* [PATCH v2 0/3] KVM: x86: fixes for speculation bug feature reporting @ 2019-08-21 8:26 Paolo Bonzini 2019-08-21 8:26 ` [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf Paolo Bonzini ` (2 more replies) 0 siblings, 3 replies; 15+ messages in thread From: Paolo Bonzini @ 2019-08-21 8:26 UTC (permalink / raw) To: linux-kernel, kvm; +Cc: jmattson, ehabkost, konrad.wilk Patches 1 and 2 are the same as the previous patch, but using svm_get_supported_cpuid and with a fix to the placement of cpuid_mask. Patch 3 is new and, unlike the previous one, will only be in kvm/next. Paolo Bonzini (3): KVM: x86: fix reporting of AMD speculation bug CPUID leaf KVM: x86: always expose VIRT_SSBD to guests KVM: x86: use Intel speculation bugs and features as derived in generic x86 code arch/x86/kvm/cpuid.c | 27 +++++++++++++++++++-------- arch/x86/kvm/svm.c | 13 +++++++++---- arch/x86/kvm/x86.c | 7 +++++++ 3 files changed, 35 insertions(+), 12 deletions(-) -- 1.8.3.1 ^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf 2019-08-21 8:26 [PATCH v2 0/3] KVM: x86: fixes for speculation bug feature reporting Paolo Bonzini @ 2019-08-21 8:26 ` Paolo Bonzini 2019-08-21 19:45 ` Jim Mattson ` (2 more replies) 2019-08-21 8:26 ` [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests Paolo Bonzini 2019-08-21 8:26 ` [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code Paolo Bonzini 2 siblings, 3 replies; 15+ messages in thread From: Paolo Bonzini @ 2019-08-21 8:26 UTC (permalink / raw) To: linux-kernel, kvm; +Cc: jmattson, ehabkost, konrad.wilk The AMD_* bits have to be set from the vendor-independent feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care about the vendor and they should be set on Intel processors as well. On top of this, SSBD, STIBP and AMD_SSB_NO bit were not set, and VIRT_SSBD does not have to be added manually because it is a cpufeature that comes directly from the host's CPUID bit. Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> --- arch/x86/kvm/cpuid.c | 21 +++++++++++++-------- 1 file changed, 13 insertions(+), 8 deletions(-) diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c index 22c2720cd948..43caeb6059b9 100644 --- a/arch/x86/kvm/cpuid.c +++ b/arch/x86/kvm/cpuid.c @@ -729,18 +729,23 @@ static inline int __do_cpuid_func(struct kvm_cpuid_entry2 *entry, u32 function, g_phys_as = phys_as; entry->eax = g_phys_as | (virt_as << 8); entry->edx = 0; + entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; + cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); /* - * IBRS, IBPB and VIRT_SSBD aren't necessarily present in - * hardware cpuid + * AMD has separate bits for each SPEC_CTRL bit. + * arch/x86/kernel/cpu/bugs.c is kind enough to + * record that in cpufeatures so use them. */ - if (boot_cpu_has(X86_FEATURE_AMD_IBPB)) + if (boot_cpu_has(X86_FEATURE_IBPB)) entry->ebx |= F(AMD_IBPB); - if (boot_cpu_has(X86_FEATURE_AMD_IBRS)) + if (boot_cpu_has(X86_FEATURE_IBRS)) entry->ebx |= F(AMD_IBRS); - if (boot_cpu_has(X86_FEATURE_VIRT_SSBD)) - entry->ebx |= F(VIRT_SSBD); - entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; - cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); + if (boot_cpu_has(X86_FEATURE_STIBP)) + entry->ebx |= F(AMD_STIBP); + if (boot_cpu_has(X86_FEATURE_SSBD)) + entry->ebx |= F(AMD_SSBD); + if (!boot_cpu_has_bug(X86_BUG_SPEC_STORE_BYPASS)) + entry->ebx |= F(AMD_SSB_NO); /* * The preference is to use SPEC CTRL MSR instead of the * VIRT_SPEC MSR. -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 15+ messages in thread
* Re: [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf 2019-08-21 8:26 ` [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf Paolo Bonzini @ 2019-08-21 19:45 ` Jim Mattson 2019-08-22 18:46 ` Konrad Rzeszutek Wilk 2019-11-30 23:27 ` Eric Biggers 2 siblings, 0 replies; 15+ messages in thread From: Jim Mattson @ 2019-08-21 19:45 UTC (permalink / raw) To: Paolo Bonzini Cc: LKML, kvm list, jmattson, Eduardo Habkost, Konrad Rzeszutek Wilk On Wed, Aug 21, 2019 at 1:27 AM Paolo Bonzini <pbonzini@redhat.com> wrote: > > The AMD_* bits have to be set from the vendor-independent > feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care > about the vendor and they should be set on Intel processors as well. > On top of this, SSBD, STIBP and AMD_SSB_NO bit were not set, and > VIRT_SSBD does not have to be added manually because it is a > cpufeature that comes directly from the host's CPUID bit. > > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Jim Mattson <jmattson@google.com> ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf 2019-08-21 8:26 ` [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf Paolo Bonzini 2019-08-21 19:45 ` Jim Mattson @ 2019-08-22 18:46 ` Konrad Rzeszutek Wilk 2019-11-30 23:27 ` Eric Biggers 2 siblings, 0 replies; 15+ messages in thread From: Konrad Rzeszutek Wilk @ 2019-08-22 18:46 UTC (permalink / raw) To: Paolo Bonzini; +Cc: linux-kernel, kvm, jmattson, ehabkost On Wed, Aug 21, 2019 at 10:26:40AM +0200, Paolo Bonzini wrote: > The AMD_* bits have to be set from the vendor-independent > feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care > about the vendor and they should be set on Intel processors as well. > On top of this, SSBD, STIBP and AMD_SSB_NO bit were not set, and > VIRT_SSBD does not have to be added manually because it is a > cpufeature that comes directly from the host's CPUID bit. > > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Thank you! ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf 2019-08-21 8:26 ` [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf Paolo Bonzini 2019-08-21 19:45 ` Jim Mattson 2019-08-22 18:46 ` Konrad Rzeszutek Wilk @ 2019-11-30 23:27 ` Eric Biggers 2019-12-09 17:55 ` Eric Biggers 2019-12-13 22:27 ` Eduardo Habkost 2 siblings, 2 replies; 15+ messages in thread From: Eric Biggers @ 2019-11-30 23:27 UTC (permalink / raw) To: Paolo Bonzini; +Cc: linux-kernel, kvm, jmattson, ehabkost, konrad.wilk Hi Paolo, On Wed, Aug 21, 2019 at 10:26:40AM +0200, Paolo Bonzini wrote: > The AMD_* bits have to be set from the vendor-independent > feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care > about the vendor and they should be set on Intel processors as well. > On top of this, SSBD, STIBP and AMD_SSB_NO bit were not set, and > VIRT_SSBD does not have to be added manually because it is a > cpufeature that comes directly from the host's CPUID bit. > > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > --- > arch/x86/kvm/cpuid.c | 21 +++++++++++++-------- > 1 file changed, 13 insertions(+), 8 deletions(-) > > diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c > index 22c2720cd948..43caeb6059b9 100644 > --- a/arch/x86/kvm/cpuid.c > +++ b/arch/x86/kvm/cpuid.c > @@ -729,18 +729,23 @@ static inline int __do_cpuid_func(struct kvm_cpuid_entry2 *entry, u32 function, > g_phys_as = phys_as; > entry->eax = g_phys_as | (virt_as << 8); > entry->edx = 0; > + entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; > + cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); > /* > - * IBRS, IBPB and VIRT_SSBD aren't necessarily present in > - * hardware cpuid > + * AMD has separate bits for each SPEC_CTRL bit. > + * arch/x86/kernel/cpu/bugs.c is kind enough to > + * record that in cpufeatures so use them. > */ > - if (boot_cpu_has(X86_FEATURE_AMD_IBPB)) > + if (boot_cpu_has(X86_FEATURE_IBPB)) > entry->ebx |= F(AMD_IBPB); > - if (boot_cpu_has(X86_FEATURE_AMD_IBRS)) > + if (boot_cpu_has(X86_FEATURE_IBRS)) > entry->ebx |= F(AMD_IBRS); > - if (boot_cpu_has(X86_FEATURE_VIRT_SSBD)) > - entry->ebx |= F(VIRT_SSBD); > - entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; > - cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); > + if (boot_cpu_has(X86_FEATURE_STIBP)) > + entry->ebx |= F(AMD_STIBP); > + if (boot_cpu_has(X86_FEATURE_SSBD)) > + entry->ebx |= F(AMD_SSBD); > + if (!boot_cpu_has_bug(X86_BUG_SPEC_STORE_BYPASS)) > + entry->ebx |= F(AMD_SSB_NO); > /* > * The preference is to use SPEC CTRL MSR instead of the > * VIRT_SPEC MSR. This patch started causing a warning about an unchecked MSR access, when starting a VM. Processor is: "AMD Ryzen Threadripper 1950X 16-Core Processor" The warning appears both in the host and guest kernel logs. On the host: [ 12.121802] unchecked MSR access error: RDMSR from 0x48 at rIP: 0xffffffff8b049765 (svm_vcpu_run+0x6a5/0x720) [ 12.121806] Call Trace: [ 12.121812] ? kvm_arch_vcpu_ioctl_run+0x902/0x1b70 [ 12.121814] ? kvm_vm_ioctl_irq_line+0x1e/0x30 [ 12.121817] ? kvm_vcpu_ioctl+0x21e/0x560 [ 12.121821] ? vfs_writev+0xc0/0xf0 [ 12.121824] ? do_vfs_ioctl+0x41d/0x690 [ 12.121826] ? ksys_ioctl+0x59/0x90 [ 12.121827] ? __x64_sys_ioctl+0x11/0x20 [ 12.121828] ? do_syscall_64+0x43/0x130 [ 12.121832] ? entry_SYSCALL_64_after_hwframe+0x44/0xa9 On the guest: [ 0.799090] unchecked MSR access error: WRMSR to 0x48 (tried to write 0x0000000000000004) at rIP: 0xffffffff81028272 (speculation_ctrl_update+0x132/0x2c0) [ 0.801823] Call Trace: [ 0.801831] ? seccomp_set_mode_filter+0x18d/0x800 [ 0.801833] speculation_ctrl_update_current+0x21/0x30 [ 0.801837] task_update_spec_tif+0x1d/0x20 [ 0.801839] ssb_prctl_set+0xb5/0xd0 [ 0.801841] arch_seccomp_spec_mitigate+0x2a/0x50 [ 0.801843] seccomp_set_mode_filter+0x788/0x800 [ 0.801845] do_seccomp+0x34/0x200 [ 0.801849] __x64_sys_seccomp+0x15/0x20 [ 0.801852] do_syscall_64+0x4a/0x1f0 [ 0.809349] entry_SYSCALL_64_after_hwframe+0x49/0xbe [ 0.810548] RIP: 0033:0x7f431db92e9d [ 0.811528] Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d b3 5f 0c 00 f7 d8 64 89 01 48 [ 0.814754] RSP: 002b:00007ffca5506788 EFLAGS: 00000246 ORIG_RAX: 000000000000013d [ 0.816075] RAX: ffffffffffffffda RBX: 0000556956c07580 RCX: 00007f431db92e9d [ 0.817367] RDX: 0000556956c023e0 RSI: 0000000000000000 RDI: 0000000000000001 [ 0.818698] RBP: 0000556956c023e0 R08: 0000000000000007 R09: 0000556956b73730 [ 0.819948] R10: 0000556956b7101a R11: 0000000000000246 R12: 000000000000002d [ 0.821184] R13: 0040000000002001 R14: 00007f431d9b4898 R15: 0000000000000000 The VM still boots though. I've actually been seeing this for a while but haven't had a chance to bisect it until now. Reverting the commit (4c6903a0f9d76) on mainline makes the warnings go away. Any ideas? Presumably something isn't working as intended. - Eric ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf 2019-11-30 23:27 ` Eric Biggers @ 2019-12-09 17:55 ` Eric Biggers 2019-12-13 22:27 ` Eduardo Habkost 1 sibling, 0 replies; 15+ messages in thread From: Eric Biggers @ 2019-12-09 17:55 UTC (permalink / raw) To: Paolo Bonzini; +Cc: linux-kernel, kvm, jmattson, ehabkost, konrad.wilk On Sat, Nov 30, 2019 at 03:27:33PM -0800, Eric Biggers wrote: > Hi Paolo, > > On Wed, Aug 21, 2019 at 10:26:40AM +0200, Paolo Bonzini wrote: > > The AMD_* bits have to be set from the vendor-independent > > feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care > > about the vendor and they should be set on Intel processors as well. > > On top of this, SSBD, STIBP and AMD_SSB_NO bit were not set, and > > VIRT_SSBD does not have to be added manually because it is a > > cpufeature that comes directly from the host's CPUID bit. > > > > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > > --- > > arch/x86/kvm/cpuid.c | 21 +++++++++++++-------- > > 1 file changed, 13 insertions(+), 8 deletions(-) > > > > diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c > > index 22c2720cd948..43caeb6059b9 100644 > > --- a/arch/x86/kvm/cpuid.c > > +++ b/arch/x86/kvm/cpuid.c > > @@ -729,18 +729,23 @@ static inline int __do_cpuid_func(struct kvm_cpuid_entry2 *entry, u32 function, > > g_phys_as = phys_as; > > entry->eax = g_phys_as | (virt_as << 8); > > entry->edx = 0; > > + entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; > > + cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); > > /* > > - * IBRS, IBPB and VIRT_SSBD aren't necessarily present in > > - * hardware cpuid > > + * AMD has separate bits for each SPEC_CTRL bit. > > + * arch/x86/kernel/cpu/bugs.c is kind enough to > > + * record that in cpufeatures so use them. > > */ > > - if (boot_cpu_has(X86_FEATURE_AMD_IBPB)) > > + if (boot_cpu_has(X86_FEATURE_IBPB)) > > entry->ebx |= F(AMD_IBPB); > > - if (boot_cpu_has(X86_FEATURE_AMD_IBRS)) > > + if (boot_cpu_has(X86_FEATURE_IBRS)) > > entry->ebx |= F(AMD_IBRS); > > - if (boot_cpu_has(X86_FEATURE_VIRT_SSBD)) > > - entry->ebx |= F(VIRT_SSBD); > > - entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; > > - cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); > > + if (boot_cpu_has(X86_FEATURE_STIBP)) > > + entry->ebx |= F(AMD_STIBP); > > + if (boot_cpu_has(X86_FEATURE_SSBD)) > > + entry->ebx |= F(AMD_SSBD); > > + if (!boot_cpu_has_bug(X86_BUG_SPEC_STORE_BYPASS)) > > + entry->ebx |= F(AMD_SSB_NO); > > /* > > * The preference is to use SPEC CTRL MSR instead of the > > * VIRT_SPEC MSR. > > This patch started causing a warning about an unchecked MSR access, when > starting a VM. > > Processor is: "AMD Ryzen Threadripper 1950X 16-Core Processor" > > The warning appears both in the host and guest kernel logs. > > On the host: > > [ 12.121802] unchecked MSR access error: RDMSR from 0x48 at rIP: 0xffffffff8b049765 (svm_vcpu_run+0x6a5/0x720) > [ 12.121806] Call Trace: > [ 12.121812] ? kvm_arch_vcpu_ioctl_run+0x902/0x1b70 > [ 12.121814] ? kvm_vm_ioctl_irq_line+0x1e/0x30 > [ 12.121817] ? kvm_vcpu_ioctl+0x21e/0x560 > [ 12.121821] ? vfs_writev+0xc0/0xf0 > [ 12.121824] ? do_vfs_ioctl+0x41d/0x690 > [ 12.121826] ? ksys_ioctl+0x59/0x90 > [ 12.121827] ? __x64_sys_ioctl+0x11/0x20 > [ 12.121828] ? do_syscall_64+0x43/0x130 > [ 12.121832] ? entry_SYSCALL_64_after_hwframe+0x44/0xa9 > > On the guest: > > [ 0.799090] unchecked MSR access error: WRMSR to 0x48 (tried to write 0x0000000000000004) at rIP: 0xffffffff81028272 (speculation_ctrl_update+0x132/0x2c0) > [ 0.801823] Call Trace: > [ 0.801831] ? seccomp_set_mode_filter+0x18d/0x800 > [ 0.801833] speculation_ctrl_update_current+0x21/0x30 > [ 0.801837] task_update_spec_tif+0x1d/0x20 > [ 0.801839] ssb_prctl_set+0xb5/0xd0 > [ 0.801841] arch_seccomp_spec_mitigate+0x2a/0x50 > [ 0.801843] seccomp_set_mode_filter+0x788/0x800 > [ 0.801845] do_seccomp+0x34/0x200 > [ 0.801849] __x64_sys_seccomp+0x15/0x20 > [ 0.801852] do_syscall_64+0x4a/0x1f0 > [ 0.809349] entry_SYSCALL_64_after_hwframe+0x49/0xbe > [ 0.810548] RIP: 0033:0x7f431db92e9d > [ 0.811528] Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d b3 5f 0c 00 f7 d8 64 89 01 48 > [ 0.814754] RSP: 002b:00007ffca5506788 EFLAGS: 00000246 ORIG_RAX: 000000000000013d > [ 0.816075] RAX: ffffffffffffffda RBX: 0000556956c07580 RCX: 00007f431db92e9d > [ 0.817367] RDX: 0000556956c023e0 RSI: 0000000000000000 RDI: 0000000000000001 > [ 0.818698] RBP: 0000556956c023e0 R08: 0000000000000007 R09: 0000556956b73730 > [ 0.819948] R10: 0000556956b7101a R11: 0000000000000246 R12: 000000000000002d > [ 0.821184] R13: 0040000000002001 R14: 00007f431d9b4898 R15: 0000000000000000 > > The VM still boots though. > > I've actually been seeing this for a while but haven't had a chance to bisect it > until now. > > Reverting the commit (4c6903a0f9d76) on mainline makes the warnings go away. > > Any ideas? Presumably something isn't working as intended. > Ping. ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf 2019-11-30 23:27 ` Eric Biggers 2019-12-09 17:55 ` Eric Biggers @ 2019-12-13 22:27 ` Eduardo Habkost 1 sibling, 0 replies; 15+ messages in thread From: Eduardo Habkost @ 2019-12-13 22:27 UTC (permalink / raw) To: Eric Biggers; +Cc: Paolo Bonzini, linux-kernel, kvm, konrad.wilk On Sat, Nov 30, 2019 at 03:27:31PM -0800, Eric Biggers wrote: > Hi Paolo, > > On Wed, Aug 21, 2019 at 10:26:40AM +0200, Paolo Bonzini wrote: > > The AMD_* bits have to be set from the vendor-independent > > feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care > > about the vendor and they should be set on Intel processors as well. > > On top of this, SSBD, STIBP and AMD_SSB_NO bit were not set, and > > VIRT_SSBD does not have to be added manually because it is a > > cpufeature that comes directly from the host's CPUID bit. > > > > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > > --- > > arch/x86/kvm/cpuid.c | 21 +++++++++++++-------- > > 1 file changed, 13 insertions(+), 8 deletions(-) > > > > diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c > > index 22c2720cd948..43caeb6059b9 100644 > > --- a/arch/x86/kvm/cpuid.c > > +++ b/arch/x86/kvm/cpuid.c > > @@ -729,18 +729,23 @@ static inline int __do_cpuid_func(struct kvm_cpuid_entry2 *entry, u32 function, > > g_phys_as = phys_as; > > entry->eax = g_phys_as | (virt_as << 8); > > entry->edx = 0; > > + entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; > > + cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); > > /* > > - * IBRS, IBPB and VIRT_SSBD aren't necessarily present in > > - * hardware cpuid > > + * AMD has separate bits for each SPEC_CTRL bit. > > + * arch/x86/kernel/cpu/bugs.c is kind enough to > > + * record that in cpufeatures so use them. > > */ > > - if (boot_cpu_has(X86_FEATURE_AMD_IBPB)) > > + if (boot_cpu_has(X86_FEATURE_IBPB)) > > entry->ebx |= F(AMD_IBPB); > > - if (boot_cpu_has(X86_FEATURE_AMD_IBRS)) > > + if (boot_cpu_has(X86_FEATURE_IBRS)) > > entry->ebx |= F(AMD_IBRS); > > - if (boot_cpu_has(X86_FEATURE_VIRT_SSBD)) > > - entry->ebx |= F(VIRT_SSBD); > > - entry->ebx &= kvm_cpuid_8000_0008_ebx_x86_features; > > - cpuid_mask(&entry->ebx, CPUID_8000_0008_EBX); > > + if (boot_cpu_has(X86_FEATURE_STIBP)) > > + entry->ebx |= F(AMD_STIBP); > > + if (boot_cpu_has(X86_FEATURE_SSBD)) > > + entry->ebx |= F(AMD_SSBD); > > + if (!boot_cpu_has_bug(X86_BUG_SPEC_STORE_BYPASS)) > > + entry->ebx |= F(AMD_SSB_NO); > > /* > > * The preference is to use SPEC CTRL MSR instead of the > > * VIRT_SPEC MSR. > > This patch started causing a warning about an unchecked MSR access, when > starting a VM. > > Processor is: "AMD Ryzen Threadripper 1950X 16-Core Processor" > > The warning appears both in the host and guest kernel logs. > > On the host: > > [ 12.121802] unchecked MSR access error: RDMSR from 0x48 at rIP: 0xffffffff8b049765 (svm_vcpu_run+0x6a5/0x720) > [ 12.121806] Call Trace: > [ 12.121812] ? kvm_arch_vcpu_ioctl_run+0x902/0x1b70 > [ 12.121814] ? kvm_vm_ioctl_irq_line+0x1e/0x30 > [ 12.121817] ? kvm_vcpu_ioctl+0x21e/0x560 > [ 12.121821] ? vfs_writev+0xc0/0xf0 > [ 12.121824] ? do_vfs_ioctl+0x41d/0x690 > [ 12.121826] ? ksys_ioctl+0x59/0x90 > [ 12.121827] ? __x64_sys_ioctl+0x11/0x20 > [ 12.121828] ? do_syscall_64+0x43/0x130 > [ 12.121832] ? entry_SYSCALL_64_after_hwframe+0x44/0xa9 For reference, this is: /* * We do not use IBRS in the kernel. If this vCPU has used the * SPEC_CTRL MSR it may have left it on; save the value and * turn it off. This is much more efficient than blindly adding * it to the atomic save/restore list. Especially as the former * (Saving guest MSRs on vmexit) doesn't even exist in KVM. * * For non-nested case: * If the L01 MSR bitmap does not intercept the MSR, then we need to * save it. * * For nested case: * If the L02 MSR bitmap does not intercept the MSR, then we need to * save it. */ if (unlikely(!msr_write_intercepted(vcpu, MSR_IA32_SPEC_CTRL))) svm->spec_ctrl = native_read_msr(MSR_IA32_SPEC_CTRL); This code looks suspicious. I don't see anything that would prevent the kernel from trying to read the MSR on CPUs that don't have X86_FEATURE_AMD_SSBD (CPUID[0x80000008].EBX[24]) set. Maybe it's a preexisting bug being triggered by the failing WRMSR below: > > On the guest: > > [ 0.799090] unchecked MSR access error: WRMSR to 0x48 (tried to write 0x0000000000000004) at rIP: 0xffffffff81028272 (speculation_ctrl_update+0x132/0x2c0) It looks like WRMSR is being rejected because of: if (!msr->host_initiated && !guest_cpuid_has(vcpu, X86_FEATURE_AMD_IBRS) && !guest_cpuid_has(vcpu, X86_FEATURE_AMD_SSBD)) return 1; My guess is that the actual bug is at do_cpuid_7_mask(), which enables SPEC_CTRL and SPEC_CTRL_SSBD even on AMD hosts, while the SVM MSR emulation code won't let guests write to MSR_IA32_SPEC_CTRL. I don't understand why it was not causing any problems before commit 4c6903a0f9d76, though Can you show output of 'x86info -r' and /proc/cpuinfo in both the host and the guest? > [ 0.801823] Call Trace: > [ 0.801831] ? seccomp_set_mode_filter+0x18d/0x800 > [ 0.801833] speculation_ctrl_update_current+0x21/0x30 > [ 0.801837] task_update_spec_tif+0x1d/0x20 > [ 0.801839] ssb_prctl_set+0xb5/0xd0 > [ 0.801841] arch_seccomp_spec_mitigate+0x2a/0x50 > [ 0.801843] seccomp_set_mode_filter+0x788/0x800 > [ 0.801845] do_seccomp+0x34/0x200 > [ 0.801849] __x64_sys_seccomp+0x15/0x20 > [ 0.801852] do_syscall_64+0x4a/0x1f0 > [ 0.809349] entry_SYSCALL_64_after_hwframe+0x49/0xbe > [ 0.810548] RIP: 0033:0x7f431db92e9d > [ 0.811528] Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d b3 5f 0c 00 f7 d8 64 89 01 48 > [ 0.814754] RSP: 002b:00007ffca5506788 EFLAGS: 00000246 ORIG_RAX: 000000000000013d > [ 0.816075] RAX: ffffffffffffffda RBX: 0000556956c07580 RCX: 00007f431db92e9d > [ 0.817367] RDX: 0000556956c023e0 RSI: 0000000000000000 RDI: 0000000000000001 > [ 0.818698] RBP: 0000556956c023e0 R08: 0000000000000007 R09: 0000556956b73730 > [ 0.819948] R10: 0000556956b7101a R11: 0000000000000246 R12: 000000000000002d > [ 0.821184] R13: 0040000000002001 R14: 00007f431d9b4898 R15: 0000000000000000 > > The VM still boots though. > > I've actually been seeing this for a while but haven't had a chance to bisect it > until now. > > Reverting the commit (4c6903a0f9d76) on mainline makes the warnings go away. > > Any ideas? Presumably something isn't working as intended. > > - Eric > -- Eduardo ^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests 2019-08-21 8:26 [PATCH v2 0/3] KVM: x86: fixes for speculation bug feature reporting Paolo Bonzini 2019-08-21 8:26 ` [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf Paolo Bonzini @ 2019-08-21 8:26 ` Paolo Bonzini 2019-08-21 19:47 ` Jim Mattson ` (2 more replies) 2019-08-21 8:26 ` [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code Paolo Bonzini 2 siblings, 3 replies; 15+ messages in thread From: Paolo Bonzini @ 2019-08-21 8:26 UTC (permalink / raw) To: linux-kernel, kvm; +Cc: jmattson, ehabkost, konrad.wilk Even though it is preferrable to use SPEC_CTRL (represented by X86_FEATURE_AMD_SSBD) instead of VIRT_SPEC, VIRT_SPEC is always supported anyway because otherwise it would be impossible to migrate from old to new CPUs. Make this apparent in the result of KVM_GET_SUPPORTED_CPUID as well. While at it, reuse X86_FEATURE_* constants for the SVM leaf too. However, we need to hide the bit on Intel processors, so move the setting to svm_set_supported_cpuid. Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Reported-by: Eduardo Habkost <ehabkost@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> --- arch/x86/kvm/svm.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c index e3d3b2128f2b..c5120a9519f3 100644 --- a/arch/x86/kvm/svm.c +++ b/arch/x86/kvm/svm.c @@ -68,10 +68,8 @@ #define SEG_TYPE_LDT 2 #define SEG_TYPE_BUSY_TSS16 3 -#define SVM_FEATURE_NPT (1 << 0) #define SVM_FEATURE_LBRV (1 << 1) #define SVM_FEATURE_SVML (1 << 2) -#define SVM_FEATURE_NRIP (1 << 3) #define SVM_FEATURE_TSC_RATE (1 << 4) #define SVM_FEATURE_VMCB_CLEAN (1 << 5) #define SVM_FEATURE_FLUSH_ASID (1 << 6) @@ -5933,6 +5931,8 @@ static void svm_cpuid_update(struct kvm_vcpu *vcpu) guest_cpuid_clear(vcpu, X86_FEATURE_X2APIC); } +#define F(x) bit(X86_FEATURE_##x) + static void svm_set_supported_cpuid(u32 func, struct kvm_cpuid_entry2 *entry) { switch (func) { @@ -5944,6 +5944,11 @@ static void svm_set_supported_cpuid(u32 func, struct kvm_cpuid_entry2 *entry) if (nested) entry->ecx |= (1 << 2); /* Set SVM bit */ break; + case 0x80000008: + if (boot_cpu_has(X86_FEATURE_LS_CFG_SSBD) || + boot_cpu_has(X86_FEATURE_AMD_SSBD)) + entry->ebx |= F(VIRT_SSBD); + break; case 0x8000000A: entry->eax = 1; /* SVM revision 1 */ entry->ebx = 8; /* Lets support 8 ASIDs in case we add proper @@ -5954,11 +5959,11 @@ static void svm_set_supported_cpuid(u32 func, struct kvm_cpuid_entry2 *entry) /* Support next_rip if host supports it */ if (boot_cpu_has(X86_FEATURE_NRIPS)) - entry->edx |= SVM_FEATURE_NRIP; + entry->edx |= F(NRIPS); /* Support NPT for the guest if enabled */ if (npt_enabled) - entry->edx |= SVM_FEATURE_NPT; + entry->edx |= F(NPT); break; case 0x8000001F: -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 15+ messages in thread
* Re: [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests 2019-08-21 8:26 ` [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests Paolo Bonzini @ 2019-08-21 19:47 ` Jim Mattson 2019-08-22 18:44 ` Konrad Rzeszutek Wilk 2019-10-01 22:16 ` Eduardo Habkost 2 siblings, 0 replies; 15+ messages in thread From: Jim Mattson @ 2019-08-21 19:47 UTC (permalink / raw) To: Paolo Bonzini Cc: LKML, kvm list, jmattson, Eduardo Habkost, Konrad Rzeszutek Wilk On Wed, Aug 21, 2019 at 1:27 AM Paolo Bonzini <pbonzini@redhat.com> wrote: > > Even though it is preferrable to use SPEC_CTRL (represented by > X86_FEATURE_AMD_SSBD) instead of VIRT_SPEC, VIRT_SPEC is always > supported anyway because otherwise it would be impossible to > migrate from old to new CPUs. Make this apparent in the > result of KVM_GET_SUPPORTED_CPUID as well. > > While at it, reuse X86_FEATURE_* constants for the SVM leaf too. > > However, we need to hide the bit on Intel processors, so move > the setting to svm_set_supported_cpuid. > > Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> > Reported-by: Eduardo Habkost <ehabkost@redhat.com> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Jim Mattson <jmattson@google.com> ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests 2019-08-21 8:26 ` [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests Paolo Bonzini 2019-08-21 19:47 ` Jim Mattson @ 2019-08-22 18:44 ` Konrad Rzeszutek Wilk 2019-10-01 22:16 ` Eduardo Habkost 2 siblings, 0 replies; 15+ messages in thread From: Konrad Rzeszutek Wilk @ 2019-08-22 18:44 UTC (permalink / raw) To: Paolo Bonzini; +Cc: linux-kernel, kvm, jmattson, ehabkost On Wed, Aug 21, 2019 at 10:26:41AM +0200, Paolo Bonzini wrote: > Even though it is preferrable to use SPEC_CTRL (represented by > X86_FEATURE_AMD_SSBD) instead of VIRT_SPEC, VIRT_SPEC is always > supported anyway because otherwise it would be impossible to > migrate from old to new CPUs. Make this apparent in the > result of KVM_GET_SUPPORTED_CPUID as well. > > While at it, reuse X86_FEATURE_* constants for the SVM leaf too. > > However, we need to hide the bit on Intel processors, so move > the setting to svm_set_supported_cpuid. > > Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Thank you! ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests 2019-08-21 8:26 ` [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests Paolo Bonzini 2019-08-21 19:47 ` Jim Mattson 2019-08-22 18:44 ` Konrad Rzeszutek Wilk @ 2019-10-01 22:16 ` Eduardo Habkost 2019-10-06 16:08 ` Paolo Bonzini 2 siblings, 1 reply; 15+ messages in thread From: Eduardo Habkost @ 2019-10-01 22:16 UTC (permalink / raw) To: Paolo Bonzini; +Cc: linux-kernel, kvm, Jim Mattson, konrad.wilk On Wed, Aug 21, 2019 at 10:26:41AM +0200, Paolo Bonzini wrote: > Even though it is preferrable to use SPEC_CTRL (represented by > X86_FEATURE_AMD_SSBD) instead of VIRT_SPEC, VIRT_SPEC is always > supported anyway because otherwise it would be impossible to > migrate from old to new CPUs. Make this apparent in the > result of KVM_GET_SUPPORTED_CPUID as well. > > While at it, reuse X86_FEATURE_* constants for the SVM leaf too. > > However, we need to hide the bit on Intel processors, so move > the setting to svm_set_supported_cpuid. > > Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> > Reported-by: Eduardo Habkost <ehabkost@redhat.com> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > --- [...] > @@ -5944,6 +5944,11 @@ static void svm_set_supported_cpuid(u32 func, struct kvm_cpuid_entry2 *entry) > if (nested) > entry->ecx |= (1 << 2); /* Set SVM bit */ > break; > + case 0x80000008: > + if (boot_cpu_has(X86_FEATURE_LS_CFG_SSBD) || > + boot_cpu_has(X86_FEATURE_AMD_SSBD)) > + entry->ebx |= F(VIRT_SSBD); > + break; Wasn't the old code at arch/x86/kvm/cpuid.c:__do_cpuid_func() supposed to be deleted? /* * The preference is to use SPEC CTRL MSR instead of the * VIRT_SPEC MSR. */ if (boot_cpu_has(X86_FEATURE_LS_CFG_SSBD) && !boot_cpu_has(X86_FEATURE_AMD_SSBD)) entry->ebx |= F(VIRT_SSBD); -- Eduardo ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests 2019-10-01 22:16 ` Eduardo Habkost @ 2019-10-06 16:08 ` Paolo Bonzini 0 siblings, 0 replies; 15+ messages in thread From: Paolo Bonzini @ 2019-10-06 16:08 UTC (permalink / raw) To: Eduardo Habkost; +Cc: linux-kernel, kvm, Jim Mattson, konrad.wilk On 02/10/19 00:16, Eduardo Habkost wrote: > Wasn't the old code at arch/x86/kvm/cpuid.c:__do_cpuid_func() > supposed to be deleted? > > /* > * The preference is to use SPEC CTRL MSR instead of the > * VIRT_SPEC MSR. > */ > if (boot_cpu_has(X86_FEATURE_LS_CFG_SSBD) && > !boot_cpu_has(X86_FEATURE_AMD_SSBD)) > entry->ebx |= F(VIRT_SSBD); Yeah, it's harmless but also useless. Paolo ^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code 2019-08-21 8:26 [PATCH v2 0/3] KVM: x86: fixes for speculation bug feature reporting Paolo Bonzini 2019-08-21 8:26 ` [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf Paolo Bonzini 2019-08-21 8:26 ` [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests Paolo Bonzini @ 2019-08-21 8:26 ` Paolo Bonzini 2019-08-21 19:48 ` Jim Mattson 2019-08-22 18:45 ` Konrad Rzeszutek Wilk 2 siblings, 2 replies; 15+ messages in thread From: Paolo Bonzini @ 2019-08-21 8:26 UTC (permalink / raw) To: linux-kernel, kvm; +Cc: jmattson, ehabkost, konrad.wilk Similar to AMD bits, set the Intel bits from the vendor-independent feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care about the vendor and they should be set on AMD processors as well. Suggested-by: Jim Mattson <jmattson@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> --- arch/x86/kvm/cpuid.c | 6 ++++++ arch/x86/kvm/x86.c | 7 +++++++ 2 files changed, 13 insertions(+) diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c index 43caeb6059b9..dd5985eb61b4 100644 --- a/arch/x86/kvm/cpuid.c +++ b/arch/x86/kvm/cpuid.c @@ -392,6 +392,12 @@ static inline void do_cpuid_7_mask(struct kvm_cpuid_entry2 *entry, int index) entry->edx &= kvm_cpuid_7_0_edx_x86_features; cpuid_mask(&entry->edx, CPUID_7_EDX); + if (boot_cpu_has(X86_FEATURE_IBPB) && boot_cpu_has(X86_FEATURE_IBRS)) + entry->edx |= F(SPEC_CTRL); + if (boot_cpu_has(X86_FEATURE_STIBP)) + entry->edx |= F(INTEL_STIBP); + if (boot_cpu_has(X86_FEATURE_SSBD)) + entry->edx |= F(SPEC_CTRL_SSBD); /* * We emulate ARCH_CAPABILITIES in software even * if the host doesn't support it. diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 93b0bd45ac73..6b81c7609d09 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -1254,6 +1254,13 @@ static u64 kvm_get_arch_capabilities(void) if (l1tf_vmx_mitigation != VMENTER_L1D_FLUSH_NEVER) data |= ARCH_CAP_SKIP_VMENTRY_L1DFLUSH; + if (!boot_cpu_has_bug(X86_BUG_CPU_MELTDOWN)) + data |= ARCH_CAP_RDCL_NO; + if (!boot_cpu_has_bug(X86_BUG_SPEC_STORE_BYPASS)) + data |= ARCH_CAP_SSB_NO; + if (!boot_cpu_has_bug(X86_BUG_MDS)) + data |= ARCH_CAP_MDS_NO; + return data; } -- 1.8.3.1 ^ permalink raw reply related [flat|nested] 15+ messages in thread
* Re: [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code 2019-08-21 8:26 ` [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code Paolo Bonzini @ 2019-08-21 19:48 ` Jim Mattson 2019-08-22 18:45 ` Konrad Rzeszutek Wilk 1 sibling, 0 replies; 15+ messages in thread From: Jim Mattson @ 2019-08-21 19:48 UTC (permalink / raw) To: Paolo Bonzini Cc: LKML, kvm list, jmattson, Eduardo Habkost, Konrad Rzeszutek Wilk On Wed, Aug 21, 2019 at 1:27 AM Paolo Bonzini <pbonzini@redhat.com> wrote: > > Similar to AMD bits, set the Intel bits from the vendor-independent > feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care > about the vendor and they should be set on AMD processors as well. > > Suggested-by: Jim Mattson <jmattson@google.com> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Jim Mattson <jmattson@google.com> ^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code 2019-08-21 8:26 ` [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code Paolo Bonzini 2019-08-21 19:48 ` Jim Mattson @ 2019-08-22 18:45 ` Konrad Rzeszutek Wilk 1 sibling, 0 replies; 15+ messages in thread From: Konrad Rzeszutek Wilk @ 2019-08-22 18:45 UTC (permalink / raw) To: Paolo Bonzini; +Cc: linux-kernel, kvm, jmattson, ehabkost On Wed, Aug 21, 2019 at 10:26:42AM +0200, Paolo Bonzini wrote: > Similar to AMD bits, set the Intel bits from the vendor-independent > feature and bug flags, because KVM_GET_SUPPORTED_CPUID does not care > about the vendor and they should be set on AMD processors as well. > > Suggested-by: Jim Mattson <jmattson@google.com> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Thank you! ^ permalink raw reply [flat|nested] 15+ messages in thread
end of thread, other threads:[~2019-12-16 20:01 UTC | newest] Thread overview: 15+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2019-08-21 8:26 [PATCH v2 0/3] KVM: x86: fixes for speculation bug feature reporting Paolo Bonzini 2019-08-21 8:26 ` [PATCH 1/3] KVM: x86: fix reporting of AMD speculation bug CPUID leaf Paolo Bonzini 2019-08-21 19:45 ` Jim Mattson 2019-08-22 18:46 ` Konrad Rzeszutek Wilk 2019-11-30 23:27 ` Eric Biggers 2019-12-09 17:55 ` Eric Biggers 2019-12-13 22:27 ` Eduardo Habkost 2019-08-21 8:26 ` [PATCH 2/3] KVM: x86: always expose VIRT_SSBD to guests Paolo Bonzini 2019-08-21 19:47 ` Jim Mattson 2019-08-22 18:44 ` Konrad Rzeszutek Wilk 2019-10-01 22:16 ` Eduardo Habkost 2019-10-06 16:08 ` Paolo Bonzini 2019-08-21 8:26 ` [PATCH 3/3] KVM: x86: use Intel speculation bugs and features as derived in generic x86 code Paolo Bonzini 2019-08-21 19:48 ` Jim Mattson 2019-08-22 18:45 ` Konrad Rzeszutek Wilk
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).