[V4,10/10] target/arm/kvm: Write CPU state back to KVM on reset

Commit Message

Salil Mehta Oct. 9, 2023, 8:36 p.m. UTC

From: Jean-Philippe Brucker <[e-mail]>

When a KVM vCPU is reset following a PSCI CPU_ON call, its power state
is not synchronized with KVM at the moment. Because the vCPU is not
marked dirty, we miss the call to kvm_arch_put_registers() that writes
to KVM's MP_STATE. Force mp_state synchronization.

Signed-off-by: Jean-Philippe Brucker <[e-mail]>
Signed-off-by: Salil Mehta <[e-mail]>
Reviewed-by: Alex Bennée <[e-mail]>
Reviewed-by: Gavin Shan <[e-mail]>
---
 target/arm/kvm.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

Comments

Salil Mehta Oct. 9, 2023, 8:54 p.m. UTC | #1

Hello,
Perhaps I need to drop this patch from this patch-set as it is ARM specific.
I pulled this from RFC V2 of Virtual CPU Hotplug Support on ARMv8 and is
not required by other architectures trying to implement Virtual CPU Hotplug
Support.

https://lore.kernel.org/qemu-devel/[e-mail]/

I will retain the Reviewed-by's collected here in the RFC V3 patch-set.

Sorry, for the noise.

Thanks
Salil.

> From: Salil Mehta <[e-mail]>
> Sent: Monday, October 9, 2023 9:36 PM
> To: [e-mail]; [e-mail]
> Cc: Salil Mehta <[e-mail]>; [e-mail]; jean-
> [e-mail]; Jonathan Cameron <[e-mail]>;
> [e-mail]; [e-mail];
> [e-mail]; [e-mail]; [e-mail];
> [e-mail]; [e-mail]; [e-mail];
> [e-mail]; [e-mail]; [e-mail];
> [e-mail]; [e-mail]; [e-mail];
> [e-mail]; [e-mail];
> [e-mail]; [e-mail];
> [e-mail]; [e-mail];
> [e-mail]; [e-mail]; zhukeqian
> <[e-mail]>; wangxiongfeng (C) <[e-mail]>;
> wangyanan (Y) <[e-mail]>; [e-mail];
> [e-mail]; [e-mail]; Linuxarm <[e-mail]>
> Subject: [PATCH V4 10/10] target/arm/kvm: Write CPU state back to KVM on
> reset
> 
> From: Jean-Philippe Brucker <[e-mail]>
> 
> When a KVM vCPU is reset following a PSCI CPU_ON call, its power state
> is not synchronized with KVM at the moment. Because the vCPU is not
> marked dirty, we miss the call to kvm_arch_put_registers() that writes
> to KVM's MP_STATE. Force mp_state synchronization.
> 
> Signed-off-by: Jean-Philippe Brucker <[e-mail]>
> Signed-off-by: Salil Mehta <[e-mail]>
> Reviewed-by: Alex Bennée <[e-mail]>
> Reviewed-by: Gavin Shan <[e-mail]>
> ---
>  target/arm/kvm.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/target/arm/kvm.c b/target/arm/kvm.c
> index b66b936a95..8cb70b9e7c 100644
> --- a/target/arm/kvm.c
> +++ b/target/arm/kvm.c
> @@ -642,11 +642,12 @@ void kvm_arm_cpu_post_load(ARMCPU *cpu)
>  void kvm_arm_reset_vcpu(ARMCPU *cpu)
>  {
>      int ret;
> +    CPUState *cs = CPU(cpu);
> 
>      /* Re-init VCPU so that all registers are set to
>       * their respective reset values.
>       */
> -    ret = kvm_arm_vcpu_init(CPU(cpu));
> +    ret = kvm_arm_vcpu_init(cs);
>      if (ret < 0) {
>          fprintf(stderr, "kvm_arm_vcpu_init failed: %s\n", strerror(-ret));
>          abort();
> @@ -663,6 +664,11 @@ void kvm_arm_reset_vcpu(ARMCPU *cpu)
>       * for the same reason we do so in kvm_arch_get_registers().
>       */
>      write_list_to_cpustate(cpu);
> +    /*
> +     * Ensure we call kvm_arch_put_registers(). The vCPU isn't marked
> dirty if
> +     * it was parked in KVM and is now booting from a PSCI CPU_ON call.
> +     */
> +    cs->vcpu_dirty = true;
>  }
> 
>  /*
> --
> 2.34.1

Patch

diff --git a/target/arm/kvm.c b/target/arm/kvm.c
index b66b936a95..8cb70b9e7c 100644
--- a/target/arm/kvm.c
+++ b/target/arm/kvm.c
@@ -642,11 +642,12 @@  void kvm_arm_cpu_post_load(ARMCPU *cpu)
 void kvm_arm_reset_vcpu(ARMCPU *cpu)
 {
     int ret;
+    CPUState *cs = CPU(cpu);
 
     /* Re-init VCPU so that all registers are set to
      * their respective reset values.
      */
-    ret = kvm_arm_vcpu_init(CPU(cpu));
+    ret = kvm_arm_vcpu_init(cs);
     if (ret < 0) {
         fprintf(stderr, "kvm_arm_vcpu_init failed: %s\n", strerror(-ret));
         abort();
@@ -663,6 +664,11 @@  void kvm_arm_reset_vcpu(ARMCPU *cpu)
      * for the same reason we do so in kvm_arch_get_registers().
      */
     write_list_to_cpustate(cpu);
+    /*
+     * Ensure we call kvm_arch_put_registers(). The vCPU isn't marked dirty if
+     * it was parked in KVM and is now booting from a PSCI CPU_ON call.
+     */
+    cs->vcpu_dirty = true;
 }
 
 /*