qemu/KVM-dirty-ring-add-missing-memory-barrier.patch
Jiabo Feng 90057151bb QEMU update to version 6.2.0-92:
- s390x/sclp: Simplify get_sclp_device()
- target/ppc: Remove msr_pr macro
- docs/system/target-arm: Re-alphabetize board list
- migration: Extend query-migrate to provide dirty page limit info
- migration: Implement dirty-limit convergence algo
- migration: Put the detection logic before auto-converge checking
- migration: Refactor auto-converge capability logic
- migration: Introduce dirty-limit capability
- qapi/migration: Introduce vcpu-dirty-limit parameters
- qapi/migration: Introduce x-vcpu-dirty-limit-period parameter
- Change the value of no_ged from true to false
- Allow UNIX socket option for VNC websocket
- tpm_emulator: Avoid double initialization during
- chardev/char-socket: Update AF_UNIX for Windows
- KVM: dirty ring: add missing memory barrier
- i386: reset KVM nested state upon CPU reset
- esp: Handle CMD_BUSRESET by resetting the SCSI bus
- dbus-vmstate: Restrict error checks to registered proxies in dbus_get_proxies
- vfio/pci: Add Ascend310b scend910b support
- target/i386: Export RFDS bit to guests
- target/i386: Add new CPU model SierraForest
- target/i386: Introduce Icelake-Server-v7 to enable TSX
- hw/virtio: handle un-configured shutdown in virtio-pci
- target/s390x: display deprecation status in '-cpu  help'
- target/i386: display deprecation status in '-cpu  help'
- pc-bios/s390-ccw: Fix booting with logical block size < physical block size

Signed-off-by: Jiabo Feng <fengjiabo1@huawei.com>
2024-06-13 14:32:49 +08:00

46 lines
1.5 KiB
Diff

From 63d3efed81ddf36acfec73910a84dee426450012 Mon Sep 17 00:00:00 2001
From: tangbinzy <tangbin_yewu@cmss.chinamobile.com>
Date: Wed, 15 May 2024 01:03:38 +0000
Subject: [PATCH] KVM: dirty ring: add missing memory barrier mainline
inclusion commit 4802bf910eee98312c4a9777ac2567e6a0445c46 category: bugfix
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---------------------------------------------------------------
The KVM_DIRTY_GFN_F_DIRTY flag ensures that the entry is valid. If
the read of the fields are not ordered after the read of the flag,
QEMU might see stale values.
Cc: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: tangbinzy <tangbin_yewu@cmss.chinamobile.com>
---
accel/kvm/kvm-all.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index c477f7a635..e55b2b6a16 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -661,7 +661,11 @@ static void kvm_dirty_ring_mark_page(KVMState *s, uint32_t as_id,
static bool dirty_gfn_is_dirtied(struct kvm_dirty_gfn *gfn)
{
- return gfn->flags == KVM_DIRTY_GFN_F_DIRTY;
+ /*
+ * Read the flags before the value. Pairs with barrier in
+ * KVM's kvm_dirty_ring_push() function.
+ */
+ return qatomic_load_acquire(&gfn->flags) == KVM_DIRTY_GFN_F_DIRTY;
}
static void dirty_gfn_set_collected(struct kvm_dirty_gfn *gfn)
--
2.27.0