[PATCH] KVM: x86 emulator: fix xchg instruction emulation

[PATCH] KVM: x86 emulator: fix xchg instruction emulation

[Wei Yongjun]

If the destination is a memory operand and the memory
cannot map to a valid page, the xchg instruction emulation
will fail. If so, we should emulate exchange as write
to fix it.

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
---
 arch/x86/kvm/x86.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index d94811e..ac63f6f 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3560,6 +3560,8 @@ static int emulator_cmpxchg_emulated(unsigned long addr,
 		goto emul_write;
 
 	page = gfn_to_page(vcpu->kvm, gpa >> PAGE_SHIFT);
+	if (is_error_page(page))
+		goto emul_write;
 
 	kaddr = kmap_atomic(page, KM_USER0);
 	kaddr += offset_in_page(gpa);
-- 
1.7.0.4

Re: [PATCH] KVM: x86 emulator: fix xchg instruction emulation

[Gleb Natapov]

On Wed, Jul 14, 2010 at 01:15:09PM +0800, Wei Yongjun wrote:
> If the destination is a memory operand and the memory
> cannot map to a valid page, the xchg instruction emulation
> will fail. If so, we should emulate exchange as write
> to fix it.
> 
> Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Acked-by: Gleb Natapov <gleb@redhat.com>

Without this patch locked instruction does not work on io regions.
They don't just fail, they stuck in endless loop.

> ---
>  arch/x86/kvm/x86.c |    2 ++
>  1 files changed, 2 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index d94811e..ac63f6f 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -3560,6 +3560,8 @@ static int emulator_cmpxchg_emulated(unsigned long addr,
>  		goto emul_write;
>  
>  	page = gfn_to_page(vcpu->kvm, gpa >> PAGE_SHIFT);
> +	if (is_error_page(page))
> +		goto emul_write;
>  
>  	kaddr = kmap_atomic(page, KM_USER0);
>  	kaddr += offset_in_page(gpa);
> -- 
> 1.7.0.4
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
			Gleb.

Re: [PATCH] KVM: x86 emulator: fix xchg instruction emulation

[Marcelo Tosatti]

On Wed, Jul 14, 2010 at 01:15:09PM +0800, Wei Yongjun wrote:
> If the destination is a memory operand and the memory
> cannot map to a valid page, the xchg instruction emulation
> will fail. If so, we should emulate exchange as write
> to fix it.
> 
> Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
> ---
>  arch/x86/kvm/x86.c |    2 ++
>  1 files changed, 2 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index d94811e..ac63f6f 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -3560,6 +3560,8 @@ static int emulator_cmpxchg_emulated(unsigned long addr,
>  		goto emul_write;
>  
>  	page = gfn_to_page(vcpu->kvm, gpa >> PAGE_SHIFT);
> +	if (is_error_page(page))
> +		goto emul_write;

Forgot to release page.

[PATCHv2] KVM: x86 emulator: fix xchg instruction emulation

[Wei Yongjun]

If the destination is a memory operand and the memory cannot
map to a valid page, the xchg instruction emulation and locked
instruction will not work on io regions and stuck in endless
loop. We should emulate exchange as write to fix it.

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Acked-by: Gleb Natapov <gleb@redhat.com>
---
 arch/x86/kvm/x86.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 86c8102..84bfb51 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3562,6 +3562,10 @@ static int emulator_cmpxchg_emulated(unsigned long addr,
 		goto emul_write;
 
 	page = gfn_to_page(vcpu->kvm, gpa >> PAGE_SHIFT);
+	if (is_error_page(page)) {
+		kvm_release_page_clean(page);
+		goto emul_write;
+	}
 
 	kaddr = kmap_atomic(page, KM_USER0);
 	kaddr += offset_in_page(gpa);
-- 
1.7.0.4

Re: [PATCHv2] KVM: x86 emulator: fix xchg instruction emulation

[Marcelo Tosatti]

On Thu, Jul 15, 2010 at 08:51:58AM +0800, Wei Yongjun wrote:
> If the destination is a memory operand and the memory cannot
> map to a valid page, the xchg instruction emulation and locked
> instruction will not work on io regions and stuck in endless
> loop. We should emulate exchange as write to fix it.
> 
> Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
> Acked-by: Gleb Natapov <gleb@redhat.com>

Applied, thanks.