From: Bin Guo <guobin@linux.alibaba.com>
To: Paolo Bonzini <pbonzini@redhat.com>,
Hyman Huang <infra.ai.cloud@bitdeer.com>
Cc: kvm@vger.kernel.org, qemu-devel@nongnu.org
Subject: [PATCH v2] accel/kvm: event-driven wakeup for dirty ring reaper thread
Date: Thu, 4 Jun 2026 15:00:29 +0800 [thread overview]
Message-ID: <20260604070029.34982-1-guobin@linux.alibaba.com> (raw)
The reaper polls with sleep(1) (TODO in the code) and only notices
dirty-limit teardown at the next 1s tick.
Replace the sleep with qemu_poll_ns() on an EventNotifier, kicked
from dirtylimit_state_finalize() after dirtylimit_state has been
cleared. The kick must follow the NULL assignment: kicking earlier
wakes the reaper while dirtylimit_in_service() still returns true,
so it just loops back to wait. A 1s fallback timeout remains as a
liveness backstop.
20 set-/cancel-vcpu-dirty-limit cycles via QMP, |reaper-wake -
cancel-ack| measured with strace on the reaper TID:
before: median 255 ms, max 502 ms
after: median 0.6 ms, max 27 ms
kvm_dirty_ring_reaper_init() returns int again (was made void in
commit 43a5e377f4) to propagate event_notifier_init() failure.
Signed-off-by: Bin Guo <guobin@linux.alibaba.com>
---
Changes in v2:
- Fix compilation error: kvm_init() has no Error **errp parameter, so
kvm_dirty_ring_reaper_init() now uses error_report() directly and
no longer accepts an errp argument.
- Remove the kick from kvm_cpu_exec() KVM_EXIT_DIRTY_RING_FULL handler:
it already reaps synchronously (all vCPUs or the ring-full one), so
a background reaper kick would only be redundant or a no-op.
- Move kick site from dirtylimit_change(false) to
dirtylimit_state_finalize() after dirtylimit_state = NULL, ensuring
the reaper actually proceeds past the dirtylimit_in_service() check.
accel/kvm/kvm-all.c | 42 ++++++++++++++++++++++++++++++++++------
accel/stubs/kvm-stub.c | 4 ++++
include/system/kvm.h | 7 +++++++
include/system/kvm_int.h | 3 +++
system/dirtylimit.c | 2 ++
5 files changed, 52 insertions(+), 6 deletions(-)
diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 96f90ebb24..be005832bc 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -1754,6 +1754,8 @@ static void kvm_set_phys_mem(KVMMemoryListener *kml,
} while (size);
}
+#define KVM_DIRTY_RING_REAPER_FALLBACK_NS (1 * NANOSECONDS_PER_SECOND)
+
static void *kvm_dirty_ring_reaper_thread(void *data)
{
KVMState *s = data;
@@ -1764,12 +1766,18 @@ static void *kvm_dirty_ring_reaper_thread(void *data)
trace_kvm_dirty_ring_reaper("init");
while (true) {
+ GPollFD pfd = {
+ .fd = event_notifier_get_fd(&r->reaper_notifier),
+ .events = G_IO_IN,
+ };
+
r->reaper_state = KVM_DIRTY_RING_REAPER_WAIT;
trace_kvm_dirty_ring_reaper("wait");
- /*
- * TODO: provide a smarter timeout rather than a constant?
- */
- sleep(1);
+
+ qemu_poll_ns(&pfd, 1, KVM_DIRTY_RING_REAPER_FALLBACK_NS);
+
+ /* Drain unconditionally so a stale event can't spin the next loop. */
+ event_notifier_test_and_clear(&r->reaper_notifier);
/* keep sleeping so that dirtylimit not be interfered by reaper */
if (dirtylimit_in_service()) {
@@ -1789,13 +1797,32 @@ static void *kvm_dirty_ring_reaper_thread(void *data)
g_assert_not_reached();
}
-static void kvm_dirty_ring_reaper_init(KVMState *s)
+static int kvm_dirty_ring_reaper_init(KVMState *s)
{
struct KVMDirtyRingReaper *r = &s->reaper;
+ int ret;
+
+ ret = event_notifier_init(&r->reaper_notifier, 0);
+ if (ret < 0) {
+ error_report("Failed to initialize dirty ring reaper notifier: %s",
+ strerror(-ret));
+ return ret;
+ }
qemu_thread_create(&r->reaper_thr, "kvm-reaper",
kvm_dirty_ring_reaper_thread,
s, QEMU_THREAD_JOINABLE);
+ return 0;
+}
+
+void kvm_dirty_ring_reaper_kick(void)
+{
+ KVMState *s = kvm_state;
+
+ if (!s || !s->kvm_dirty_ring_size) {
+ return;
+ }
+ event_notifier_set(&s->reaper.reaper_notifier);
}
static int kvm_dirty_ring_init(KVMState *s)
@@ -3097,7 +3124,10 @@ static int kvm_init(AccelState *as, MachineState *ms)
}
if (s->kvm_dirty_ring_size) {
- kvm_dirty_ring_reaper_init(s);
+ ret = kvm_dirty_ring_reaper_init(s);
+ if (ret < 0) {
+ goto err;
+ }
}
if (kvm_check_extension(kvm_state, KVM_CAP_BINARY_STATS_FD)) {
diff --git a/accel/stubs/kvm-stub.c b/accel/stubs/kvm-stub.c
index c4617caac6..b878598552 100644
--- a/accel/stubs/kvm-stub.c
+++ b/accel/stubs/kvm-stub.c
@@ -134,6 +134,10 @@ uint32_t kvm_dirty_ring_size(void)
return 0;
}
+void kvm_dirty_ring_reaper_kick(void)
+{
+}
+
bool kvm_hwpoisoned_mem(void)
{
return false;
diff --git a/include/system/kvm.h b/include/system/kvm.h
index 5fa33eddda..e127a5eb37 100644
--- a/include/system/kvm.h
+++ b/include/system/kvm.h
@@ -553,6 +553,13 @@ bool kvm_dirty_ring_enabled(void);
uint32_t kvm_dirty_ring_size(void);
+/**
+ * kvm_dirty_ring_reaper_kick - wake the background dirty ring reaper.
+ *
+ * Safe from any thread; no-op when the dirty ring is not in use.
+ */
+void kvm_dirty_ring_reaper_kick(void);
+
void kvm_mark_guest_state_protected(void);
/**
diff --git a/include/system/kvm_int.h b/include/system/kvm_int.h
index 0876aac938..c14ebc927f 100644
--- a/include/system/kvm_int.h
+++ b/include/system/kvm_int.h
@@ -12,6 +12,7 @@
#include "system/memory.h"
#include "qapi/qapi-types-common.h"
#include "qemu/accel.h"
+#include "qemu/event_notifier.h"
#include "qemu/queue.h"
#include "system/kvm.h"
#include "accel/accel-ops.h"
@@ -100,6 +101,8 @@ struct KVMDirtyRingReaper {
QemuThread reaper_thr;
volatile uint64_t reaper_iteration; /* iteration number of reaper thr */
volatile enum KVMDirtyRingReaperState reaper_state; /* reap thr state */
+ /* Wakeup channel: kicked when dirty-limit is torn down. */
+ EventNotifier reaper_notifier;
};
struct KVMState
{
diff --git a/system/dirtylimit.c b/system/dirtylimit.c
index c934ceb0de..a33256ade7 100644
--- a/system/dirtylimit.c
+++ b/system/dirtylimit.c
@@ -239,6 +239,8 @@ void dirtylimit_state_finalize(void)
g_free(dirtylimit_state);
dirtylimit_state = NULL;
+ kvm_dirty_ring_reaper_kick();
+
trace_dirtylimit_state_finalize();
}
--
2.50.1 (Apple Git-155)
next reply other threads:[~2026-06-04 7:00 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-04 7:00 Bin Guo [this message]
2026-06-05 8:22 ` [PATCH v2] accel/kvm: event-driven wakeup for dirty ring reaper thread Hy Man
2026-06-18 0:58 ` Hy Man
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=20260604070029.34982-1-guobin@linux.alibaba.com \
--to=guobin@linux.alibaba.com \
--cc=infra.ai.cloud@bitdeer.com \
--cc=kvm@vger.kernel.org \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox