* [PATCH -v5] x86/kvm: Implement HWCR support
@ 2019-04-08 9:09 Borislav Petkov
2019-04-08 14:41 ` Sean Christopherson
0 siblings, 1 reply; 9+ messages in thread
From: Borislav Petkov @ 2019-04-08 9:09 UTC (permalink / raw)
To: KVM
Cc: lkml, Jim Mattson, Joerg Roedel, Paolo Bonzini,
Radim Krčmář,
Sean Christopherson, Tom Lendacky, Tony Luck, Yazen Ghannam
Hi all,
here's v5 which keeps the HWCR functionality in kvm/x86.c so that
emulation of AMD guests on Intel hw still can work.
--
From: Borislav Petkov <bp@suse.de>
The AMD hardware configuration register has some useful bits which can
be used by guests. Implement McStatusWrEn which can be used by guests
when injecting MCEs with the in-kernel mce-inject module.
For that, we need to set bit 18 - McStatusWrEn - first, before writing
the MCi_STATUS registers (otherwise we #GP).
Add the required machinery to do so.
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Jim Mattson <jmattson@google.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: KVM <kvm@vger.kernel.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Yazen Ghannam <Yazen.Ghannam@amd.com>
---
arch/x86/include/asm/kvm_host.h | 3 +++
arch/x86/kvm/x86.c | 45 ++++++++++++++++++++++++++++-----
2 files changed, 42 insertions(+), 6 deletions(-)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 159b5988292f..541c431df806 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -780,6 +780,9 @@ struct kvm_vcpu_arch {
/* Flush the L1 Data cache for L1TF mitigation on VMENTER */
bool l1tf_flush_l1d;
+
+ /* AMD MSRC001_0015 Hardware Configuration */
+ u64 msr_hwcr;
};
struct kvm_lpage_info {
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 099b851dabaf..e07f21c1d2d4 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2273,6 +2273,30 @@ static void kvmclock_sync_fn(struct work_struct *work)
KVMCLOCK_SYNC_PERIOD);
}
+/*
+ * On AMD, HWCR[McStatusWrEn] controls whether setting MCi_STATUS results in #GP.
+ */
+static bool __set_mci_status(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
+{
+ if (guest_cpuid_is_amd(vcpu)) {
+ struct msr_data tmp;
+
+ tmp.index = MSR_K7_HWCR;
+
+ if (kvm_get_msr_common(vcpu, &tmp))
+ return false;
+
+ /* McStatusWrEn enabled? */
+ if (tmp.data & BIT_ULL(18))
+ return true;
+ }
+
+ if (msr_info->data != 0)
+ return false;
+
+ return true;
+}
+
static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
{
u64 mcg_cap = vcpu->arch.mcg_cap;
@@ -2304,9 +2328,13 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
if ((offset & 0x3) == 0 &&
data != 0 && (data | (1 << 10)) != ~(u64)0)
return -1;
- if (!msr_info->host_initiated &&
- (offset & 0x3) == 1 && data != 0)
- return -1;
+
+ /* MCi_STATUS */
+ if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
+ if (!__set_mci_status(vcpu, msr_info))
+ return -1;
+ }
+
vcpu->arch.mce_banks[offset] = data;
break;
}
@@ -2455,8 +2483,11 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
data &= ~(u64)0x40; /* ignore flush filter disable */
data &= ~(u64)0x100; /* ignore ignne emulation enable */
data &= ~(u64)0x8; /* ignore TLB cache disable */
- data &= ~(u64)0x40000; /* ignore Mc status write enable */
- if (data != 0) {
+
+ /* Handle McStatusWrEn */
+ if (data == BIT_ULL(18)) {
+ vcpu->arch.msr_hwcr = data;
+ } else if (data != 0) {
vcpu_unimpl(vcpu, "unimplemented HWCR wrmsr: 0x%llx\n",
data);
return 1;
@@ -2730,7 +2761,6 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
case MSR_K8_SYSCFG:
case MSR_K8_TSEG_ADDR:
case MSR_K8_TSEG_MASK:
- case MSR_K7_HWCR:
case MSR_VM_HSAVE_PA:
case MSR_K8_INT_PENDING_MSG:
case MSR_AMD64_NB_CFG:
@@ -2894,6 +2924,9 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
case MSR_MISC_FEATURES_ENABLES:
msr_info->data = vcpu->arch.msr_misc_features_enables;
break;
+ case MSR_K7_HWCR:
+ msr_info->data = vcpu->arch.msr_hwcr;
+ break;
default:
if (kvm_pmu_is_valid_msr(vcpu, msr_info->index))
return kvm_pmu_get_msr(vcpu, msr_info->index, &msr_info->data);
--
2.21.0
--
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH -v5] x86/kvm: Implement HWCR support
2019-04-08 9:09 [PATCH -v5] x86/kvm: Implement HWCR support Borislav Petkov
@ 2019-04-08 14:41 ` Sean Christopherson
2019-04-08 14:57 ` Borislav Petkov
0 siblings, 1 reply; 9+ messages in thread
From: Sean Christopherson @ 2019-04-08 14:41 UTC (permalink / raw)
To: Borislav Petkov
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Paolo Bonzini,
Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
On Mon, Apr 08, 2019 at 11:09:46AM +0200, Borislav Petkov wrote:
> Hi all,
>
> here's v5 which keeps the HWCR functionality in kvm/x86.c so that
> emulation of AMD guests on Intel hw still can work.
>
> --
> From: Borislav Petkov <bp@suse.de>
>
> The AMD hardware configuration register has some useful bits which can
> be used by guests. Implement McStatusWrEn which can be used by guests
> when injecting MCEs with the in-kernel mce-inject module.
>
> For that, we need to set bit 18 - McStatusWrEn - first, before writing
> the MCi_STATUS registers (otherwise we #GP).
>
> Add the required machinery to do so.
>
> Signed-off-by: Borislav Petkov <bp@suse.de>
> Cc: Jim Mattson <jmattson@google.com>
> Cc: Joerg Roedel <joro@8bytes.org>
> Cc: KVM <kvm@vger.kernel.org>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Radim Krčmář <rkrcmar@redhat.com>
> Cc: Sean Christopherson <sean.j.christopherson@intel.com>
> Cc: Tom Lendacky <thomas.lendacky@amd.com>
> Cc: Tony Luck <tony.luck@intel.com>
> Cc: Yazen Ghannam <Yazen.Ghannam@amd.com>
> ---
> arch/x86/include/asm/kvm_host.h | 3 +++
> arch/x86/kvm/x86.c | 45 ++++++++++++++++++++++++++++-----
> 2 files changed, 42 insertions(+), 6 deletions(-)
>
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 159b5988292f..541c431df806 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -780,6 +780,9 @@ struct kvm_vcpu_arch {
>
> /* Flush the L1 Data cache for L1TF mitigation on VMENTER */
> bool l1tf_flush_l1d;
> +
> + /* AMD MSRC001_0015 Hardware Configuration */
> + u64 msr_hwcr;
> };
>
> struct kvm_lpage_info {
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 099b851dabaf..e07f21c1d2d4 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -2273,6 +2273,30 @@ static void kvmclock_sync_fn(struct work_struct *work)
> KVMCLOCK_SYNC_PERIOD);
> }
>
> +/*
> + * On AMD, HWCR[McStatusWrEn] controls whether setting MCi_STATUS results in #GP.
> + */
> +static bool __set_mci_status(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
The name and return value semantics are confusing since kvm_set_msr*()
returns '1' to signal fault, '0' to signal success. Most MSR helpers have
"valid" somewhere in the function name to avoid confusion regarding the
return value.
> +{
> + if (guest_cpuid_is_amd(vcpu)) {
> + struct msr_data tmp;
> +
> + tmp.index = MSR_K7_HWCR;
> +
> + if (kvm_get_msr_common(vcpu, &tmp))
No need to get through kvm_get_msr_common(), vcpu->arch.msr_hwcr can be
queried directly. Going that route would likely eliminate the need for a
helper func, i.e. avoid the naming confusion and the comment below.
> + return false;
> +
> + /* McStatusWrEn enabled? */
> + if (tmp.data & BIT_ULL(18))
> + return true;
> + }
> +
> + if (msr_info->data != 0)
Irrevelant if this helper gets dropped, but why not "return !msr_info->data;"?
> + return false;
> +
> + return true;
> +}
> +
> static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> {
> u64 mcg_cap = vcpu->arch.mcg_cap;
> @@ -2304,9 +2328,13 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> if ((offset & 0x3) == 0 &&
> data != 0 && (data | (1 << 10)) != ~(u64)0)
> return -1;
> - if (!msr_info->host_initiated &&
> - (offset & 0x3) == 1 && data != 0)
> - return -1;
> +
> + /* MCi_STATUS */
> + if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
> + if (!__set_mci_status(vcpu, msr_info))
> + return -1;
> + }
> +
> vcpu->arch.mce_banks[offset] = data;
> break;
> }
> @@ -2455,8 +2483,11 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> data &= ~(u64)0x40; /* ignore flush filter disable */
> data &= ~(u64)0x100; /* ignore ignne emulation enable */
> data &= ~(u64)0x8; /* ignore TLB cache disable */
> - data &= ~(u64)0x40000; /* ignore Mc status write enable */
> - if (data != 0) {
> +
> + /* Handle McStatusWrEn */
> + if (data == BIT_ULL(18)) {
> + vcpu->arch.msr_hwcr = data;
> + } else if (data != 0) {
> vcpu_unimpl(vcpu, "unimplemented HWCR wrmsr: 0x%llx\n",
> data);
> return 1;
> @@ -2730,7 +2761,6 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> case MSR_K8_SYSCFG:
> case MSR_K8_TSEG_ADDR:
> case MSR_K8_TSEG_MASK:
> - case MSR_K7_HWCR:
> case MSR_VM_HSAVE_PA:
> case MSR_K8_INT_PENDING_MSG:
> case MSR_AMD64_NB_CFG:
> @@ -2894,6 +2924,9 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> case MSR_MISC_FEATURES_ENABLES:
> msr_info->data = vcpu->arch.msr_misc_features_enables;
> break;
> + case MSR_K7_HWCR:
> + msr_info->data = vcpu->arch.msr_hwcr;
> + break;
> default:
> if (kvm_pmu_is_valid_msr(vcpu, msr_info->index))
> return kvm_pmu_get_msr(vcpu, msr_info->index, &msr_info->data);
> --
> 2.21.0
>
>
> --
> Regards/Gruss,
> Boris.
>
> Good mailing practices for 400: avoid top-posting and trim the reply.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH -v5] x86/kvm: Implement HWCR support
2019-04-08 14:41 ` Sean Christopherson
@ 2019-04-08 14:57 ` Borislav Petkov
2019-04-18 12:28 ` [PATCH -v5.1] " Borislav Petkov
0 siblings, 1 reply; 9+ messages in thread
From: Borislav Petkov @ 2019-04-08 14:57 UTC (permalink / raw)
To: Sean Christopherson
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Paolo Bonzini,
Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
On Mon, Apr 08, 2019 at 07:41:15AM -0700, Sean Christopherson wrote:
> > +{
> > + if (guest_cpuid_is_amd(vcpu)) {
> > + struct msr_data tmp;
> > +
> > + tmp.index = MSR_K7_HWCR;
> > +
> > + if (kvm_get_msr_common(vcpu, &tmp))
>
> No need to get through kvm_get_msr_common(), vcpu->arch.msr_hwcr can be
> queried directly. Going that route would likely eliminate the need for a
> helper func, i.e. avoid the naming confusion and the comment below.
Agree with every point except this: the helper function abstracts
the guest cpuid check and the rest of the handling nicely. If I put
everything in an already too big default: label in set_msr_mce(), then
the code there will become even more unreadable than it is.
Thx.
--
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH -v5.1] x86/kvm: Implement HWCR support
2019-04-08 14:57 ` Borislav Petkov
@ 2019-04-18 12:28 ` Borislav Petkov
2019-04-18 13:56 ` Sean Christopherson
2019-04-18 16:34 ` Paolo Bonzini
0 siblings, 2 replies; 9+ messages in thread
From: Borislav Petkov @ 2019-04-18 12:28 UTC (permalink / raw)
To: Sean Christopherson
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Paolo Bonzini,
Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
Hi all,
ok here's v5.1 with most of Sean's feedback addressed. The function
checking whether HWCR[18] is set, I've renamed to can_set_mci_status()
and left it to return bool because it really is used in boolean context,
answering the question "Can I set MCi_STATUS MSRs?"
And now it all looks simple and clean, just how I like it! :-)
Thx.
---
From: Borislav Petkov <bp@suse.de>
The hardware configuration register has some useful bits which can be
used by guests. Implement McStatusWrEn which can be used by guests when
injecting MCEs with the in-kernel mce-inject module.
For that, we need to set bit 18 - McStatusWrEn - first, before writing
the MCi_STATUS registers (otherwise we #GP).
Add the required machinery to do so.
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Jim Mattson <jmattson@google.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: KVM <kvm@vger.kernel.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Yazen Ghannam <Yazen.Ghannam@amd.com>
---
arch/x86/include/asm/kvm_host.h | 3 +++
arch/x86/kvm/x86.c | 33 +++++++++++++++++++++++++++------
2 files changed, 30 insertions(+), 6 deletions(-)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 159b5988292f..541c431df806 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -780,6 +780,9 @@ struct kvm_vcpu_arch {
/* Flush the L1 Data cache for L1TF mitigation on VMENTER */
bool l1tf_flush_l1d;
+
+ /* AMD MSRC001_0015 Hardware Configuration */
+ u64 msr_hwcr;
};
struct kvm_lpage_info {
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 099b851dabaf..10f6acc6494c 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2273,6 +2273,18 @@ static void kvmclock_sync_fn(struct work_struct *work)
KVMCLOCK_SYNC_PERIOD);
}
+/*
+ * On AMD, HWCR[McStatusWrEn] controls whether setting MCi_STATUS results in #GP.
+ */
+static bool can_set_mci_status(struct kvm_vcpu *vcpu)
+{
+ /* McStatusWrEn enabled? */
+ if (guest_cpuid_is_amd(vcpu))
+ return !!(vcpu->arch.msr_hwcr & BIT_ULL(18));
+
+ return false;
+}
+
static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
{
u64 mcg_cap = vcpu->arch.mcg_cap;
@@ -2304,9 +2316,13 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
if ((offset & 0x3) == 0 &&
data != 0 && (data | (1 << 10)) != ~(u64)0)
return -1;
- if (!msr_info->host_initiated &&
- (offset & 0x3) == 1 && data != 0)
- return -1;
+
+ /* MCi_STATUS */
+ if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
+ if (!can_set_mci_status(vcpu))
+ return -1;
+ }
+
vcpu->arch.mce_banks[offset] = data;
break;
}
@@ -2455,8 +2471,11 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
data &= ~(u64)0x40; /* ignore flush filter disable */
data &= ~(u64)0x100; /* ignore ignne emulation enable */
data &= ~(u64)0x8; /* ignore TLB cache disable */
- data &= ~(u64)0x40000; /* ignore Mc status write enable */
- if (data != 0) {
+
+ /* Handle McStatusWrEn */
+ if (data == BIT_ULL(18)) {
+ vcpu->arch.msr_hwcr = data;
+ } else if (data != 0) {
vcpu_unimpl(vcpu, "unimplemented HWCR wrmsr: 0x%llx\n",
data);
return 1;
@@ -2730,7 +2749,6 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
case MSR_K8_SYSCFG:
case MSR_K8_TSEG_ADDR:
case MSR_K8_TSEG_MASK:
- case MSR_K7_HWCR:
case MSR_VM_HSAVE_PA:
case MSR_K8_INT_PENDING_MSG:
case MSR_AMD64_NB_CFG:
@@ -2894,6 +2912,9 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
case MSR_MISC_FEATURES_ENABLES:
msr_info->data = vcpu->arch.msr_misc_features_enables;
break;
+ case MSR_K7_HWCR:
+ msr_info->data = vcpu->arch.msr_hwcr;
+ break;
default:
if (kvm_pmu_is_valid_msr(vcpu, msr_info->index))
return kvm_pmu_get_msr(vcpu, msr_info->index, &msr_info->data);
--
2.21.0
--
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH -v5.1] x86/kvm: Implement HWCR support
2019-04-18 12:28 ` [PATCH -v5.1] " Borislav Petkov
@ 2019-04-18 13:56 ` Sean Christopherson
2019-04-18 14:19 ` Borislav Petkov
2019-04-18 16:34 ` Paolo Bonzini
1 sibling, 1 reply; 9+ messages in thread
From: Sean Christopherson @ 2019-04-18 13:56 UTC (permalink / raw)
To: Borislav Petkov
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Paolo Bonzini,
Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
On Thu, Apr 18, 2019 at 02:28:50PM +0200, Borislav Petkov wrote:
> Hi all,
>
> ok here's v5.1 with most of Sean's feedback addressed. The function
> checking whether HWCR[18] is set, I've renamed to can_set_mci_status()
> and left it to return bool because it really is used in boolean context,
> answering the question "Can I set MCi_STATUS MSRs?"
>
> And now it all looks simple and clean, just how I like it! :-)
>
> Thx.
>
> ---
> From: Borislav Petkov <bp@suse.de>
>
> The hardware configuration register has some useful bits which can be
> used by guests. Implement McStatusWrEn which can be used by guests when
> injecting MCEs with the in-kernel mce-inject module.
>
> For that, we need to set bit 18 - McStatusWrEn - first, before writing
> the MCi_STATUS registers (otherwise we #GP).
>
> Add the required machinery to do so.
>
> Signed-off-by: Borislav Petkov <bp@suse.de>
> Cc: Jim Mattson <jmattson@google.com>
> Cc: Joerg Roedel <joro@8bytes.org>
> Cc: KVM <kvm@vger.kernel.org>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Radim Krčmář <rkrcmar@redhat.com>
> Cc: Sean Christopherson <sean.j.christopherson@intel.com>
> Cc: Tom Lendacky <thomas.lendacky@amd.com>
> Cc: Tony Luck <tony.luck@intel.com>
> Cc: Yazen Ghannam <Yazen.Ghannam@amd.com>
> ---
> arch/x86/include/asm/kvm_host.h | 3 +++
> arch/x86/kvm/x86.c | 33 +++++++++++++++++++++++++++------
> 2 files changed, 30 insertions(+), 6 deletions(-)
>
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 159b5988292f..541c431df806 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -780,6 +780,9 @@ struct kvm_vcpu_arch {
>
> /* Flush the L1 Data cache for L1TF mitigation on VMENTER */
> bool l1tf_flush_l1d;
> +
> + /* AMD MSRC001_0015 Hardware Configuration */
> + u64 msr_hwcr;
> };
>
> struct kvm_lpage_info {
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 099b851dabaf..10f6acc6494c 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -2273,6 +2273,18 @@ static void kvmclock_sync_fn(struct work_struct *work)
> KVMCLOCK_SYNC_PERIOD);
> }
>
> +/*
> + * On AMD, HWCR[McStatusWrEn] controls whether setting MCi_STATUS results in #GP.
> + */
> +static bool can_set_mci_status(struct kvm_vcpu *vcpu)
> +{
> + /* McStatusWrEn enabled? */
> + if (guest_cpuid_is_amd(vcpu))
> + return !!(vcpu->arch.msr_hwcr & BIT_ULL(18));
> +
> + return false;
> +}
> +
> static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> {
> u64 mcg_cap = vcpu->arch.mcg_cap;
> @@ -2304,9 +2316,13 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> if ((offset & 0x3) == 0 &&
> data != 0 && (data | (1 << 10)) != ~(u64)0)
> return -1;
> - if (!msr_info->host_initiated &&
> - (offset & 0x3) == 1 && data != 0)
> - return -1;
> +
> + /* MCi_STATUS */
> + if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
> + if (!can_set_mci_status(vcpu))
This doesn't allow writing '0' regardless of msr_hwcr.BIT(18), which was
previously supported. And there's no need for multiple if statements.
> + return -1;
> + }
> +
> vcpu->arch.mce_banks[offset] = data;
> break;
> }
> @@ -2455,8 +2471,11 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> data &= ~(u64)0x40; /* ignore flush filter disable */
> data &= ~(u64)0x100; /* ignore ignne emulation enable */
> data &= ~(u64)0x8; /* ignore TLB cache disable */
> - data &= ~(u64)0x40000; /* ignore Mc status write enable */
> - if (data != 0) {
> +
> + /* Handle McStatusWrEn */
> + if (data == BIT_ULL(18)) {
> + vcpu->arch.msr_hwcr = data;
> + } else if (data != 0) {
> vcpu_unimpl(vcpu, "unimplemented HWCR wrmsr: 0x%llx\n",
> data);
> return 1;
> @@ -2730,7 +2749,6 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> case MSR_K8_SYSCFG:
> case MSR_K8_TSEG_ADDR:
> case MSR_K8_TSEG_MASK:
> - case MSR_K7_HWCR:
> case MSR_VM_HSAVE_PA:
> case MSR_K8_INT_PENDING_MSG:
> case MSR_AMD64_NB_CFG:
> @@ -2894,6 +2912,9 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> case MSR_MISC_FEATURES_ENABLES:
> msr_info->data = vcpu->arch.msr_misc_features_enables;
> break;
> + case MSR_K7_HWCR:
> + msr_info->data = vcpu->arch.msr_hwcr;
> + break;
> default:
> if (kvm_pmu_is_valid_msr(vcpu, msr_info->index))
> return kvm_pmu_get_msr(vcpu, msr_info->index, &msr_info->data);
> --
> 2.21.0
>
> --
> Regards/Gruss,
> Boris.
>
> Good mailing practices for 400: avoid top-posting and trim the reply.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH -v5.1] x86/kvm: Implement HWCR support
2019-04-18 13:56 ` Sean Christopherson
@ 2019-04-18 14:19 ` Borislav Petkov
2019-04-18 14:38 ` Sean Christopherson
2019-04-18 16:36 ` Paolo Bonzini
0 siblings, 2 replies; 9+ messages in thread
From: Borislav Petkov @ 2019-04-18 14:19 UTC (permalink / raw)
To: Sean Christopherson
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Paolo Bonzini,
Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
On Thu, Apr 18, 2019 at 06:56:06AM -0700, Sean Christopherson wrote:
> This doesn't allow writing '0' regardless of msr_hwcr.BIT(18), which was
> previously supported.
Restored the old conditional:
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 10f6acc6494c..f74f1280745b 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2318,7 +2318,8 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
return -1;
/* MCi_STATUS */
- if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
+ if (!msr_info->host_initiated &&
+ (offset & 0x3) == 1 && data != 0) {
if (!can_set_mci_status(vcpu))
return -1;
}
> And there's no need for multiple if statements.
It is a bit more readable this way.
Actually, I'd break that if above into smaller if-statements with
flipped logic to make it even more readable:
if (msr_info->host_initiated)
goto set_msr;
if (!(offset & 0x3))
goto set_msr;
if (!data)
goto set_msr;
if (!can_set_mci_status(vcpu))
return -1;
set_msr:
...
Anything else? Can I send v5.2 now?
--
Regards/Gruss,
Boris.
Good mailing practices for 400: avoid top-posting and trim the reply.
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH -v5.1] x86/kvm: Implement HWCR support
2019-04-18 14:19 ` Borislav Petkov
@ 2019-04-18 14:38 ` Sean Christopherson
2019-04-18 16:36 ` Paolo Bonzini
1 sibling, 0 replies; 9+ messages in thread
From: Sean Christopherson @ 2019-04-18 14:38 UTC (permalink / raw)
To: Borislav Petkov
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Paolo Bonzini,
Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
On Thu, Apr 18, 2019 at 04:19:49PM +0200, Borislav Petkov wrote:
> On Thu, Apr 18, 2019 at 06:56:06AM -0700, Sean Christopherson wrote:
> > This doesn't allow writing '0' regardless of msr_hwcr.BIT(18), which was
> > previously supported.
>
> Restored the old conditional:
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 10f6acc6494c..f74f1280745b 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -2318,7 +2318,8 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> return -1;
>
> /* MCi_STATUS */
> - if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
> + if (!msr_info->host_initiated &&
> + (offset & 0x3) == 1 && data != 0) {
> if (!can_set_mci_status(vcpu))
> return -1;
> }
>
> > And there's no need for multiple if statements.
>
> It is a bit more readable this way.
>
> Actually, I'd break that if above into smaller if-statements with
> flipped logic to make it even more readable:
>
> if (msr_info->host_initiated)
> goto set_msr;
>
> if (!(offset & 0x3))
> goto set_msr;
>
> if (!data)
> goto set_msr;
>
> if (!can_set_mci_status(vcpu))
> return -1;
At that point it probably makes sense to rework the whole block as there
are existing issues in both the CTL and STATUS flows:
- Fault cases should return '1', not '-1'.
- host_initiated should be checked for the CTL case
My personal preference would be to combine the host_initiated and !data
checks for brevity, so something like:
if (msr_info->host_initiated || !data)
goto set_msr;
offset = msr - MSR_IA32_MC0_CTL;
if ((offset & 0x3) == 0 && (data | (1 << 10)) != ~(u64)0)
return 1;
if ((offset & 0x3) == 1 && !can_set_mci_status(vcpu))
return 1;
set_msr:
>
> set_msr:
> ...
>
>
> Anything else? Can I send v5.2 now?
>
> --
> Regards/Gruss,
> Boris.
>
> Good mailing practices for 400: avoid top-posting and trim the reply.
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH -v5.1] x86/kvm: Implement HWCR support
2019-04-18 12:28 ` [PATCH -v5.1] " Borislav Petkov
2019-04-18 13:56 ` Sean Christopherson
@ 2019-04-18 16:34 ` Paolo Bonzini
1 sibling, 0 replies; 9+ messages in thread
From: Paolo Bonzini @ 2019-04-18 16:34 UTC (permalink / raw)
To: Borislav Petkov, Sean Christopherson
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
On 18/04/19 14:28, Borislav Petkov wrote:
> The hardware configuration register has some useful bits which can be
> used by guests. Implement McStatusWrEn which can be used by guests when
> injecting MCEs with the in-kernel mce-inject module.
>
> For that, we need to set bit 18 - McStatusWrEn - first, before writing
> the MCi_STATUS registers (otherwise we #GP).
>
> Add the required machinery to do so.
Just one little thing missing:
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index eb49d2cee68a..6e23343c6b36 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -1172,6 +1172,8 @@ bool kvm_rdpmc(struct kvm_vcpu *vcpu)
MSR_MISC_FEATURES_ENABLES,
MSR_AMD64_VIRT_SPEC_CTRL,
MSR_IA32_POWER_CTL,
+
+ MSR_K7_HWCR,
};
static unsigned num_emulated_msrs;
Queued with this change.
Paolo
>
> Signed-off-by: Borislav Petkov <bp@suse.de>
> Cc: Jim Mattson <jmattson@google.com>
> Cc: Joerg Roedel <joro@8bytes.org>
> Cc: KVM <kvm@vger.kernel.org>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Radim Krčmář <rkrcmar@redhat.com>
> Cc: Sean Christopherson <sean.j.christopherson@intel.com>
> Cc: Tom Lendacky <thomas.lendacky@amd.com>
> Cc: Tony Luck <tony.luck@intel.com>
> Cc: Yazen Ghannam <Yazen.Ghannam@amd.com>
> ---
> arch/x86/include/asm/kvm_host.h | 3 +++
> arch/x86/kvm/x86.c | 33 +++++++++++++++++++++++++++------
> 2 files changed, 30 insertions(+), 6 deletions(-)
>
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 159b5988292f..541c431df806 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -780,6 +780,9 @@ struct kvm_vcpu_arch {
>
> /* Flush the L1 Data cache for L1TF mitigation on VMENTER */
> bool l1tf_flush_l1d;
> +
> + /* AMD MSRC001_0015 Hardware Configuration */
> + u64 msr_hwcr;
> };
>
> struct kvm_lpage_info {
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 099b851dabaf..10f6acc6494c 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -2273,6 +2273,18 @@ static void kvmclock_sync_fn(struct work_struct *work)
> KVMCLOCK_SYNC_PERIOD);
> }
>
> +/*
> + * On AMD, HWCR[McStatusWrEn] controls whether setting MCi_STATUS results in #GP.
> + */
> +static bool can_set_mci_status(struct kvm_vcpu *vcpu)
> +{
> + /* McStatusWrEn enabled? */
> + if (guest_cpuid_is_amd(vcpu))
> + return !!(vcpu->arch.msr_hwcr & BIT_ULL(18));
> +
> + return false;
> +}
> +
> static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> {
> u64 mcg_cap = vcpu->arch.mcg_cap;
> @@ -2304,9 +2316,13 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> if ((offset & 0x3) == 0 &&
> data != 0 && (data | (1 << 10)) != ~(u64)0)
> return -1;
> - if (!msr_info->host_initiated &&
> - (offset & 0x3) == 1 && data != 0)
> - return -1;
> +
> + /* MCi_STATUS */
> + if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
> + if (!can_set_mci_status(vcpu))
> + return -1;
> + }
> +
> vcpu->arch.mce_banks[offset] = data;
> break;
> }
> @@ -2455,8 +2471,11 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> data &= ~(u64)0x40; /* ignore flush filter disable */
> data &= ~(u64)0x100; /* ignore ignne emulation enable */
> data &= ~(u64)0x8; /* ignore TLB cache disable */
> - data &= ~(u64)0x40000; /* ignore Mc status write enable */
> - if (data != 0) {
> +
> + /* Handle McStatusWrEn */
> + if (data == BIT_ULL(18)) {
> + vcpu->arch.msr_hwcr = data;
> + } else if (data != 0) {
> vcpu_unimpl(vcpu, "unimplemented HWCR wrmsr: 0x%llx\n",
> data);
> return 1;
> @@ -2730,7 +2749,6 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> case MSR_K8_SYSCFG:
> case MSR_K8_TSEG_ADDR:
> case MSR_K8_TSEG_MASK:
> - case MSR_K7_HWCR:
> case MSR_VM_HSAVE_PA:
> case MSR_K8_INT_PENDING_MSG:
> case MSR_AMD64_NB_CFG:
> @@ -2894,6 +2912,9 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> case MSR_MISC_FEATURES_ENABLES:
> msr_info->data = vcpu->arch.msr_misc_features_enables;
> break;
> + case MSR_K7_HWCR:
> + msr_info->data = vcpu->arch.msr_hwcr;
> + break;
> default:
> if (kvm_pmu_is_valid_msr(vcpu, msr_info->index))
> return kvm_pmu_get_msr(vcpu, msr_info->index, &msr_info->data);
> -- 2.21.0
Queued, thanks.
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH -v5.1] x86/kvm: Implement HWCR support
2019-04-18 14:19 ` Borislav Petkov
2019-04-18 14:38 ` Sean Christopherson
@ 2019-04-18 16:36 ` Paolo Bonzini
1 sibling, 0 replies; 9+ messages in thread
From: Paolo Bonzini @ 2019-04-18 16:36 UTC (permalink / raw)
To: Borislav Petkov, Sean Christopherson
Cc: KVM, lkml, Jim Mattson, Joerg Roedel, Radim Krčmář,
Tom Lendacky, Tony Luck, Yazen Ghannam
On 18/04/19 16:19, Borislav Petkov wrote:
> On Thu, Apr 18, 2019 at 06:56:06AM -0700, Sean Christopherson wrote:
>> This doesn't allow writing '0' regardless of msr_hwcr.BIT(18), which was
>> previously supported.
>
> Restored the old conditional:
>
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 10f6acc6494c..f74f1280745b 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -2318,7 +2318,8 @@ static int set_msr_mce(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
> return -1;
>
> /* MCi_STATUS */
> - if ((offset & 0x3) == 1 && !msr_info->host_initiated) {
> + if (!msr_info->host_initiated &&
> + (offset & 0x3) == 1 && data != 0) {
> if (!can_set_mci_status(vcpu))
> return -1;
> }
>
>> And there's no need for multiple if statements.
>
> It is a bit more readable this way.
>
> Actually, I'd break that if above into smaller if-statements with
> flipped logic to make it even more readable:
>
> if (msr_info->host_initiated)
> goto set_msr;
>
> if (!(offset & 0x3))
> goto set_msr;
>
> if (!data)
> goto set_msr;
>
> if (!can_set_mci_status(vcpu))
> return -1;
>
> set_msr:
> ...
>
>
> Anything else? Can I send v5.2 now?
No need, squashing this in before running away for vacation.
Paolo
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2019-04-18 16:36 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-04-08 9:09 [PATCH -v5] x86/kvm: Implement HWCR support Borislav Petkov
2019-04-08 14:41 ` Sean Christopherson
2019-04-08 14:57 ` Borislav Petkov
2019-04-18 12:28 ` [PATCH -v5.1] " Borislav Petkov
2019-04-18 13:56 ` Sean Christopherson
2019-04-18 14:19 ` Borislav Petkov
2019-04-18 14:38 ` Sean Christopherson
2019-04-18 16:36 ` Paolo Bonzini
2019-04-18 16:34 ` Paolo Bonzini
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).