diff mbox series

KVM: riscv: Support enabling dirty log gradually in small chunks

Message ID 20251103062825.9084-1-dayss1224@gmail.com
State Accepted
Headers show
Series KVM: riscv: Support enabling dirty log gradually in small chunks | expand

Commit Message

Dong Yang Nov. 3, 2025, 6:28 a.m. UTC 
From: Dong Yang <dayss1224@gmail.com>

There is already support of enabling dirty log gradually in small chunks
for x86 in commit 3c9bd4006bfc ("KVM: x86: enable dirty log gradually in
small chunks") and c862626 ("KVM: arm64: Support enabling dirty log
gradually in small chunks"). This adds support for riscv.

x86 and arm64 writes protect both huge pages and normal pages now, so
riscv protect also protects both huge pages and normal pages.

On a nested virtualization setup (RISC-V KVM running inside a QEMU VM
on an [Intel® Core™ i5-12500H] host), I did some tests with a 2G Linux
VM using different backing page sizes. The time taken for
memory_global_dirty_log_start in the L2 QEMU is listed below:

Page Size      Before    After Optimization
  4K            4490.23ms         31.94ms
  2M             48.97ms          45.46ms
  1G             28.40ms          30.93ms

Signed-off-by: Quan Zhou <zhouquan@iscas.ac.cn>
Signed-off-by: Dong Yang <dayss1224@gmail.com>
---
 Documentation/virt/kvm/api.rst    | 2 +-
 arch/riscv/include/asm/kvm_host.h | 3 +++
 arch/riscv/kvm/mmu.c              | 5 ++++-
 3 files changed, 8 insertions(+), 2 deletions(-)

Comments

Anup Patel Nov. 7, 2025, 11:03 a.m. UTC | #1 
On Mon, Nov 3, 2025 at 11:58 AM <dayss1224@gmail.com> wrote:
>
> From: Dong Yang <dayss1224@gmail.com>
>
> There is already support of enabling dirty log gradually in small chunks
> for x86 in commit 3c9bd4006bfc ("KVM: x86: enable dirty log gradually in
> small chunks") and c862626 ("KVM: arm64: Support enabling dirty log
> gradually in small chunks"). This adds support for riscv.
>
> x86 and arm64 writes protect both huge pages and normal pages now, so
> riscv protect also protects both huge pages and normal pages.
>
> On a nested virtualization setup (RISC-V KVM running inside a QEMU VM
> on an [Intel® Core™ i5-12500H] host), I did some tests with a 2G Linux
> VM using different backing page sizes. The time taken for
> memory_global_dirty_log_start in the L2 QEMU is listed below:
>
> Page Size      Before    After Optimization
>   4K            4490.23ms         31.94ms
>   2M             48.97ms          45.46ms
>   1G             28.40ms          30.93ms
>
> Signed-off-by: Quan Zhou <zhouquan@iscas.ac.cn>
> Signed-off-by: Dong Yang <dayss1224@gmail.com>

LGTM.

Reviewed-by: Anup Patel <anup@brainfault.org>

Queued this patch for Linux-6.19

Thanks,
Anup


> ---
>  Documentation/virt/kvm/api.rst    | 2 +-
>  arch/riscv/include/asm/kvm_host.h | 3 +++
>  arch/riscv/kvm/mmu.c              | 5 ++++-
>  3 files changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
> index 57061fa29e6a..3b621c3ae67c 100644
> --- a/Documentation/virt/kvm/api.rst
> +++ b/Documentation/virt/kvm/api.rst
> @@ -8028,7 +8028,7 @@ will be initialized to 1 when created.  This also improves performance because
>  dirty logging can be enabled gradually in small chunks on the first call
>  to KVM_CLEAR_DIRTY_LOG.  KVM_DIRTY_LOG_INITIALLY_SET depends on
>  KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
> -x86 and arm64 for now).
> +x86, arm64 and riscv for now).
>
>  KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
>  KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
> diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
> index 4d794573e3db..848b63f87001 100644
> --- a/arch/riscv/include/asm/kvm_host.h
> +++ b/arch/riscv/include/asm/kvm_host.h
> @@ -59,6 +59,9 @@
>                                          BIT(IRQ_VS_TIMER) | \
>                                          BIT(IRQ_VS_EXT))
>
> +#define KVM_DIRTY_LOG_MANUAL_CAPS   (KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE | \
> +       KVM_DIRTY_LOG_INITIALLY_SET)
> +
>  struct kvm_vm_stat {
>         struct kvm_vm_stat_generic generic;
>  };
> diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
> index 525fb5a330c0..a194eee256d8 100644
> --- a/arch/riscv/kvm/mmu.c
> +++ b/arch/riscv/kvm/mmu.c
> @@ -161,8 +161,11 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
>          * allocated dirty_bitmap[], dirty pages will be tracked while
>          * the memory slot is write protected.
>          */
> -       if (change != KVM_MR_DELETE && new->flags & KVM_MEM_LOG_DIRTY_PAGES)
> +       if (change != KVM_MR_DELETE && new->flags & KVM_MEM_LOG_DIRTY_PAGES) {
> +               if (kvm_dirty_log_manual_protect_and_init_set(kvm))
> +                       return;
>                 mmu_wp_memory_region(kvm, new->id);
> +       }
>  }
>
>  int kvm_arch_prepare_memory_region(struct kvm *kvm,
> --
> 2.34.1
>
diff mbox series

Patch

diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
index 57061fa29e6a..3b621c3ae67c 100644
--- a/Documentation/virt/kvm/api.rst
+++ b/Documentation/virt/kvm/api.rst
@@ -8028,7 +8028,7 @@  will be initialized to 1 when created.  This also improves performance because
 dirty logging can be enabled gradually in small chunks on the first call
 to KVM_CLEAR_DIRTY_LOG.  KVM_DIRTY_LOG_INITIALLY_SET depends on
 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
-x86 and arm64 for now).
+x86, arm64 and riscv for now).
 
 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
index 4d794573e3db..848b63f87001 100644
--- a/arch/riscv/include/asm/kvm_host.h
+++ b/arch/riscv/include/asm/kvm_host.h
@@ -59,6 +59,9 @@ 
 					 BIT(IRQ_VS_TIMER) | \
 					 BIT(IRQ_VS_EXT))
 
+#define KVM_DIRTY_LOG_MANUAL_CAPS   (KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE | \
+	KVM_DIRTY_LOG_INITIALLY_SET)
+
 struct kvm_vm_stat {
 	struct kvm_vm_stat_generic generic;
 };
diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
index 525fb5a330c0..a194eee256d8 100644
--- a/arch/riscv/kvm/mmu.c
+++ b/arch/riscv/kvm/mmu.c
@@ -161,8 +161,11 @@  void kvm_arch_commit_memory_region(struct kvm *kvm,
 	 * allocated dirty_bitmap[], dirty pages will be tracked while
 	 * the memory slot is write protected.
 	 */
-	if (change != KVM_MR_DELETE && new->flags & KVM_MEM_LOG_DIRTY_PAGES)
+	if (change != KVM_MR_DELETE && new->flags & KVM_MEM_LOG_DIRTY_PAGES) {
+		if (kvm_dirty_log_manual_protect_and_init_set(kvm))
+			return;
 		mmu_wp_memory_region(kvm, new->id);
+	}
 }
 
 int kvm_arch_prepare_memory_region(struct kvm *kvm,