From: Avihai Horon <avihaih@nvidia.com>
To: Peter Xu <peterx@redhat.com>, qemu-devel@nongnu.org
Cc: "Juraj Marcin" <jmarcin@redhat.com>,
"Kirti Wankhede" <kwankhede@nvidia.com>,
"Maciej S . Szmigiero" <mail@maciej.szmigiero.name>,
"Daniel P . Berrangé" <berrange@redhat.com>,
"Joao Martins" <joao.m.martins@oracle.com>,
"Alex Williamson" <alex@shazbot.org>,
"Yishai Hadas" <yishaih@nvidia.com>,
"Fabiano Rosas" <farosas@suse.de>,
"Pranav Tyagi" <prtyagi@redhat.com>,
"Zhiyi Guo" <zhguo@redhat.com>,
"Markus Armbruster" <armbru@redhat.com>,
"Cédric Le Goater" <clg@redhat.com>
Subject: Re: [PATCH RFC 04/12] vfio/migration: Cache stop size in VFIOMigration
Date: Wed, 25 Mar 2026 16:15:45 +0200 [thread overview]
Message-ID: <894ad48a-6779-41f7-bf9f-08008c8e4bde@nvidia.com> (raw)
In-Reply-To: <20260319231302.123135-5-peterx@redhat.com>
On 3/20/2026 1:12 AM, Peter Xu wrote:
> External email: Use caution opening links or attachments
>
>
> Add a field to cache stop size. Note that there's an initial value change
> in vfio_save_setup for the stop size default, but it shouldn't matter if it
> is followed with a math of MIN() against VFIO_MIG_DEFAULT_DATA_BUFFER_SIZE.
>
> Signed-off-by: Peter Xu <peterx@redhat.com>
> ---
> hw/vfio/vfio-migration-internal.h | 1 +
> hw/vfio/migration.c | 43 +++++++++++++++++--------------
> 2 files changed, 24 insertions(+), 20 deletions(-)
>
> diff --git a/hw/vfio/vfio-migration-internal.h b/hw/vfio/vfio-migration-internal.h
> index 814fbd9eba..08df32c055 100644
> --- a/hw/vfio/vfio-migration-internal.h
> +++ b/hw/vfio/vfio-migration-internal.h
> @@ -47,6 +47,7 @@ typedef struct VFIOMigration {
> uint64_t mig_flags;
> uint64_t precopy_init_size;
> uint64_t precopy_dirty_size;
> + uint64_t stopcopy_size;
> bool multifd_transfer;
> VFIOMultifd *multifd;
> bool initial_data_sent;
> diff --git a/hw/vfio/migration.c b/hw/vfio/migration.c
> index 851ea783f3..827d3ded63 100644
> --- a/hw/vfio/migration.c
> +++ b/hw/vfio/migration.c
> @@ -41,6 +41,12 @@
> */
> #define VFIO_MIG_DEFAULT_DATA_BUFFER_SIZE (1 * MiB)
>
> +/*
> + * Migration size of VFIO devices can be as little as a few KBs or as big as
> + * many GBs. This value should be big enough to cover the worst case.
> + */
> +#define VFIO_MIG_STOP_COPY_SIZE (100 * GiB)
> +
> static unsigned long bytes_transferred;
>
> static const char *mig_state_to_str(enum vfio_device_mig_state state)
> @@ -314,8 +320,7 @@ static void vfio_migration_cleanup(VFIODevice *vbasedev)
> migration->data_fd = -1;
> }
>
> -static int vfio_query_stop_copy_size(VFIODevice *vbasedev,
> - uint64_t *stop_copy_size)
> +static int vfio_query_stop_copy_size(VFIODevice *vbasedev)
> {
> uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature) +
> sizeof(struct vfio_device_feature_mig_data_size),
> @@ -323,16 +328,22 @@ static int vfio_query_stop_copy_size(VFIODevice *vbasedev,
> struct vfio_device_feature *feature = (struct vfio_device_feature *)buf;
> struct vfio_device_feature_mig_data_size *mig_data_size =
> (struct vfio_device_feature_mig_data_size *)feature->data;
> + VFIOMigration *migration = vbasedev->migration;
>
> feature->argsz = sizeof(buf);
> feature->flags =
> VFIO_DEVICE_FEATURE_GET | VFIO_DEVICE_FEATURE_MIG_DATA_SIZE;
>
> if (ioctl(vbasedev->fd, VFIO_DEVICE_FEATURE, feature)) {
> + /*
> + * If getting pending migration size fails, VFIO_MIG_STOP_COPY_SIZE
> + * is reported so downtime limit won't be violated.
> + */
> + migration->stopcopy_size = VFIO_MIG_STOP_COPY_SIZE;
> return -errno;
> }
>
> - *stop_copy_size = mig_data_size->stop_copy_length;
> + migration->stopcopy_size = mig_data_size->stop_copy_length;
>
> return 0;
> }
> @@ -415,6 +426,9 @@ static void vfio_update_estimated_pending_data(VFIOMigration *migration,
> return;
> }
>
> + /* The total size remaining requires separate accounting */
> + migration->stopcopy_size -= data_size;
stopcopy_size is also an estimation, so I think it's safer to have:
migration->stopcopy_size -= MIN(migration->stopcopy_size, data_size);
Thanks.
> +
> if (migration->precopy_init_size) {
> uint64_t init_size = MIN(migration->precopy_init_size, data_size);
>
> @@ -469,7 +483,6 @@ static int vfio_save_setup(QEMUFile *f, void *opaque, Error **errp)
> {
> VFIODevice *vbasedev = opaque;
> VFIOMigration *migration = vbasedev->migration;
> - uint64_t stop_copy_size = VFIO_MIG_DEFAULT_DATA_BUFFER_SIZE;
> int ret;
>
> if (!vfio_multifd_setup(vbasedev, false, errp)) {
> @@ -478,9 +491,9 @@ static int vfio_save_setup(QEMUFile *f, void *opaque, Error **errp)
>
> qemu_put_be64(f, VFIO_MIG_FLAG_DEV_SETUP_STATE);
>
> - vfio_query_stop_copy_size(vbasedev, &stop_copy_size);
> + vfio_query_stop_copy_size(vbasedev);
> migration->data_buffer_size = MIN(VFIO_MIG_DEFAULT_DATA_BUFFER_SIZE,
> - stop_copy_size);
> + migration->stopcopy_size);
> migration->data_buffer = g_try_malloc0(migration->data_buffer_size);
> if (!migration->data_buffer) {
> error_setg(errp, "%s: Failed to allocate migration data buffer",
> @@ -576,32 +589,22 @@ static void vfio_state_pending_estimate(void *opaque, uint64_t *must_precopy,
> migration->precopy_dirty_size);
> }
>
> -/*
> - * Migration size of VFIO devices can be as little as a few KBs or as big as
> - * many GBs. This value should be big enough to cover the worst case.
> - */
> -#define VFIO_MIG_STOP_COPY_SIZE (100 * GiB)
> -
> static void vfio_state_pending_exact(void *opaque, uint64_t *must_precopy,
> uint64_t *can_postcopy)
> {
> VFIODevice *vbasedev = opaque;
> VFIOMigration *migration = vbasedev->migration;
> - uint64_t stop_copy_size = VFIO_MIG_STOP_COPY_SIZE;
>
> - /*
> - * If getting pending migration size fails, VFIO_MIG_STOP_COPY_SIZE is
> - * reported so downtime limit won't be violated.
> - */
> - vfio_query_stop_copy_size(vbasedev, &stop_copy_size);
> - *must_precopy += stop_copy_size;
> + vfio_query_stop_copy_size(vbasedev);
> + *must_precopy += migration->stopcopy_size;
>
> if (vfio_device_state_is_precopy(vbasedev)) {
> vfio_query_precopy_size(migration);
> }
>
> trace_vfio_state_pending_exact(vbasedev->name, *must_precopy, *can_postcopy,
> - stop_copy_size, migration->precopy_init_size,
> + migration->stopcopy_size,
> + migration->precopy_init_size,
> migration->precopy_dirty_size);
> }
>
> --
> 2.50.1
>
next prev parent reply other threads:[~2026-03-25 14:16 UTC|newest]
Thread overview: 67+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-03-19 23:12 [PATCH RFC 00/12] migration/vfio: Fix a few issues on API misuse or statistic reports Peter Xu
2026-03-19 23:12 ` [PATCH RFC 01/12] migration: Fix low possibility downtime violation Peter Xu
2026-03-20 12:26 ` Prasad Pandit
2026-03-27 14:35 ` Juraj Marcin
2026-03-30 11:52 ` Prasad Pandit
2026-03-31 12:49 ` Juraj Marcin
2026-04-06 7:21 ` Prasad Pandit
2026-04-01 19:11 ` Peter Xu
2026-03-27 15:05 ` Juraj Marcin
2026-03-19 23:12 ` [PATCH RFC 02/12] migration/qapi: Rename MigrationStats to MigrationRAMStats Peter Xu
2026-03-19 23:26 ` Peter Xu
2026-03-20 6:54 ` Markus Armbruster
2026-04-01 19:38 ` Peter Xu
2026-04-01 19:47 ` Peter Xu
2026-03-19 23:12 ` [PATCH RFC 03/12] vfio/migration: Throttle vfio_save_block() on data size to read Peter Xu
2026-03-25 14:10 ` Avihai Horon
2026-04-01 20:36 ` Peter Xu
2026-04-06 11:21 ` Avihai Horon
2026-04-07 15:18 ` Peter Xu
2026-03-19 23:12 ` [PATCH RFC 04/12] vfio/migration: Cache stop size in VFIOMigration Peter Xu
2026-03-25 14:15 ` Avihai Horon [this message]
2026-04-01 20:41 ` Peter Xu
2026-04-06 11:28 ` Avihai Horon
2026-03-19 23:12 ` [PATCH RFC 05/12] migration/treewide: Merge @state_pending_{exact|estimate} APIs Peter Xu
2026-03-24 10:35 ` Prasad Pandit
2026-04-01 20:53 ` Peter Xu
2026-03-25 15:20 ` Avihai Horon
2026-04-01 21:22 ` Peter Xu
2026-04-06 11:54 ` Avihai Horon
2026-03-27 15:17 ` Juraj Marcin
2026-03-19 23:12 ` [PATCH RFC 06/12] migration: Use the new save_query_pending() API directly Peter Xu
2026-03-24 9:35 ` Prasad Pandit
2026-03-27 15:24 ` Juraj Marcin
2026-04-01 22:28 ` Peter Xu
2026-03-19 23:12 ` [PATCH RFC 07/12] migration: Introduce stopcopy_bytes in save_query_pending() Peter Xu
2026-03-24 11:05 ` Prasad Pandit
2026-03-25 16:54 ` Avihai Horon
2026-04-02 14:09 ` Peter Xu
2026-04-06 12:20 ` Avihai Horon
2026-04-07 15:30 ` Peter Xu
2026-03-27 16:43 ` Juraj Marcin
2026-04-02 15:16 ` Peter Xu
2026-04-07 15:19 ` Juraj Marcin
2026-04-07 15:32 ` Peter Xu
2026-03-19 23:12 ` [PATCH RFC 08/12] vfio/migration: Fix incorrect reporting for VFIO pending data Peter Xu
2026-03-25 17:32 ` Avihai Horon
2026-04-02 15:28 ` Peter Xu
2026-04-02 15:55 ` Peter Xu
2026-04-06 12:34 ` Avihai Horon
2026-04-07 15:45 ` Peter Xu
2026-03-19 23:12 ` [PATCH RFC 09/12] migration: Make iteration counter out of RAM Peter Xu
2026-03-20 6:12 ` Yong Huang
2026-03-20 9:49 ` Prasad Pandit
2026-04-02 15:35 ` Peter Xu
2026-03-27 16:49 ` Juraj Marcin
2026-04-02 15:42 ` Peter Xu
2026-03-19 23:13 ` [PATCH RFC 10/12] migration: Introduce a helper to return switchover bw estimate Peter Xu
2026-03-23 10:26 ` Prasad Pandit
2026-03-27 17:07 ` Juraj Marcin
2026-04-07 17:27 ` Peter Xu
2026-04-08 14:33 ` Juraj Marcin
2026-03-19 23:13 ` [PATCH RFC 11/12] migration: Calculate expected downtime on demand Peter Xu
2026-03-27 17:17 ` Juraj Marcin
2026-04-07 17:33 ` Peter Xu
2026-03-19 23:13 ` [PATCH RFC 12/12] migration: Fix calculation of expected_downtime to take VFIO info Peter Xu
2026-03-23 12:05 ` Prasad Pandit
2026-04-07 17:40 ` Peter Xu
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=894ad48a-6779-41f7-bf9f-08008c8e4bde@nvidia.com \
--to=avihaih@nvidia.com \
--cc=alex@shazbot.org \
--cc=armbru@redhat.com \
--cc=berrange@redhat.com \
--cc=clg@redhat.com \
--cc=farosas@suse.de \
--cc=jmarcin@redhat.com \
--cc=joao.m.martins@oracle.com \
--cc=kwankhede@nvidia.com \
--cc=mail@maciej.szmigiero.name \
--cc=peterx@redhat.com \
--cc=prtyagi@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=yishaih@nvidia.com \
--cc=zhguo@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.