From: Fabiano Rosas <farosas@suse.de>
To: Hao Xiang <hao.xiang@bytedance.com>,
quintela@redhat.com, peterx@redhat.com,
marcandre.lureau@redhat.com, bryan.zhang@bytedance.com,
qemu-devel@nongnu.org
Cc: Hao Xiang <hao.xiang@bytedance.com>
Subject: Re: [PATCH 13/16] migration/multifd: Add migration option set packet size.
Date: Mon, 30 Oct 2023 12:03:39 -0300 [thread overview]
Message-ID: <87bkcgnodw.fsf@suse.de> (raw)
In-Reply-To: <20231025193822.2813204-14-hao.xiang@bytedance.com>
Hao Xiang <hao.xiang@bytedance.com> writes:
> The current multifd packet size is 128 * 4kb. This change adds
> an option to set the packet size. Both sender and receiver needs
> to set the same packet size for things to work.
>
> Signed-off-by: Hao Xiang <hao.xiang@bytedance.com>
> ---
> migration/options.c | 34 ++++++++++++++++++++++++++++++++++
> migration/options.h | 1 +
> qapi/migration.json | 20 +++++++++++++++++---
> 3 files changed, 52 insertions(+), 3 deletions(-)
>
> diff --git a/migration/options.c b/migration/options.c
> index 9ee0ad5d89..6cb3d19470 100644
> --- a/migration/options.c
> +++ b/migration/options.c
> @@ -83,6 +83,12 @@
> */
> #define DEFAULT_MIGRATE_MULTIFD_NORMAL_PAGE_RATIO 101
>
> +/*
> + * Parameter for multifd packet size.
> + */
> +#define DEFAULT_MIGRATE_MULTIFD_PACKET_SIZE (512 * 1024)
> +#define MAX_MIGRATE_MULTIFD_PACKET_SIZE (1024 * 4 * 1024)
> +
> #define DEFINE_PROP_MIG_CAP(name, x) \
> DEFINE_PROP_BOOL(name, MigrationState, capabilities[x], false)
>
> @@ -183,6 +189,9 @@ Property migration_properties[] = {
> DEFINE_PROP_UINT8("multifd-normal-page-ratio", MigrationState,
> parameters.multifd_normal_page_ratio,
> DEFAULT_MIGRATE_MULTIFD_NORMAL_PAGE_RATIO),
> + DEFINE_PROP_SIZE("multifd-packet-size", MigrationState,
> + parameters.multifd_packet_size,
> + DEFAULT_MIGRATE_MULTIFD_PACKET_SIZE),
>
> /* Migration capabilities */
> DEFINE_PROP_MIG_CAP("x-xbzrle", MIGRATION_CAPABILITY_XBZRLE),
> @@ -822,6 +831,13 @@ uint8_t migrate_multifd_normal_page_ratio(void)
> return s->parameters.multifd_normal_page_ratio;
> }
>
> +uint64_t migrate_multifd_packet_size(void)
> +{
> + MigrationState *s = migrate_get_current();
> +
> + return s->parameters.multifd_packet_size;
> +}
> +
> MultiFDCompression migrate_multifd_compression(void)
> {
> MigrationState *s = migrate_get_current();
> @@ -958,6 +974,8 @@ MigrationParameters *qmp_query_migrate_parameters(Error **errp)
> params->x_checkpoint_delay = s->parameters.x_checkpoint_delay;
> params->has_block_incremental = true;
> params->block_incremental = s->parameters.block_incremental;
> + params->has_multifd_packet_size = true;
> + params->multifd_packet_size = s->parameters.multifd_packet_size;
> params->has_multifd_channels = true;
> params->multifd_channels = s->parameters.multifd_channels;
> params->has_multifd_compression = true;
> @@ -1016,6 +1034,7 @@ void migrate_params_init(MigrationParameters *params)
> params->has_downtime_limit = true;
> params->has_x_checkpoint_delay = true;
> params->has_block_incremental = true;
> + params->has_multifd_packet_size = true;
> params->has_multifd_channels = true;
> params->has_multifd_compression = true;
> params->has_multifd_zlib_level = true;
> @@ -1104,6 +1123,15 @@ bool migrate_params_check(MigrationParameters *params, Error **errp)
>
> /* x_checkpoint_delay is now always positive */
>
> + if (params->has_multifd_packet_size &&
> + ((params->multifd_packet_size < DEFAULT_MIGRATE_MULTIFD_PACKET_SIZE) ||
> + (params->multifd_packet_size > MAX_MIGRATE_MULTIFD_PACKET_SIZE))) {
> + error_setg(errp, QERR_INVALID_PARAMETER_VALUE,
> + "multifd_packet_size",
> + "a value between 524288 and 4194304");
> + return false;
> + }
> +
> if (params->has_multifd_channels && (params->multifd_channels < 1)) {
> error_setg(errp, QERR_INVALID_PARAMETER_VALUE,
> "multifd_channels",
> @@ -1281,6 +1309,9 @@ static void migrate_params_test_apply(MigrateSetParameters *params,
> if (params->has_block_incremental) {
> dest->block_incremental = params->block_incremental;
> }
> + if (params->has_multifd_packet_size) {
> + dest->multifd_packet_size = params->multifd_packet_size;
> + }
> if (params->has_multifd_channels) {
> dest->multifd_channels = params->multifd_channels;
> }
> @@ -1408,6 +1439,9 @@ static void migrate_params_apply(MigrateSetParameters *params, Error **errp)
> if (params->has_block_incremental) {
> s->parameters.block_incremental = params->block_incremental;
> }
> + if (params->has_multifd_packet_size) {
> + s->parameters.multifd_packet_size = params->multifd_packet_size;
> + }
> if (params->has_multifd_channels) {
> s->parameters.multifd_channels = params->multifd_channels;
> }
> diff --git a/migration/options.h b/migration/options.h
> index dafb09d6ea..1170971aef 100644
> --- a/migration/options.h
> +++ b/migration/options.h
> @@ -93,6 +93,7 @@ const char *migrate_tls_hostname(void);
> uint64_t migrate_xbzrle_cache_size(void);
> const char *migrate_multifd_dsa_accel(void);
> uint8_t migrate_multifd_normal_page_ratio(void);
> +uint64_t migrate_multifd_packet_size(void);
>
> /* parameters setters */
>
> diff --git a/qapi/migration.json b/qapi/migration.json
> index a667527081..a492b73060 100644
> --- a/qapi/migration.json
> +++ b/qapi/migration.json
> @@ -835,6 +835,10 @@
> # @multifd-normal-page-ratio: Test hook setting the normal page ratio.
> # (Since 8.1)
> #
> +# @multifd-packet-size: Packet size used to migrate data. This value
> +# needs to be a multiple of qemu_target_page_size(). The default
> +# value is (512 * 1024) (Since 8.0)
User doesn't know what qemu_target_page_size means. I'd also not have
them do arithmetic, specially since the option takes a single number.
We also need to enforce this in the code in the next patch.
> +#
> # Features:
> #
> # @unstable: Members @x-checkpoint-delay and @x-vcpu-dirty-limit-period
> @@ -859,7 +863,7 @@
> 'multifd-zlib-level', 'multifd-zstd-level',
> 'block-bitmap-mapping',
> { 'name': 'x-vcpu-dirty-limit-period', 'features': ['unstable'] },
> - 'vcpu-dirty-limit', 'multifd-normal-page-ratio'] }
> + 'vcpu-dirty-limit', 'multifd-normal-page-ratio', 'multifd-packet-size'] }
>
> ##
> # @MigrateSetParameters:
> @@ -1007,6 +1011,10 @@
> # @multifd-normal-page-ratio: Test hook setting the normal page ratio.
> # (Since 8.1)
> #
> +# @multifd-packet-size: Packet size used to migrate data. This value
> +# needs to be a multiple of qemu_target_page_size(). The default
> +# value is (512 * 1024) (Since 8.0)
> +#
> # Features:
> #
> # @unstable: Members @x-checkpoint-delay and @x-vcpu-dirty-limit-period
> @@ -1050,7 +1058,8 @@
> 'features': [ 'unstable' ] },
> '*vcpu-dirty-limit': 'uint64',
> '*multifd-dsa-accel': 'StrOrNull',
> - '*multifd-normal-page-ratio': 'uint8'} }
> + '*multifd-normal-page-ratio': 'uint8',
> + '*multifd-packet-size' : 'uint64'} }
>
> ##
> # @migrate-set-parameters:
> @@ -1218,6 +1227,10 @@
> # @multifd-normal-page-ratio: Test hook setting the normal page ratio.
> # (Since 8.1)
> #
> +# @multifd-packet-size: Packet size used to migrate data. This value
> +# needs to be a multiple of qemu_target_page_size(). The default
> +# value is (512 * 1024) (Since 8.0)
> +#
> # Features:
> #
> # @unstable: Members @x-checkpoint-delay and @x-vcpu-dirty-limit-period
> @@ -1258,7 +1271,8 @@
> 'features': [ 'unstable' ] },
> '*vcpu-dirty-limit': 'uint64',
> '*multifd-dsa-accel': 'str',
> - '*multifd-normal-page-ratio': 'uint8'} }
> + '*multifd-normal-page-ratio': 'uint8',
> + '*multifd-packet-size': 'uint64'} }
>
> ##
> # @query-migrate-parameters:
next prev parent reply other threads:[~2023-10-30 15:04 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-10-25 19:38 [PATCH 00/16] Use Intel DSA accelerator to offload zero page checking in multifd live migration Hao Xiang
2023-10-25 19:38 ` [PATCH 01/16] Cherry pick a set of patches that enables multifd zero page feature Hao Xiang
2023-10-27 12:30 ` Fabiano Rosas
2023-10-27 13:21 ` Peter Maydell
2023-10-28 1:13 ` [External] " Hao Xiang
2023-10-28 1:06 ` Hao Xiang
2023-10-30 13:58 ` Fabiano Rosas
2023-11-06 18:53 ` Hao Xiang
2023-10-25 19:38 ` [PATCH 02/16] meson: Introduce new instruction set enqcmd to the build system Hao Xiang
2023-10-25 19:38 ` [PATCH 03/16] util/dsa: Add dependency idxd Hao Xiang
2023-10-25 19:38 ` [PATCH 04/16] util/dsa: Implement DSA device start and stop logic Hao Xiang
2023-10-25 19:38 ` [PATCH 05/16] util/dsa: Implement DSA task enqueue and dequeue Hao Xiang
2023-10-25 19:38 ` [PATCH 06/16] util/dsa: Implement DSA task asynchronous completion thread model Hao Xiang
2023-10-25 19:38 ` [PATCH 07/16] util/dsa: Implement zero page checking in DSA task Hao Xiang
2023-10-25 19:38 ` [PATCH 08/16] util/dsa: Implement DSA task asynchronous submission and wait for completion Hao Xiang
2023-10-25 19:38 ` [PATCH 09/16] migration/multifd: Add new migration option for multifd DSA offloading Hao Xiang
2023-10-30 14:41 ` Fabiano Rosas
2023-11-06 21:58 ` [External] " Hao Xiang
2023-10-25 19:38 ` [PATCH 10/16] migration/multifd: Enable DSA offloading in multifd sender path Hao Xiang
2023-10-30 14:37 ` Fabiano Rosas
2023-10-25 19:38 ` [PATCH 11/16] migration/multifd: Add test hook to set normal page ratio Hao Xiang
2023-10-25 19:38 ` [PATCH 12/16] migration/multifd: Enable set normal page ratio test hook in multifd Hao Xiang
2023-10-25 19:38 ` [PATCH 13/16] migration/multifd: Add migration option set packet size Hao Xiang
2023-10-30 15:03 ` Fabiano Rosas [this message]
2023-10-25 19:38 ` [PATCH 14/16] migration/multifd: Enable set packet size migration option Hao Xiang
2023-10-25 19:38 ` [PATCH 15/16] util/dsa: Add unit test coverage for Intel DSA task submission and completion Hao Xiang
2023-10-25 19:38 ` [PATCH 16/16] migration/multifd: Add integration tests for multifd with Intel DSA offloading Hao Xiang
2023-10-30 15:26 ` Fabiano Rosas
2023-10-30 15:26 ` [PATCH 00/16] Use Intel DSA accelerator to offload zero page checking in multifd live migration Fabiano Rosas
2023-10-31 1:02 ` [External] " Hao Xiang
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=87bkcgnodw.fsf@suse.de \
--to=farosas@suse.de \
--cc=bryan.zhang@bytedance.com \
--cc=hao.xiang@bytedance.com \
--cc=marcandre.lureau@redhat.com \
--cc=peterx@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=quintela@redhat.com \
/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).