on
> and the new representation as members of MigrationParameters.
>
> Signed-off-by: Fabiano Rosas
Acked-by: Peter Xu
--
Peter Xu
gt; is added.
>
> For this to work, the has_* fields of s->parameters need to be already
> set beforehand, so move migrate_mark_all_params_present() to the init
> routine.
>
> Signed-off-by: Fabiano Rosas
Reviewed-by: Peter Xu
--
Peter Xu
On Mon, Jun 30, 2025 at 04:58:58PM -0300, Fabiano Rosas wrote:
> The QERR_INVALID_PARAMETER_VALUE macro is documented as not to be used
> in new code. Remove the usage from migration/options.c.
>
> Signed-off-by: Fabiano Rosas
Reviewed-by: Peter Xu
--
Peter Xu
bject (s->parameters).
>
> Add a separate boolean to track the status of the block_bitmap_mapping
> option.
>
> No functional change intended.
>
> (this was verified to not regress iotest 300, which is the test that
> 3cba22c9ad refers to)
>
> CC: Kevin Wolf
> Signed-off-by: Fabiano Rosas
Acked-by: Peter Xu
--
Peter Xu
e
> completion.
>
> Documenting this enum only serves to duplicate documentation between
> MigrationParameter and MigrationParameters.
>
> Add an exception to QAPIs pragma.json and stop documenting it.
>
> Signed-off-by: Fabiano Rosas
Acked-by: Peter Xu
--
Peter Xu
ationParameters were replaced
> with the ones from MigrateSetParameters which was more complete.
>
> Signed-off-by: Fabiano Rosas
Acked-by: Peter Xu
--
Peter Xu
+*tls_opt = opt;
Does tls_opt ever change? I wonder if this line is not needed, instead
tls_opt_to_str() can take an "StrOrNull *opt" directly.
> >>> +}
> >>
> >> I'd prefer something like
> >>
> >>if (!opt || opt->type == QTYPE_QSTRING) {
> >>return;
> >>}
> >>qobject_unref(opt->u.n);
> >>opt->type = QTYPE_QSTRING;
> >>opt->u.s = g_strdup("");
> >>*tls_opt = opt;
> >>
> >> But this is clearly a matter of taste.
>
> This is better indeed. I was moving back-and-forth between
> implementations and the code ended up a bit weird. Thanks!
>
--
Peter Xu
o->get) { <-- [1]
return;
}
name = g_strdup_printf("legacy-%s", prop->name);
object_class_property_add(OBJECT_CLASS(dc), name, "str",
prop->info->print ? qdev_get_legacy_property : prop->info->get,
NULL, NULL, (Property *)prop);
}
When with no get(), it seems it'll be wrongly treated as legacy property..
which further means whoever tries to get() on the property will invoke
qdev_get_legacy_property(), and likely crash on accessing info->print()..
The other issue is legacy property doesn't look like to provide a setter
function.. as it's passing NULL to object_class_property_add(set=XXX).
Likely we'll need to provide get() if without changing qdev code.
>
> > +
> > bool migrate_auto_converge(void)
> > {
> > MigrationState *s = migrate_get_current();
>
--
Peter Xu
On Thu, Jul 03, 2025 at 11:21:38PM +0200, Cédric Le Goater wrote:
> On 7/3/25 21:45, Peter Xu wrote:
> > On Wed, Jul 02, 2025 at 03:41:08PM -0400, Steven Sistare wrote:
> > > The irq producer is not closed, but it is detached from the kvm consumer.
> > > It's event
ration.c
> > +++ b/migration/migration.c
> > @@ -4050,6 +4050,7 @@ static void migration_instance_finalize(Object *obj)
> > {
> > MigrationState *ms = MIGRATION_OBJ(obj);
> >
> > +
> > qapi_free_BitmapMigrationNodeAliasList(ms->parameters.block_bitmap_mapping);
> > qemu_mutex_destroy
commit message, either when repost, or when
merge, would be nice, on explaning irq won't get lost.
Thanks,
--
Peter Xu
misc.h"
> > #include "migration/options.h"
> > #include "migration/qemu-file.h"
> > #include "migration/savevm.h"
> > #include "migration/vmstate.h"
> > +#include "system/kvm.h"
> > #include "system/runstate.h"
> > #include "trace.h"
> > @@ -264,3 +266,9 @@ bool cpr_incoming_needed(void *opaque)
> > MigMode mode = migrate_mode();
> > return mode == MIG_MODE_CPR_TRANSFER;
> > }
> > +
> > +void cpr_kvm_close(void)
> > +{
> > +kvm_close();
> > +vfio_kvm_device_close();
> > +}
> > diff --git a/migration/migration.c b/migration/migration.c
> > index 4098870..8f23cff 100644
> > --- a/migration/migration.c
> > +++ b/migration/migration.c
> > @@ -337,6 +337,7 @@ void migration_object_init(void)
> > ram_mig_init();
> > dirty_bitmap_mig_init();
> > +cpr_transfer_init();
> > /* Initialize cpu throttle timers */
> > cpu_throttle_init();
> > diff --git a/hw/vfio/meson.build b/hw/vfio/meson.build
> > index 73d29f9..98134a7 100644
> > --- a/hw/vfio/meson.build
> > +++ b/hw/vfio/meson.build
> > @@ -17,6 +17,8 @@ vfio_ss.add(when: 'CONFIG_VFIO_IGD', if_true:
> > files('igd.c'))
> > specific_ss.add_all(when: 'CONFIG_VFIO', if_true: vfio_ss)
> > +system_ss.add(when: 'CONFIG_VFIO', if_false: files('vfio-stubs.c'))
> > +
> > system_ss.add(when: 'CONFIG_VFIO_XGMAC', if_true:
> > files('calxeda-xgmac.c'))
> > system_ss.add(when: 'CONFIG_VFIO_AMD_XGBE', if_true: files('amd-xgbe.c'))
> > system_ss.add(when: 'CONFIG_VFIO', if_true: files(
--
Peter Xu
ot;precopy" keyword is retained.
>
> [1]: https://lore.kernel.org/all/20250613140801.474264-7-pet...@redhat.com/
>
> Signed-off-by: Juraj Marcin
> ---
> This patch is based on the following series:
> https://lore.kernel.org/all/20250613140801.474264-1-pet...@redhat.com/
queued, thanks.
--
Peter Xu
> >const char *suffix)
> > {
> > Error *err;
> > int saved_errno = errno;
> >
> > if (errp == NULL) {
> > return;
> > }
> > assert(*errp == NULL);
> > ...
> > }
> >
> > So here we can keep local_err but use error_propagate().
>
> Please correct me if I am wrong, as far as I have checked,
> qemu_loadvm_state() is called at 3 places
> - load_snapshot
> - qmp_xen_load_devices_state
> - process_incoming_migration_co
> and in all these functions, Error **errp is passed.
> I did not find a function that passes NULL.
> Is it still required to declare and pass a local_err object?
In that case (this should include all the callers of the three call sites
never passing in NULL) we're almost asserting errp!=NULL, then it should be
OK.
Thanks,
--
Peter Xu
queued.
On Fri, Jun 13, 2025 at 10:12:03AM -0400, Peter Xu wrote:
> This series is based on the other series I posted here:
queued.
--
Peter Xu
On Wed, Jun 25, 2025 at 01:13:48PM +0200, Juraj Marcin wrote:
> Sure, I will send it today.
Thanks. In case it helps, can fetch head here:
https://gitlab.com/peterx/qemu/-/tree/migration-staging
--
Peter Xu
ading, &vmstate_versioned, &obj, 2, NULL);
> g_assert_cmpint(obj.a, ==, 10);
> g_assert_cmpint(obj.b, ==, 20);
> g_assert_cmpint(obj.c, ==, 30);
> @@ -480,7 +480,7 @@ static void test_load_noskip(void)
>
> QEMUFile *loading = open_test_file(false);
> TestStruct obj = { .skip_c_e = false };
> -vmstate_load_state(loading, &vmstate_skipping, &obj, 2);
> +vmstate_load_state(loading, &vmstate_skipping, &obj, 2, NULL);
> g_assert(!qemu_file_get_error(loading));
> g_assert_cmpint(obj.a, ==, 10);
> g_assert_cmpint(obj.b, ==, 20);
> @@ -504,7 +504,7 @@ static void test_load_skip(void)
>
> QEMUFile *loading = open_test_file(false);
> TestStruct obj = { .skip_c_e = true, .c = 300, .e = 500 };
> -vmstate_load_state(loading, &vmstate_skipping, &obj, 2);
> +vmstate_load_state(loading, &vmstate_skipping, &obj, 2, NULL);
> g_assert(!qemu_file_get_error(loading));
> g_assert_cmpint(obj.a, ==, 10);
> g_assert_cmpint(obj.b, ==, 20);
> @@ -773,7 +773,7 @@ static void test_load_q(void)
> TestQtailq tgt;
>
> QTAILQ_INIT(&tgt.q);
> -vmstate_load_state(fload, &vmstate_q, &tgt, 1);
> +vmstate_load_state(fload, &vmstate_q, &tgt, 1, NULL);
> char eof = qemu_get_byte(fload);
> g_assert(!qemu_file_get_error(fload));
> g_assert_cmpint(tgt.i16, ==, obj_q.i16);
> @@ -1127,7 +1127,7 @@ static void test_gtree_load_domain(void)
>
> fload = open_test_file(false);
>
> -vmstate_load_state(fload, &vmstate_domain, dest_domain, 1);
> +vmstate_load_state(fload, &vmstate_domain, dest_domain, 1, NULL);
> eof = qemu_get_byte(fload);
> g_assert(!qemu_file_get_error(fload));
> g_assert_cmpint(orig_domain->id, ==, dest_domain->id);
> @@ -1241,7 +1241,7 @@ static void test_gtree_load_iommu(void)
> qemu_fclose(fsave);
>
> fload = open_test_file(false);
> -vmstate_load_state(fload, &vmstate_iommu, dest_iommu, 1);
> +vmstate_load_state(fload, &vmstate_iommu, dest_iommu, 1, NULL);
> eof = qemu_get_byte(fload);
> g_assert(!qemu_file_get_error(fload));
> g_assert_cmpint(orig_iommu->id, ==, dest_iommu->id);
> @@ -1376,7 +1376,7 @@ static void test_load_qlist(void)
> qemu_fclose(fsave);
>
> fload = open_test_file(false);
> -vmstate_load_state(fload, &vmstate_container, dest_container, 1);
> +vmstate_load_state(fload, &vmstate_container, dest_container, 1, NULL);
> eof = qemu_get_byte(fload);
> g_assert(!qemu_file_get_error(fload));
> g_assert_cmpint(eof, ==, QEMU_VM_EOF);
>
> --
> 2.49.0
>
--
Peter Xu
; -error_report("%s: incoming version_id %d is too old "
> - "for local minimum version_id %d",
> - vmsd->name, version_id, vmsd->minimum_version_id);
> +error_setg(errp, "%s: incoming version_id %d is too old "
> + "for local minimum version_id %d",
> + vmsd->name, version_id, vmsd->minimum_version_id);
> trace_vmstate_load_state_end(vmsd->name, "too old", -EINVAL);
> return -EINVAL;
> }
> @@ -213,15 +213,15 @@ int vmstate_load_state(QEMUFile *f, const
> VMStateDescription *vmsd,
> }
> if (ret < 0) {
> qemu_file_set_error(f, ret);
> -error_report("Failed to load %s:%s", vmsd->name,
> - field->name);
> +error_setg(errp, "Failed to load %s:%s", vmsd->name,
> + field->name);
> trace_vmstate_load_field_error(field->name, ret);
> return ret;
> }
> }
> } else if (field->flags & VMS_MUST_EXIST) {
> -error_report("Input validation failed: %s/%s",
> - vmsd->name, field->name);
> +error_setg(errp, "Input validation failed: %s/%s",
> + vmsd->name, field->name);
> return -1;
> }
> field++;
>
> --
> 2.49.0
>
--
Peter Xu
) until
> docs/devel/migration/main.rst:511
> - A ``save_live_complete_precopy`` function that must transmit the
Good catch. If I prepare this fixup to be squashed, can I get your R-b?
>From 58147b11276fa193c25f35e63f41a8e34d444dd9 Mon Sep 17 00:00:00 2001
From: Peter Xu
D
timer,
and the display is refreshed immediately if the timer is expired.
Signed-off-by: Juraj Marcin
Reviewed-by: Marc-André Lureau
Reviewed-by: Peter Xu
Reviewed-by: Daniel P. Berrangé
Link: https://lore.kernel.org/r/20250521151616.3951178-1-jmar...@redhat.com
Signed-off-by: Peter Xu
---
ui
d.
Tested-by: Alexey Kardashevskiy
Reviewed-by: Alexey Kardashevskiy
Reviewed-by: Pankaj Gupta
Signed-off-by: Chenyi Qiang
Link: https://lore.kernel.org/r/20250612082747.51539-5-chenyi.qi...@intel.com
[peterx: squash fixup from Chenyi to fix builds]
Signed-off-by: Peter Xu
---
M
om
[peterx: null-initialize opts_target, per Steve]
Signed-off-by: Peter Xu
---
tests/qtest/migration/cpr-tests.c | 14 --
1 file changed, 12 insertions(+), 2 deletions(-)
diff --git a/tests/qtest/migration/cpr-tests.c
b/tests/qtest/migration/cpr-tests.c
index 5536e14610..5e764a6787 10
makes
it cleaner.
Reviewed-by: David Hildenbrand
Reviewed-by: Pankaj Gupta
Reviewed-by: Xiaoyao Li
Signed-off-by: Chenyi Qiang
Link: https://lore.kernel.org/r/20250612082747.51539-4-chenyi.qi...@intel.com
Signed-off-by: Peter Xu
---
include/system/memory.h | 74
The following changes since commit 43ba160cb4bbb193560eb0d2d7decc4b5fc599fe:
Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging
(2025-06-21 10:00:46 -0400)
are available in the Git repository at:
https://gitlab.com/peterx/qemu.git tags/migration-staging-pull-request
f
: Peter Xu
---
include/system/memory.h | 6 +++---
hw/virtio/virtio-mem.c | 30 +-
system/memory.c | 10 +++---
3 files changed, 27 insertions(+), 19 deletions(-)
diff --git a/include/system/memory.h b/include/system/memory.h
index da97753e28..60983d4977
org/r/20250612082747.51539-6-chenyi.qi...@intel.com
Signed-off-by: Peter Xu
---
include/system/ramblock.h | 1 +
accel/kvm/kvm-all.c | 9 +
system/physmem.c | 23 +--
3 files changed, 31 insertions(+), 2 deletions(-)
diff --git a/include/system/rambloc
ram block writeback)
Link: https://lore.kernel.org/r/1cc43f59-336f-4a12-84ad-db89e0a17...@akamai.com
Signed-off-by: Peter Xu
---
migration/ram.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/migration/ram.c b/migration/ram.c
index fd8d83b63c..024d8b3d03 100644
--- a/migra
: Alexey Kardashevskiy
Reviewed-by: Pankaj Gupta
Reviewed-by: David Hildenbrand
Reviewed-by: Zhao Liu
Reviewed-by: Xiaoyao Li
Signed-off-by: Chenyi Qiang
Link: https://lore.kernel.org/r/20250612082747.51539-2-chenyi.qi...@intel.com
Signed-off-by: Peter Xu
---
include/system/memory.h | 30
created externally and passed via fd.
Reviewed-by: Jason J. Herne
Reviewed-by: Steve Sistare
Signed-off-by: Jaehoon Kim
Link: https://lore.kernel.org/r/20250611205610.147008-3-jh...@linux.ibm.com
Signed-off-by: Peter Xu
---
migration/cpr-transfer.c | 7 +--
1 file changed, 5 insertions(+), 2
qmp/ to include/qobject/")
Signed-off-by: Fabiano Rosas
Message-id: 20200310152141.13959-1-peter.mayd...@linaro.org
Reviewed-by: Markus Armbruster
Link: https://lore.kernel.org/r/20250523123023.19284-1-faro...@suse.de
Signed-off-by: Peter Xu
---
tests/qtest/migra
y: Fabiano Rosas
Link: https://lore.kernel.org/r/20250514115827.3216082-1-yanfei...@bytedance.com
[peterx: add comments]
Signed-off-by: Peter Xu
---
migration/ram.c | 22 +++---
1 file changed, 15 insertions(+), 7 deletions(-)
diff --git a/migration/ram.c b/migration/ram.c
index
On Mon, Jun 23, 2025 at 02:51:46PM +0200, Igor Mammedov wrote:
> On Fri, 20 Jun 2025 12:53:06 -0400
> Peter Xu wrote:
>
> > On Fri, Jun 20, 2025 at 05:14:16PM +0200, Igor Mammedov wrote:
> > > This patch brings back Jan's idea [1] of BQL-free IO access,
> >
On Mon, Jun 23, 2025 at 05:43:06AM +, CLEMENT MATHIEU--DRIF wrote:
> Hi Peter
>
> On 20/06/2025 4:35 pm, Peter Xu wrote:
> > Caution: External email. Do not open attachments or click links, unless
> > this email comes from a known sender and you know the content is safe.
> +!(read && mr->lockless_ro_io == true)) {
> bql_lock();
> release_lock = true;
> }
> @@ -2935,7 +2936,7 @@ static MemTxResult
> flatview_write_continue_step(MemTxAttrs attrs,
> if (!memory_access_is_direct(mr, true, attrs)) {
> uint64_t val;
> MemTxResult result;
> -bool release_lock = prepare_mmio_access(mr);
> +bool release_lock = prepare_mmio_access(mr, false);
>
> *l = memory_access_size(mr, *l, mr_addr);
> /*
> @@ -3032,7 +3033,7 @@ static MemTxResult
> flatview_read_continue_step(MemTxAttrs attrs, uint8_t *buf,
> /* I/O case */
> uint64_t val;
> MemTxResult result;
> -bool release_lock = prepare_mmio_access(mr);
> +bool release_lock = prepare_mmio_access(mr, true);
>
> *l = memory_access_size(mr, *l, mr_addr);
> result = memory_region_dispatch_read(mr, mr_addr, &val,
> size_memop(*l),
> --
> 2.43.5
>
--
Peter Xu
issions in IOMMUAccessFlags
>
> include/exec/memattrs.h | 3 +++
> include/hw/pci/pci.h| 9 +
> include/system/memory.h | 23 +++++++++--
> 3 files changed, 33 insertions(+), 2 deletions(-)
>
> --
> 2.49.0
>
--
Peter Xu
o 6 failures (and you can ignore the rust warnings):
https://gitlab.com/peterx/qemu/-/pipelines/1878653850
WASM complains a bit more than others, but worth double check from your
side too.
Thanks,
--
Peter Xu
est with kernel-doc format.
If you want, you can provide "git --fixup" appended to the reply (one fixup
for each patch that needs fixing) to avoid a full repost.
Thanks,
--
Peter Xu
it's just that it'll be a smaller portion to be shared.
Other than that, even if I don't know how to test this.. I read the patches
today and they look all good. The duplication is a pure question I have
above, but even if so it can also be done on top.
I do plan to pick this up. Paolo/David, any comments before I do?
Thanks,
--
Peter Xu
k_writeback(rb);
> +if (memory_region_is_nonvolatile(rb->mr)) {
> +qemu_ram_block_writeback(rb);
> + }
> }
>
> xbzrle_load_cleanup();
> --
> 2.40.1
>
--
Peter Xu
lock_writeback(rb);
> +}
> }
>
> xbzrle_load_cleanup();
> --
> 2.40.1
>
>
>
>
--
Peter Xu
ayer.
Note that we need a stub for mark_postcopy_blocktime_begin() for Windows
builds.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/migration.h| 2 +-
migration/postcopy-ram.h | 2 ++
migration/migration.c| 25 +++---
migration/postcopy-ram.c | 56 ++
On Mon, Jun 09, 2025 at 06:36:07PM -0400, Peter Xu wrote:
> Add the latency distribution too for blocktime, using order-of-two buckets.
> It accounts for all the faults, from either vCPU or non-vCPU threads. With
> prior rework, it's very easy to achieve by adding an array to accoun
Add a field to count how many remote faults one vCPU has taken. So far
it's still not used, but will be soon.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 5 +
1 file changed, 5 insertions(+)
diff --git a/migration/postcopy-ram.c b/migration/pos
Now with 64bits, the offseting using start_time is not needed anymore,
because the array can always remember the whole timestamp.
Then drop the unused parameter in get_low_time_offset() altogether.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 10
vCPUs especially after this rewrite. If add preempt mode into
picture, I think one should enable all three features for postcopy by
default at some point.
Comments welcomed, thanks.
Peter Xu (14):
migration: Add option to set postcopy-blocktime
migration/postcopy: Push blocktime start/end int
with this optimization:
Before: 268.00us (+-1.87%)
After: 232.67us (+-2.01%)
The test was done with a 16GB VM with 80 vCPUs, running a workload that
busy random writes to 13GB memory.
Cc: Yanfei Xu
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/ram.c | 4 +++-
1 file
Add a global property to allow enabling postcopy-blocktime feature.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/options.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/migration/options.c b/migration/options.c
index 162c72cda4..4e923a2e07 100644
--- a/migration
.
One tracepoint is added when walking all the fault entries.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 259 ---
migration/trace-events | 5 +-
2 files changed, 216 insertions(+), 48 deletions(-)
diff --git a/migration/p
;t created instead this time; the old
"if" trick isn't needed when we're sure it will only happen once now.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 15 ++-
1 file changed, 10 insertions(+), 5 deletions(-)
diff --git
Now with the mutex protection it's not needed anymore.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 23 ++-
1 file changed, 10 insertions(+), 13 deletions(-)
diff --git a/migration/postcopy-ram.c b/migration/postcopy-ram.c
e variable at all: since nobody is
touching smp_cpus_down except ourselves, we can safely do the calculation
at the end before decrementing smp_cpus_down.
Hopefully this makes the logic easier to read, side benefit is we drop one
temp var.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
cktime_start.
- Rename vcpu_blocktime to vcpu_blocktime_total.
- Touch up the trace-events to not dump blocktime ctx pointer
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 50
migration/trace-events | 4 ++--
2 files chan
ed".
Cc: Markus Armbruster
Cc: Dr. David Alan Gilbert
Reviewed-by: Fabiano Rosas
Tested-by: Mario Casquero
Signed-off-by: Peter Xu
---
qapi/migration.json | 12 -
migration/migration-hmp-cmds.c| 5 +++
migration/postcopy-ram.c | 64 ++
get_vcpu_blocktime_list() so we don't need to walk the list twice,
meanwhile add the entry checks in qtests for all postcopy tests.
Cc: Markus Armbruster
Cc: Dr. David Alan Gilbert
Reviewed-by: Fabiano Rosas
Tested-by: Mario Casquero
Signed-off-by: Peter Xu
---
qapi/migration
access a guest page that was missing.
These kind of faults are not accounted by blocktime so far.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 68 ++--
migration/trace-events | 3 +-
2 files changed, 59 insertions(+), 12 d
With 64-bit fields, it is trivial. The caution is when exposing any values
in QMP, it was still declared with milliseconds (ms). Hence it's needed to
do the convertion when exporting the values to existing QMP queries.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/pos
sec -8 sec ]: 0
[8 sec - 16 sec ]: 0
Cc: Markus Armbruster
Acked-by: Dr. David Alan Gilbert
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
qapi/migration.json | 12 ++-
migration/migration-hmp-cmds.c| 32 +++
It's not possible to happen in bg-snapshot case.
Reviewed-by: Juraj Marcin
Signed-off-by: Peter Xu
---
migration/migration.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/migration/migration.c b/migration/migration.c
index 4098870bce..e33e39ac74 100644
ses (precopy v.s. postcopy).
(4) Trivial touch up on threshold_size comparision
Which changes:
"(!pending_size || pending_size < s->threshold_size)"
into:
"(pending_size <= s->threshold_size)"
Reviewed-by: Juraj Marcin
Signed-off-by: Peter
welcomed, thanks.
Peter Xu (11):
migration/hmp: Reorg "info migrate" once more
migration/hmp: Fix postcopy-blocktime per-vCPU results
migration/docs: Move docs for postcopy blocktime feature
migration/bg-snapshot: Do not check for SKIP in iterator
migration: Drop save_live_complete
n Gilbert
Tested-by: Li Zhijian
Reviewed-by: Li Zhijian
Acked-by: Dr. David Alan Gilbert
Reviewed-by: Juraj Marcin
Tested-by: Mario Casquero
Signed-off-by: Peter Xu
---
migration/migration-hmp-cmds.c | 59 ++
1 file changed, 31 insertions(+), 28 deletions(-)
Move it out of vanilla postcopy session, but instead a standalone feature.
When at it, removing the NOTE because it's incorrect now after introduction
of max-postcopy-bandwidth, which can control the throughput even for
postcopy phase.
Reviewed-by: Juraj Marcin
Signed-off-by: Peter Xu
---
Take notes on start/end state of dirty pages for the whole system.
Reviewed-by: Juraj Marcin
Signed-off-by: Peter Xu
---
migration/ram.c| 5 +
migration/trace-events | 1 +
2 files changed, 6 insertions(+)
diff --git a/migration/ram.c b/migration/ram.c
index c66ad3cf8b..a1d0e8ada2
Signed-off-by: Peter Xu
---
include/migration/register.h | 4 ++--
hw/ppc/spapr.c | 2 +-
hw/s390x/s390-stattrib.c | 2 +-
hw/vfio/migration.c| 2 +-
migration/block-dirty-bitmap.c | 2 +-
migration/ram.c| 2 +-
migration/savevm.c
The hook is only defined in two vmstate users ("ram" and "block dirty
bitmap"), meanwhile both of them define the hook exactly the same as the
precopy version. Hence, this postcopy version isn't needed.
No functional change intended.
Signed-off-by: Peter Xu
---
inc
Since we use the same save_complete() hook for both precopy and postcopy,
add a set of helpers to invoke the hook() to dedup the code.
Signed-off-by: Peter Xu
---
migration/savevm.c | 78 ++
1 file changed, 44 insertions(+), 34 deletions(-)
diff
The check over PAGE_DIRTY_FOUND isn't necessary. We could indent one less
and assert that instead.
Reviewed-by: Juraj Marcin
Signed-off-by: Peter Xu
---
migration/ram.c | 20 +++-
1 file changed, 11 insertions(+), 9 deletions(-)
diff --git a/migration/ram.c b/migration/
52, 50, 38, 40, 37, 40, 49,
40, 35, 35, 35, 81, 19, 18, 19, 18, 30,
22, 3, 0, 0, 0, 0, 0, 0, 0, 0]
Cc: Dr. David Alan Gilbert
Cc: Alexey Perevalov
Cc: Markus Armbruster
Tested-by: Mario Casquero
Signed-off-by: Peter Xu
---
migration/migration-hmp-cmds.c | 22 +-
1 fi
On Thu, Jun 12, 2025 at 05:58:14PM -0300, Fabiano Rosas wrote:
> Fabiano Rosas writes:
>
> > Peter Xu writes:
> >
> >> On Mon, Jun 02, 2025 at 10:37:59PM -0300, Fabiano Rosas wrote:
> >>> QAPI_CLONE_MEMBERS is a better option than copying parameters one
x1.local/
But I guess I got the use case wrong; looks like it worked for you.
In that case please switch to memory_region_is_nonvolatile(), and add
proper Fixes, and copy stable.
Thanks,
--
Peter Xu
fixed both issues and queued the series, thanks all. For the other
patch I did null-initialize for the var.
https://gitlab.com/peterx/qemu/-/commits/migration-staging
--
Peter Xu
en revert the last
patch of 1st series, as "postcopy-latency' metrics wasn't around before
applying the 2nd series, or you'll need to use some kernel tracepoints.
This is definitely an awkward series to test when having the two mangled.
Again, feel free to skip that, just FYI!
Thanks,
--
Peter Xu
On Mon, Jun 09, 2025 at 12:18:44PM -0400, Peter Xu wrote:
> v2:
> - Collected R-bs
> - Avoid using "\b" in HMP dumps [Markus, Dave]
>
> The series is based on a small patch from Yanfei Xu here:
>
> Based-on: <20250514115827.3216082-1-yanfei...@bytedanc
On Wed, Jun 11, 2025 at 04:25:13PM +0200, Cédric Le Goater wrote:
> Peter, Fabiano,
>
> The first 2 patches are migration patches. Do you agree if I take them
> through the VFIO queue ?
Yep, please go ahead.
--
Peter Xu
On Wed, Jun 11, 2025 at 04:41:21PM +0400, Marc-André Lureau wrote:
> Hi
>
> On Wed, Jun 11, 2025 at 4:34 PM Peter Xu wrote:
>
> > On Wed, May 21, 2025 at 05:16:13PM +0200, Juraj Marcin wrote:
> > > From: Juraj Marcin
> > >
> > > If a virtual machi
meat.
Note that this is not a request to have you test everything! Just to
mention the bits from test perspective, so just take it as FYI. I
appreciate your help already to test on the recv side!
Thanks,
--
Peter Xu
nges
> and changing the update interval when the VM state changes to RUNNING.
> The update_displaychangelistener() function updates the internal timer,
> and the display is refreshed immediately if the timer is expired.
>
> Signed-off-by: Juraj Marcin
[add Dan]
Ping - anyone is willing to pick this one up?
Thanks,
--
Peter Xu
t; g) config not touching the parameter, value is reset to default:
>
> { 'execute': 'migrate', 'arguments': { ..., 'config': { 'multifd': true } } }
> => "multifd-channels": 2, HMP: multifd-channels: 2
>
> h) config overrides all:
>
> { 'execute': 'migrate', 'arguments': { ..., 'config': {'multifd-channels':
> 16 } } }
> => QMP: "multifd-channels": 16, HMP: multifd-channels: 16
>
> I'll update the variable names and code comments to be more
> precise. Sorry for the noise.
Good to know it's even working. Thanks for digging it.
--
Peter Xu
On Tue, Jun 10, 2025 at 01:53:40PM +, Dr. David Alan Gilbert wrote:
> * Peter Xu (pet...@redhat.com) wrote:
> > On Tue, Jun 10, 2025 at 12:08:23AM +, Dr. David Alan Gilbert wrote:
> > > > diff --git a/qapi/migration.json b/qapi/migration.json
> > > > in
ick comment while we can wait for others to look at the details: when
it involves both qemu and tests changes, please consider splitting that
into two patches. The test patch can be prefixed with "tests/migration:".
Thanks,
--
Peter Xu
es in the
report, maybe I should also mark all of these fields experimental from the
start? So we don't necessarily need to maintain the ABI - the expectation
is even if a mgmt would like to fetch those they should only fetch and dump
it into log so that human can read later only for debugging purposes.
--
Peter Xu
sec -8 sec ]: 0
[8 sec - 16 sec ]: 0
Cc: Dr. David Alan Gilbert
Cc: Markus Armbruster
Signed-off-by: Peter Xu
---
This patch is based on:
[PATCH v2 00/13] migration/postcopy: Blocktime tracking overhaul
Based-on: <20250609191259.9053-1-pet...@redhat.com>
On Mon, Jun 09, 2025 at 06:05:16PM -0400, Peter Xu wrote:
> On Mon, Jun 09, 2025 at 03:12:54PM -0400, Peter Xu wrote:
> > +static void migration_dump_blocktime(Monitor *mon, MigrationInfo *info)
> > +{
> > +if (info->has_postcopy_blocktime) {
> > +
access a guest page that was missing.
These kind of faults are not accounted by blocktime so far.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 68 ++--
migration/trace-events | 3 +-
2 files changed, 59 insertions(+), 12 d
On Mon, Jun 09, 2025 at 03:12:54PM -0400, Peter Xu wrote:
> +static void migration_dump_blocktime(Monitor *mon, MigrationInfo *info)
> +{
> +if (info->has_postcopy_blocktime) {
> +monitor_printf(mon, "Postcopy Blocktime (ms): %" PRIu32 "\n&qu
On Mon, Jun 09, 2025 at 04:41:06PM -0300, Fabiano Rosas wrote:
> Peter Xu writes:
>
> > On Mon, Jun 09, 2025 at 03:02:06PM -0300, Fabiano Rosas wrote:
> >> Peter Xu writes:
> >>
> >> > On Mon, Jun 09, 2025 at 11:37:02AM -0300, Fab
;t created instead this time; the old
"if" trick isn't needed when we're sure it will only happen once now.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 15 ++-
1 file changed, 10 insertions(+), 5 deletions(-)
diff --git
.
One tracepoint is added when walking all the fault entries.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 259 ---
migration/trace-events | 5 +-
2 files changed, 216 insertions(+), 48 deletions(-)
diff --git a/migration/p
statistics for the latency
report, e.g. how many faults were resolved in (2^N, 2^(N+1)) us window, etc.
Comments welcomed, thanks.
Peter Xu (13):
migration: Add option to set postcopy-blocktime
migration/postcopy: Push blocktime start/end into page req mutex
migration/postcopy
Add a global property to allow enabling postcopy-blocktime feature.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/options.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/migration/options.c b/migration/options.c
index 162c72cda4..4e923a2e07 100644
--- a/migration
Add a field to count how many remote faults one vCPU has taken. So far
it's still not used, but will be soon.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 5 +
1 file changed, 5 insertions(+)
diff --git a/migration/postcopy-ram.c b/migration/pos
get_vcpu_blocktime_list() so we don't need to walk the list twice,
meanwhile add the entry checks in qtests for all postcopy tests.
Cc: Markus Armbruster
Cc: Dr. David Alan Gilbert
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
qapi/migration.json
With 64-bit fields, it is trivial. The caution is when exposing any values
in QMP, it was still declared with milliseconds (ms). Hence it's needed to
do the convertion when exporting the values to existing QMP queries.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/pos
Now with 64bits, the offseting using start_time is not needed anymore,
because the array can always remember the whole timestamp.
Then drop the unused parameter in get_low_time_offset() altogether.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 10
e variable at all: since nobody is
touching smp_cpus_down except ourselves, we can safely do the calculation
at the end before decrementing smp_cpus_down.
Hopefully this makes the logic easier to read, side benefit is we drop one
temp var.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
Now with the mutex protection it's not needed anymore.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 23 ++-
1 file changed, 10 insertions(+), 13 deletions(-)
diff --git a/migration/postcopy-ram.c b/migration/postcopy-ram.c
ed".
Cc: Markus Armbruster
Cc: Dr. David Alan Gilbert
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
qapi/migration.json | 7 +++
migration/migration-hmp-cmds.c| 5 +++
migration/postcopy-ram.c | 64 +--
tests/qtest/mig
ayer.
Note that we need a stub for mark_postcopy_blocktime_begin() for Windows
builds.
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/migration.h| 2 +-
migration/postcopy-ram.h | 2 ++
migration/migration.c| 24 ++---
migration/postcopy-ram.c | 56 ++
cktime_start.
- Rename vcpu_blocktime to vcpu_blocktime_total.
- Touch up the trace-events to not dump blocktime ctx pointer
Reviewed-by: Fabiano Rosas
Signed-off-by: Peter Xu
---
migration/postcopy-ram.c | 50
migration/trace-events | 4 ++--
2 files chan
1 - 100 of 2470 matches
Mail list logo
