon)
info->value->inserted->ro,
info->value->inserted->drv,
info->value->inserted->encrypted);
+
+ monitor_printf(mon, " bps=%" PRId64 " bps_rd=%" PRId64
+ " bps_wr=%" PRId64 " iops=%" PRId64
+ " iops_rd=%" PRId64 " iops_wr=%" PRId64,
+ info->value->inserted->bps,
+ info->value->inserted->bps_rd,
+ info->value->inserted->bps_wr,
+ info->value->inserted->iops,
+ info->value->inserted->iops_rd,
+ info->value->inserted->iops_wr);
} else {
monitor_printf(mon, " [not inserted]");
}
diff --git a/qapi-schema.json b/qapi-schema.json
index cb1ba77..734076b 100644
--- a/qapi-schema.json
+++ b/qapi-schema.json
@@ -370,13 +370,27 @@
#
# @encrypted: true if the backing device is encrypted
#
+# @bps: #optional if total throughput limit in bytes per second is specified
+#
+# @bps_rd: #optional if read throughput limit in bytes per second is specified
+#
+# @bps_wr: #optional if write throughput limit in bytes per second is specified
+#
+# @iops: #optional if total I/O operations per second is specified
+#
+# @iops_rd: #optional if read I/O operations per second is specified
+#
+# @iops_wr: #optional if write I/O operations per second is specified
+#
# Since: 0.14.0
#
# Notes: This interface is only found in @BlockInfo.
##
{ 'type': 'BlockDeviceInfo',
'data': { 'file': 'str', 'ro': 'bool', 'drv': 'str',
- '*backing_file': 'str', 'encrypted': 'bool' } }
+ '*backing_file': 'str', 'encrypted': 'bool',
+ 'bps': 'int', 'bps_rd': 'int', 'bps_wr': 'int',
+ 'iops': 'int', 'iops_rd': 'int', 'iops_wr': 'int'} }
##
# @BlockDeviceIoStatus:
diff --git a/qerror.c b/qerror.c
index 4b48b39..807fb55 100644
--- a/qerror.c
+++ b/qerror.c
@@ -238,6 +238,10 @@ static const QErrorStringTable qerror_table[] = {
.error_fmt = QERR_QGA_COMMAND_FAILED,
.desc = "Guest agent command failed, error was '%(message)'",
},
+ {
+ .error_fmt = QERR_INVALID_PARAMETER_COMBINATION,
+ .desc = "Invalid paramter combination",
+ },
{}
};
diff --git a/qerror.h b/qerror.h
index d4bfcfd..777a36a 100644
--- a/qerror.h
+++ b/qerror.h
@@ -198,4 +198,7 @@ QError *qobject_to_qerror(const QObject *obj);
#define QERR_QGA_COMMAND_FAILED \
"{ 'class': 'QgaCommandFailed', 'data': { 'message': %s } }"
+#define QERR_INVALID_PARAMETER_COMBINATION \
+ "{ 'class': 'InvalidParameterCombination', 'data': {} }"
+
#endif /* QERROR_H */
diff --git a/qmp-commands.hx b/qmp-commands.hx
index 97975a5..cdc3c18 100644
--- a/qmp-commands.hx
+++ b/qmp-commands.hx
@@ -851,6 +851,44 @@ Example:
EQMP
{
+ .name = "block_set_io_throttle",
+ .args_type = "device:B,bps:i?,bps_rd:i?,bps_wr:i?,iops:i?,iops_rd:i?,iops_wr:i?",
+ .params = "device [bps] [bps_rd] [bps_wr] [iops] [iops_rd] [iops_wr]",
+ .help = "change I/O throttle limits for a block drive",
+ .user_print = monitor_user_noop,
+ .mhandler.cmd_new = do_block_set_io_throttle,
+ },
+
+SQMP
+block_set_io_throttle
+------------
+
+Change I/O throttle limits for a block drive.
+
+Arguments:
+
+- "device": device name (json-string)
+- "bps": total throughput limit in bytes per second(json-int, optional)
+- "bps_rd": read throughput limit in bytes per second(json-int, optional)
+- "bps_wr": read throughput limit in bytes per second(json-int, optional)
+- "iops": total I/O operations per second(json-int, optional)
+- "iops_rd": read I/O operations per second(json-int, optional)
+- "iops_wr": write I/O operations per second(json-int, optional)
+
+Example:
+
+-> { "execute": "block_set_io_throttle", "arguments": { "device": "virtio0",
+ "bps": "1000000",
+ "bps_rd": "0",
+ "bps_wr": "0",
+ "iops": "0",
+ "iops_rd": "0",
+ "iops_wr": "0" } }
+<- { "return": {} }
+
+EQMP
+
+ {
.name = "set_password",
.args_type = "protocol:s,password:s,connected:s?",
.params = "protocol password action-if-connected",
@@ -1152,6 +1190,13 @@ Each json-object contain the following:
"tftp", "vdi", "vmdk", "vpc", "vvfat"
- "backing_file": backing file name (json-string, optional)
- "encrypted": true if encrypted, false otherwise (json-bool)
+ - "bps": limit total bytes per second (json-int)
+ - "bps_rd": limit read bytes per second (json-int)
+ - "bps_wr": limit write bytes per second (json-int)
+ - "iops": limit total I/O operations per second (json-int)
+ - "iops_rd": limit read operations per second (json-int)
+ - "iops_wr": limit write operations per second (json-int)
+
- "io-status": I/O operation status, only present if the device supports it
and the VM is configured to stop on errors. It's always reset
to "ok" when the "cont" command is issued (json_string, optional)
@@ -1171,7 +1216,13 @@ Example:
"ro":false,
"drv":"qcow2",
"encrypted":false,
- "file":"disks/test.img"
+ "file":"disks/test.img",
+ "bps":1000000,
+ "bps_rd":0,
+ "bps_wr":0,
+ "iops":1000000,
+ "iops_rd":0,
+ "iops_wr":0,
},
"type":"unknown"
},
--
1.7.6
^ permalink raw reply related [flat|nested] 9+ messages in thread