* [PATCH] KVM: VMX: fix tlb flush with invalid root
@ 2010-07-03 8:02 Xiao Guangrong
2010-07-05 0:52 ` Sheng Yang
2010-07-05 1:02 ` Marcelo Tosatti
0 siblings, 2 replies; 4+ messages in thread
From: Xiao Guangrong @ 2010-07-03 8:02 UTC (permalink / raw)
To: Avi Kivity; +Cc: Marcelo Tosatti, LKML, KVM list
Commit 341d9b535b6c simplify reload logic while entry guest mode, it
can avoid unnecessary sync-root if KVM_REQ_MMU_RELOAD and
KVM_REQ_MMU_SYNC both set.
But, it cause a issue that when we handle 'KVM_REQ_TLB_FLUSH', the
root is invalid, it is triggered during my test:
Kernel BUG at ffffffffa00212b8 [verbose debug info unavailable]
......
[<ffffffff810f5caf>] ? fget_light+0x111/0x28e
[<ffffffff81103963>] sys_ioctl+0x47/0x6a
[<ffffffff81002c1b>] system_call_fastpath+0x16/0x1b
Code: f0 eb 21 f7 c2 00 00 00 04 74 22 48 8d 45 f0 48 c7 45 f0 00 00 00 00 48 c7 45 f8 00 00 00 00 b9 02 00 00 00 66 0f 38 80 08 77 02 <0f> 0b c9 c3 55 48 89 e5 0f 1f 44 00 00 ba 00 68 00 00 48 8b 8f
RIP [<ffffffffa00212b8>] vmx_flush_tlb+0xdf/0xe3 [kvm_intel]
RSP <ffff8800be269ca8>
Fixed by directly return if the root is not ready.
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
---
arch/x86/include/asm/kvm_host.h | 2 ++
arch/x86/kvm/mmu.c | 2 --
arch/x86/kvm/vmx.c | 5 ++++-
3 files changed, 6 insertions(+), 3 deletions(-)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 2bda624..8f522ec 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -40,6 +40,8 @@
0xFFFFFF0000000000ULL)
#define INVALID_PAGE (~(hpa_t)0)
+#define VALID_PAGE(x) ((x) != INVALID_PAGE)
+
#define UNMAPPED_GVA (~(gpa_t)0)
/* KVM Hugepage definitions for x86 */
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index a0c5c31..399ddb0 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -92,8 +92,6 @@ module_param(oos_shadow, bool, 0644);
#define PT_FIRST_AVAIL_BITS_SHIFT 9
#define PT64_SECOND_AVAIL_BITS_SHIFT 52
-#define VALID_PAGE(x) ((x) != INVALID_PAGE)
-
#define PT64_LEVEL_BITS 9
#define PT64_LEVEL_SHIFT(level) \
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 806ab12..ebaaeaf 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -1831,8 +1831,11 @@ static void exit_lmode(struct kvm_vcpu *vcpu)
static void vmx_flush_tlb(struct kvm_vcpu *vcpu)
{
vpid_sync_context(to_vmx(vcpu));
- if (enable_ept)
+ if (enable_ept) {
+ if (!VALID_PAGE(vcpu->arch.mmu.root_hpa))
+ return;
ept_sync_context(construct_eptp(vcpu->arch.mmu.root_hpa));
+ }
}
static void vmx_decache_cr0_guest_bits(struct kvm_vcpu *vcpu)
--
1.6.1.2
^ permalink raw reply related [flat|nested] 4+ messages in thread* Re: [PATCH] KVM: VMX: fix tlb flush with invalid root
2010-07-03 8:02 [PATCH] KVM: VMX: fix tlb flush with invalid root Xiao Guangrong
@ 2010-07-05 0:52 ` Sheng Yang
2010-07-05 2:58 ` Xiao Guangrong
2010-07-05 1:02 ` Marcelo Tosatti
1 sibling, 1 reply; 4+ messages in thread
From: Sheng Yang @ 2010-07-05 0:52 UTC (permalink / raw)
To: Xiao Guangrong; +Cc: kvm, Avi Kivity, Marcelo Tosatti, LKML
On Saturday 03 July 2010 16:02:42 Xiao Guangrong wrote:
> Commit 341d9b535b6c simplify reload logic while entry guest mode, it
> can avoid unnecessary sync-root if KVM_REQ_MMU_RELOAD and
> KVM_REQ_MMU_SYNC both set.
Which commit you are talking about? Can't find 341d9b535b6c...
--
regards
Yang, Sheng
>
> But, it cause a issue that when we handle 'KVM_REQ_TLB_FLUSH', the
> root is invalid, it is triggered during my test:
>
> Kernel BUG at ffffffffa00212b8 [verbose debug info unavailable]
> ......
>
> [<ffffffff810f5caf>] ? fget_light+0x111/0x28e
> [<ffffffff81103963>] sys_ioctl+0x47/0x6a
> [<ffffffff81002c1b>] system_call_fastpath+0x16/0x1b
> Code: f0 eb 21 f7 c2 00 00 00 04 74 22 48 8d 45 f0 48 c7 45 f0 00 00 00 00
> 48 c7 45 f8 00 00 00 00 b9 02 00 00 00 66 0f 38 80 08 77 02 <0f> 0b c9 c3
> 55 48 89 e5 0f 1f 44 00 00 ba 00 68 00 00 48 8b 8f RIP
> [<ffffffffa00212b8>] vmx_flush_tlb+0xdf/0xe3 [kvm_intel]
> RSP <ffff8800be269ca8>
>
> Fixed by directly return if the root is not ready.
>
> Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
> ---
> arch/x86/include/asm/kvm_host.h | 2 ++
> arch/x86/kvm/mmu.c | 2 --
> arch/x86/kvm/vmx.c | 5 ++++-
> 3 files changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/arch/x86/include/asm/kvm_host.h
> b/arch/x86/include/asm/kvm_host.h index 2bda624..8f522ec 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -40,6 +40,8 @@
> 0xFFFFFF0000000000ULL)
>
> #define INVALID_PAGE (~(hpa_t)0)
> +#define VALID_PAGE(x) ((x) != INVALID_PAGE)
> +
> #define UNMAPPED_GVA (~(gpa_t)0)
>
> /* KVM Hugepage definitions for x86 */
> diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
> index a0c5c31..399ddb0 100644
> --- a/arch/x86/kvm/mmu.c
> +++ b/arch/x86/kvm/mmu.c
> @@ -92,8 +92,6 @@ module_param(oos_shadow, bool, 0644);
> #define PT_FIRST_AVAIL_BITS_SHIFT 9
> #define PT64_SECOND_AVAIL_BITS_SHIFT 52
>
> -#define VALID_PAGE(x) ((x) != INVALID_PAGE)
> -
> #define PT64_LEVEL_BITS 9
>
> #define PT64_LEVEL_SHIFT(level) \
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 806ab12..ebaaeaf 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -1831,8 +1831,11 @@ static void exit_lmode(struct kvm_vcpu *vcpu)
> static void vmx_flush_tlb(struct kvm_vcpu *vcpu)
> {
> vpid_sync_context(to_vmx(vcpu));
> - if (enable_ept)
> + if (enable_ept) {
> + if (!VALID_PAGE(vcpu->arch.mmu.root_hpa))
> + return;
> ept_sync_context(construct_eptp(vcpu->arch.mmu.root_hpa));
> + }
> }
>
> static void vmx_decache_cr0_guest_bits(struct kvm_vcpu *vcpu)
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] KVM: VMX: fix tlb flush with invalid root
2010-07-05 0:52 ` Sheng Yang
@ 2010-07-05 2:58 ` Xiao Guangrong
0 siblings, 0 replies; 4+ messages in thread
From: Xiao Guangrong @ 2010-07-05 2:58 UTC (permalink / raw)
To: Sheng Yang; +Cc: kvm, Avi Kivity, Marcelo Tosatti, LKML
Sheng Yang wrote:
> On Saturday 03 July 2010 16:02:42 Xiao Guangrong wrote:
>> Commit 341d9b535b6c simplify reload logic while entry guest mode, it
>> can avoid unnecessary sync-root if KVM_REQ_MMU_RELOAD and
>> KVM_REQ_MMU_SYNC both set.
>
> Which commit you are talking about? Can't find 341d9b535b6c...
>
Please checkout to the 'next' branch.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] KVM: VMX: fix tlb flush with invalid root
2010-07-03 8:02 [PATCH] KVM: VMX: fix tlb flush with invalid root Xiao Guangrong
2010-07-05 0:52 ` Sheng Yang
@ 2010-07-05 1:02 ` Marcelo Tosatti
1 sibling, 0 replies; 4+ messages in thread
From: Marcelo Tosatti @ 2010-07-05 1:02 UTC (permalink / raw)
To: Xiao Guangrong; +Cc: Avi Kivity, LKML, KVM list
On Sat, Jul 03, 2010 at 04:02:42PM +0800, Xiao Guangrong wrote:
> Commit 341d9b535b6c simplify reload logic while entry guest mode, it
> can avoid unnecessary sync-root if KVM_REQ_MMU_RELOAD and
> KVM_REQ_MMU_SYNC both set.
>
> But, it cause a issue that when we handle 'KVM_REQ_TLB_FLUSH', the
> root is invalid, it is triggered during my test:
>
> Kernel BUG at ffffffffa00212b8 [verbose debug info unavailable]
> ......
>
> [<ffffffff810f5caf>] ? fget_light+0x111/0x28e
> [<ffffffff81103963>] sys_ioctl+0x47/0x6a
> [<ffffffff81002c1b>] system_call_fastpath+0x16/0x1b
> Code: f0 eb 21 f7 c2 00 00 00 04 74 22 48 8d 45 f0 48 c7 45 f0 00 00 00 00 48 c7 45 f8 00 00 00 00 b9 02 00 00 00 66 0f 38 80 08 77 02 <0f> 0b c9 c3 55 48 89 e5 0f 1f 44 00 00 ba 00 68 00 00 48 8b 8f
> RIP [<ffffffffa00212b8>] vmx_flush_tlb+0xdf/0xe3 [kvm_intel]
> RSP <ffff8800be269ca8>
>
> Fixed by directly return if the root is not ready.
>
> Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Applied, thanks.
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2010-07-05 3:02 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-07-03 8:02 [PATCH] KVM: VMX: fix tlb flush with invalid root Xiao Guangrong
2010-07-05 0:52 ` Sheng Yang
2010-07-05 2:58 ` Xiao Guangrong
2010-07-05 1:02 ` Marcelo Tosatti
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox