Fix SVM hang at startup
Subsystems (2)
cover Fix SVM hang at startup
This series is fixing a SVM hang occurring when starting a SVM requiring more secure memory than available. The hang happens in the SVM when calling UV_ESM.
The following is happening:
1. SVM calls UV_ESM
2. Ultravisor (UV) calls H_SVM_INIT_START
3. Hypervisor (HV) calls UV_REGISTER_MEM_SLOT
4. UV returns error because there is not enough free secure memory
5. HV enter the error path in kvmppc_h_svm_init_start()
6. In the return path, since kvm->arch.secure_guest is not yet set hrfid is called
7. As the HV doesn't know the SVM calling context hrfid is jumping to unknown address in the SVM leading to various expections.
This series fixes the setting of kvm->arch.secure_guest in kvmppc_h_svm_init_start() to ensure that UV_RETURN is called on the return path to get back to the UV.
In addition to ensure that a malicious VM will not call UV reserved Hcall, a check of the Secure bit in the calling MSR is addded to reject such a call.
It is assumed that the UV will filtered out such Hcalls made by a malicious SVM.
Laurent Dufour (2):
KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN
arch/powerpc/kvm/book3s_hv.c | 32 ++++++++++++++++++++---------- arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- 2 files changed, 23 insertions(+), 12 deletions(-)
Signature · 1 line
2.25.2
PATCH 1/2 KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
The Hcall named H_SVM_* are reserved to the Ultravisor. However, nothing prevent a malicious VM or SVM to call them. This could lead to weird result and should be filtered out.
Checking the Secure bit of the calling MSR ensure that the call is coming from either the Ultravisor or a SVM. But any system call made from a SVM are going through the Ultravisor, and the Ultravisor should filter out these malicious call. This way, only the Ultravisor is able to make such a Hcall.
--- arch/powerpc/kvm/book3s_hv.c | 32 +++++++++++++++++++++----------- 1 file changed, 21 insertions(+), 11 deletions(-)
arch/powerpc/kvm/book3s_hv.c +21 -11
@@ -1074,25 +1074,35 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu) kvmppc_get_gpr(vcpu, 6));break ;case H_SVM_PAGE_IN : - ret =kvmppc_h_svm_page_in (vcpu->kvm, -kvmppc_get_gpr (vcpu,4 ), -kvmppc_get_gpr (vcpu,5 ), -kvmppc_get_gpr (vcpu,6 )); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_page_in (vcpu->kvm, +kvmppc_get_gpr (vcpu,4 ), +kvmppc_get_gpr (vcpu,5 ), +kvmppc_get_gpr (vcpu,6 ));break ;case H_SVM_PAGE_OUT : - ret =kvmppc_h_svm_page_out (vcpu->kvm, -kvmppc_get_gpr (vcpu,4 ), -kvmppc_get_gpr (vcpu,5 ), -kvmppc_get_gpr (vcpu,6 )); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_page_out (vcpu->kvm, +kvmppc_get_gpr (vcpu,4 ), +kvmppc_get_gpr (vcpu,5 ), +kvmppc_get_gpr (vcpu,6 ));break ;case H_SVM_INIT_START : - ret =kvmppc_h_svm_init_start (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_start (vcpu->kvm);break ;case H_SVM_INIT_DONE : - ret =kvmppc_h_svm_init_done (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_done (vcpu->kvm);break ;case H_SVM_INIT_ABORT : - ret =kvmppc_h_svm_init_abort (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_abort (vcpu->kvm);break ;default :
Signature · 1 line
2.25.2
Re: [PATCH 1/2] KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
Quoted 9 lines · On Fri, 20 Mar 2020 11:26:42 +0100 Laurent Dufour <[email protected]> wrote:
The Hcall named H_SVM_* are reserved to the Ultravisor. However, nothing prevent a malicious VM or SVM to call them. This could lead to weird result and should be filtered out.
Checking the Secure bit of the calling MSR ensure that the call is coming from either the Ultravisor or a SVM. But any system call made from a SVM are going through the Ultravisor, and the Ultravisor should filter out these malicious call. This way, only the Ultravisor is able to make such a Hcall.
"Ultravisor should filter" ? And what if it doesn't (eg. because of a bug) ?
Shouldn't we also check the HV bit of the calling MSR as well to disambiguate SVM and UV ?
Quoted 18 lines
Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>--- arch/powerpc/kvm/book3s_hv.c | 32 +++++++++++++++++++++----------- 1 file changed, 21 insertions(+), 11 deletions(-)arch/powerpc/kvm/book3s_hv.c +0
-0@@ -1074,25 +1074,35 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu) kvmppc_get_gpr(vcpu, 6));break ;case H_SVM_PAGE_IN :
arch/powerpc/kvm/book3s_hv.c int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu) · line 1082
Quoted 10 lines
arch/powerpc/kvm/book3s_hv.c +6
-4- ret =kvmppc_h_svm_page_in (vcpu->kvm, -kvmppc_get_gpr (vcpu,4 ), -kvmppc_get_gpr (vcpu,5 ), -kvmppc_get_gpr (vcpu,6 )); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_page_in (vcpu->kvm, +kvmppc_get_gpr (vcpu,4 ), +kvmppc_get_gpr (vcpu,5 ), +kvmppc_get_gpr (vcpu,6 ));
If calling kvmppc_h_svm_page_in() produces a "weird result" when the MSR_S bit isn't set, then I think it should do the checking itself, ie. pass vcpu.
This would also prevent adding that many lines in kvmppc_pseries_do_hcall() which is a big enough function already. The checking could be done in a helper in book3s_hv_uvmem.c and used by all UV specific hcalls.
Quoted 33 lines
arch/powerpc/kvm/book3s_hv.c +15
-7break ;case H_SVM_PAGE_OUT : - ret =kvmppc_h_svm_page_out (vcpu->kvm, -kvmppc_get_gpr (vcpu,4 ), -kvmppc_get_gpr (vcpu,5 ), -kvmppc_get_gpr (vcpu,6 )); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_page_out (vcpu->kvm, +kvmppc_get_gpr (vcpu,4 ), +kvmppc_get_gpr (vcpu,5 ), +kvmppc_get_gpr (vcpu,6 ));break ;case H_SVM_INIT_START : - ret =kvmppc_h_svm_init_start (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_start (vcpu->kvm);break ;case H_SVM_INIT_DONE : - ret =kvmppc_h_svm_init_done (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_done (vcpu->kvm);break ;case H_SVM_INIT_ABORT : - ret =kvmppc_h_svm_init_abort (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_abort (vcpu->kvm);break ;default :
Re: [PATCH 1/2] KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
Le 20/03/2020 à 13:22, Greg Kurz a écrit :
Quoted 7 lines
Quoted 4 lines · On Fri, 20 Mar 2020 11:26:42 +0100 Laurent Dufour <[email protected]> wrote:
The Hcall named H_SVM_* are reserved to the Ultravisor. However, nothing prevent a malicious VM or SVM to call them. This could lead to weird result and should be filtered out.
Quoted 7 lines
Quoted 5 lines
Checking the Secure bit of the calling MSR ensure that the call is coming from either the Ultravisor or a SVM. But any system call made from a SVM are going through the Ultravisor, and the Ultravisor should filter out these malicious call. This way, only the Ultravisor is able to make such a Hcall.
"Ultravisor should filter" ? And what if it doesn't (eg. because of a bug) ?
If it doesn't, a malicious SVM would be able to call UV reserved Hcall like H_SVM_INIT_ABORT, etc... which is not a good idea.
Quoted 3 lines
Shouldn't we also check the HV bit of the calling MSR as well to disambiguate SVM and UV ?
That's another way to do so, but since the SVM Hcall are going through the UV, it seems the right place (the UV) to do the filtering.
Quoted 33 lines
Quoted 18 lines
Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>--- arch/powerpc/kvm/book3s_hv.c | 32 +++++++++++++++++++++----------- 1 file changed, 21 insertions(+), 11 deletions(-)arch/powerpc/kvm/book3s_hv.c +0
-0@@ -1074,25 +1074,35 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu) kvmppc_get_gpr(vcpu, 6));break ;case H_SVM_PAGE_IN :arch/powerpc/kvm/book3s_hv.c int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu) · line 1082
Quoted 10 lines
arch/powerpc/kvm/book3s_hv.c +6
-4- ret =kvmppc_h_svm_page_in (vcpu->kvm, -kvmppc_get_gpr (vcpu,4 ), -kvmppc_get_gpr (vcpu,5 ), -kvmppc_get_gpr (vcpu,6 )); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_page_in (vcpu->kvm, +kvmppc_get_gpr (vcpu,4 ), +kvmppc_get_gpr (vcpu,5 ), +kvmppc_get_gpr (vcpu,6 ));If calling
kvmppc_h_svm_page_in()produces a "weird result" when theMSR_Sbit isn't set, then I think it should do the checking itself, ie. pass vcpu.
Quoted 3 lines
This would also prevent adding that many lines in
kvmppc_pseries_do_hcall()which is a big enough function already. The checking could be done in a helper inbook3s_hv_uvmem.cand used by all UV specific hcalls.
I'm not convinced that would be better, and I followed the way checks for other Hcalls has been made (see H_TLB_INVALIDATE,..).
I agree kvmppc_pseries_do_hcall() is long but this is just a big switch(), quite linear.
Quoted 35 lines
Quoted 34 lines
arch/powerpc/kvm/book3s_hv.c +15
-7break ;case H_SVM_PAGE_OUT : - ret =kvmppc_h_svm_page_out (vcpu->kvm, -kvmppc_get_gpr (vcpu,4 ), -kvmppc_get_gpr (vcpu,5 ), -kvmppc_get_gpr (vcpu,6 )); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_page_out (vcpu->kvm, +kvmppc_get_gpr (vcpu,4 ), +kvmppc_get_gpr (vcpu,5 ), +kvmppc_get_gpr (vcpu,6 ));break ;case H_SVM_INIT_START : - ret =kvmppc_h_svm_init_start (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_start (vcpu->kvm);break ;case H_SVM_INIT_DONE : - ret =kvmppc_h_svm_init_done (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_done (vcpu->kvm);break ;case H_SVM_INIT_ABORT : - ret =kvmppc_h_svm_init_abort (vcpu->kvm); + ret =H_UNSUPPORTED ; +if (kvmppc_get_srr1 (vcpu) &MSR_S ) + ret =kvmppc_h_svm_init_abort (vcpu->kvm);break ;default :
Re: [PATCH 1/2] KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
In reply to Greg KurzQuoted 13 lines · Greg Kurz
Quoted 9 lines · On Fri, 20 Mar 2020 11:26:42 +0100 Laurent Dufour <[email protected]> wrote:
The Hcall named H_SVM_* are reserved to the Ultravisor. However, nothing prevent a malicious VM or SVM to call them. This could lead to weird result and should be filtered out.
Checking the Secure bit of the calling MSR ensure that the call is coming from either the Ultravisor or a SVM. But any system call made from a SVM are going through the Ultravisor, and the Ultravisor should filter out these malicious call. This way, only the Ultravisor is able to make such a Hcall.
Quoted 4 lines
"Ultravisor should filter" ? And what if it doesn't (eg. because of a bug) ?
Shouldn't we also check the HV bit of the calling MSR as well to disambiguate SVM and UV ?
The trouble with doing that (checking the HV bit) is that KVM does not expect to see the HV bit set on an interrupt that occurred while we were in the guest, and if it is set, it indicates a serious problem, i.e. that an interrupt occurred while we were in the code that transitions from host context to guest context, or from guest context to host context. In those cases we don't know how much of the transition has been completed and therefore whether we have guest values or host values in the CPU registers (GPRs, FPRs/VSRs, SPRs). If we do see HV set then KVM reports a severe error to userspace which should cause userspace to terminate the guest.
Therefore the UV should *always* have the HV bit clear in HSRR1/SRR1 when transitioning to KVM.
Paul.
Re: [PATCH 1/2] KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
Quoted 30 lines · On Tue, 24 Mar 2020 10:43:23 +1100 Paul Mackerras <[email protected]> wrote:
Quoted 13 lines · Greg Kurz
Quoted 9 lines · On Fri, 20 Mar 2020 11:26:42 +0100 Laurent Dufour <[email protected]> wrote:
The Hcall named H_SVM_* are reserved to the Ultravisor. However, nothing prevent a malicious VM or SVM to call them. This could lead to weird result and should be filtered out.
Checking the Secure bit of the calling MSR ensure that the call is coming from either the Ultravisor or a SVM. But any system call made from a SVM are going through the Ultravisor, and the Ultravisor should filter out these malicious call. This way, only the Ultravisor is able to make such a Hcall.
Quoted 4 lines
"Ultravisor should filter" ? And what if it doesn't (eg. because of a bug) ?
Shouldn't we also check the HV bit of the calling MSR as well to disambiguate SVM and UV ?
The trouble with doing that (checking the HV bit) is that KVM does not expect to see the HV bit set on an interrupt that occurred while we were in the guest, and if it is set, it indicates a serious problem, i.e. that an interrupt occurred while we were in the code that transitions from host context to guest context, or from guest context to host context. In those cases we don't know how much of the transition has been completed and therefore whether we have guest values or host values in the CPU registers (GPRs, FPRs/VSRs, SPRs). If we do see HV set then KVM reports a severe error to userspace which should cause userspace to terminate the guest.
Quoted 3 lines
Therefore the UV should *always* have the HV bit clear in HSRR1/SRR1 when transitioning to KVM.
Indeed... thanks for the clarification. So I guess we'll just assume that the UV doesn't reflect these SVM specific hcalls if they happened to be issued by the guest then.
Cheers,
--
Greg
Quoted 1 line
Paul.
Re: [PATCH 1/2] KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
Le 24/03/2020 à 13:00, Greg Kurz a écrit :
Quoted 33 lines
Quoted 30 lines · On Tue, 24 Mar 2020 10:43:23 +1100 Paul Mackerras <[email protected]> wrote:
Quoted 13 lines · Greg Kurz
Quoted 9 lines · On Fri, 20 Mar 2020 11:26:42 +0100 Laurent Dufour <[email protected]> wrote:
The Hcall named H_SVM_* are reserved to the Ultravisor. However, nothing prevent a malicious VM or SVM to call them. This could lead to weird result and should be filtered out.
Checking the Secure bit of the calling MSR ensure that the call is coming from either the Ultravisor or a SVM. But any system call made from a SVM are going through the Ultravisor, and the Ultravisor should filter out these malicious call. This way, only the Ultravisor is able to make such a Hcall.
Quoted 4 lines
"Ultravisor should filter" ? And what if it doesn't (eg. because of a bug) ?
Shouldn't we also check the HV bit of the calling MSR as well to disambiguate SVM and UV ?
The trouble with doing that (checking the HV bit) is that KVM does not expect to see the HV bit set on an interrupt that occurred while we were in the guest, and if it is set, it indicates a serious problem, i.e. that an interrupt occurred while we were in the code that transitions from host context to guest context, or from guest context to host context. In those cases we don't know how much of the transition has been completed and therefore whether we have guest values or host values in the CPU registers (GPRs, FPRs/VSRs, SPRs). If we do see HV set then KVM reports a severe error to userspace which should cause userspace to terminate the guest.
Quoted 7 lines
Quoted 3 lines
Therefore the UV should *always* have the HV bit clear in HSRR1/SRR1 when transitioning to KVM.
Indeed... thanks for the clarification. So I guess we'll just assume that the UV doesn't reflect these SVM specific hcalls if they happened to be issued by the guest then.
As mentioned in the series's description:
"It is assumed that the UV will filtered out such Hcalls made by a malicious SVM."
Quoted 7 lines
Cheers,
--
GregQuoted 1 line
Paul.
Re: [PATCH 1/2] KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
In reply to Laurent DufourQuoted 10 lines · Laurent Dufour
The Hcall named H_SVM_* are reserved to the Ultravisor. However, nothing prevent a malicious VM or SVM to call them. This could lead to weird result and should be filtered out.
Checking the Secure bit of the calling MSR ensure that the call is coming from either the Ultravisor or a SVM. But any system call made from a SVM are going through the Ultravisor, and the Ultravisor should filter out these malicious call. This way, only the Ultravisor is able to make such a Hcall.
Quoted 5 lines
Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>
Quoted 4 lines
--- arch/powerpc/kvm/book3s_hv.c | 32 +++++++++++++++++++++----------- 1 file changed, 21 insertions(+), 11 deletions(-)
PATCH 2/2 KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN
In reply to Laurent DufourWhen the call to UV_REGISTER_MEM_SLOT is failing, for instance because there is not enough free secured memory, the Hypervisor (HV) has to call UV_RETURN to report the error to the Ultravisor (UV). Then the UV will call H_SVM_INIT_ABORT to abort the securing phase and go back to the calling VM.
If the kvm->arch.secure_guest is not set, in the return path rfid is called but there is no valid context to get back to the SVM since the Hcall has been routed by the Ultravisor.
Move the setting of kvm->arch.secure_guest earlier in kvmppc_h_svm_init_start() so in the return path, UV_RETURN will be called instead of rfid.
--- arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
arch/powerpc/kvm/book3s_hv_uvmem.c +2 -1
@@ -209,6 +209,8 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm)int ret =H_SUCCESS ;int srcu_idx; + kvm->arch.secure_guest =KVMPPC_SECURE_INIT_START ; +if (!kvmppc_uvmem_bitmap)return H_UNSUPPORTED ;
@@ -233,7 +235,6 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) goto out; } } - kvm->arch.secure_guest |=KVMPPC_SECURE_INIT_START ; out:srcu_read_unlock (&kvm->srcu, srcu_idx);return ret;
Signature · 1 line
2.25.2
Re: [PATCH 2/2] KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN
Quoted 36 lines · Laurent Dufour
When the call to
UV_REGISTER_MEM_SLOTis failing, for instance because there is not enough free secured memory, the Hypervisor (HV) has to callUV_RETURNto report the error to the Ultravisor (UV). Then the UV will callH_SVM_INIT_ABORTto abort the securing phase and go back to the calling VM.If the
kvm->arch.secure_guestis not set, in the return path rfid is called but there is no valid context to get back to the SVM since the Hcall has been routed by the Ultravisor.Move the setting of
kvm->arch.secure_guestearlier inkvmppc_h_svm_init_start()so in the return path,UV_RETURNwill be called instead of rfid.Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>--- arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)arch/powerpc/kvm/book3s_hv_uvmem.c +2
-0@@ -209,6 +209,8 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm)int ret =H_SUCCESS ;int srcu_idx; + kvm->arch.secure_guest =KVMPPC_SECURE_INIT_START ; +if (!kvmppc_uvmem_bitmap)return H_UNSUPPORTED ;
arch/powerpc/kvm/book3s_hv_uvmem.c unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) · line 236
Quoted 4 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +0
-1@@ -233,7 +235,6 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) goto out; } } - kvm->arch.secure_guest |=KVMPPC_SECURE_INIT_START ;
There is an assumption that memory slots would have been registered with UV if KVMPPC_SECURE_INIT_START has been done. KVM_PPC_SVM_OFF ioctl will skip unregistration and other steps during reboot if KVMPPC_SECURE_INIT_START hasn't been done.
Have you checked if that path isn't affected by this change?
Regards,
Bharata.
Re: [PATCH 2/2] KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN
Le 20/03/2020 à 12:24, Bharata B Rao a écrit :
Quoted 47 lines · Laurent Dufour
Quoted 36 lines
When the call to
UV_REGISTER_MEM_SLOTis failing, for instance because there is not enough free secured memory, the Hypervisor (HV) has to callUV_RETURNto report the error to the Ultravisor (UV). Then the UV will callH_SVM_INIT_ABORTto abort the securing phase and go back to the calling VM.If the
kvm->arch.secure_guestis not set, in the return path rfid is called but there is no valid context to get back to the SVM since the Hcall has been routed by the Ultravisor.Move the setting of
kvm->arch.secure_guestearlier inkvmppc_h_svm_init_start()so in the return path,UV_RETURNwill be called instead of rfid.Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>--- arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)arch/powerpc/kvm/book3s_hv_uvmem.c +2
-0@@ -209,6 +209,8 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm)int ret =H_SUCCESS ;int srcu_idx; + kvm->arch.secure_guest =KVMPPC_SECURE_INIT_START ; +if (!kvmppc_uvmem_bitmap)return H_UNSUPPORTED ;arch/powerpc/kvm/book3s_hv_uvmem.c unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) · line 236
Quoted 4 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +0
-1@@ -233,7 +235,6 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) goto out; } } - kvm->arch.secure_guest |=KVMPPC_SECURE_INIT_START ;There is an assumption that memory slots would have been registered with UV if
KVMPPC_SECURE_INIT_STARThas been done.KVM_PPC_SVM_OFFioctl will skip unregistration and other steps during reboot ifKVMPPC_SECURE_INIT_STARThasn't been done.
Quoted 1 line
Have you checked if that path isn't affected by this change?
I checked that and didn't find any issue there.
My only concern was that block:
kvm_for_each_vcpu(i, vcpu, kvm) {
spin_lock(&vcpu->arch.vpa_update_lock);
unpin_vpa_reset(kvm, &vcpu->arch.dtl);
unpin_vpa_reset(kvm, &vcpu->arch.slb_shadow);
unpin_vpa_reset(kvm, &vcpu->arch.vpa);
spin_unlock(&vcpu->arch.vpa_update_lock);
}But that seems to be safe.
However I'm not a familiar with the KVM's code, do you think an additional KVMPPC_SECURE_INIT_* value needed here?
Thanks,
Laurent.
Re: [PATCH 2/2] KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN
Quoted 59 lines · Laurent Dufour
Le 20/03/2020 à 12:24, Bharata B Rao a écrit :
Quoted 45 lines · Laurent Dufour
Quoted 34 lines
When the call to
UV_REGISTER_MEM_SLOTis failing, for instance because there is not enough free secured memory, the Hypervisor (HV) has to callUV_RETURNto report the error to the Ultravisor (UV). Then the UV will callH_SVM_INIT_ABORTto abort the securing phase and go back to the calling VM.If the
kvm->arch.secure_guestis not set, in the return path rfid is called but there is no valid context to get back to the SVM since the Hcall has been routed by the Ultravisor.Move the setting of
kvm->arch.secure_guestearlier inkvmppc_h_svm_init_start()so in the return path,UV_RETURNwill be called instead of rfid.Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>--- arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)arch/powerpc/kvm/book3s_hv_uvmem.c +2
-0@@ -209,6 +209,8 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm)int ret =H_SUCCESS ;int srcu_idx; + kvm->arch.secure_guest =KVMPPC_SECURE_INIT_START ; +if (!kvmppc_uvmem_bitmap)return H_UNSUPPORTED ;arch/powerpc/kvm/book3s_hv_uvmem.c unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) · line 236
Quoted 4 lines
arch/powerpc/kvm/book3s_hv_uvmem.c +0
-1@@ -233,7 +235,6 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) goto out; } } - kvm->arch.secure_guest |=KVMPPC_SECURE_INIT_START ;There is an assumption that memory slots would have been registered with UV if
KVMPPC_SECURE_INIT_STARThas been done.KVM_PPC_SVM_OFFioctl will skip unregistration and other steps during reboot ifKVMPPC_SECURE_INIT_STARThasn't been done.Quoted 1 line
Have you checked if that path isn't affected by this change?
I checked that and didn't find any issue there.
My only concern was that block:
kvm_for_each_vcpu(i, vcpu, kvm) { spin_lock(&vcpu->arch.vpa_update_lock); unpin_vpa_reset(kvm, &vcpu->arch.dtl); unpin_vpa_reset(kvm, &vcpu->arch.slb_shadow); unpin_vpa_reset(kvm, &vcpu->arch.vpa); spin_unlock(&vcpu->arch.vpa_update_lock); }
Quoted 1 line
But that seems to be safe.
Yes, looks like.
Quoted 3 lines
However I'm not a familiar with the KVM's code, do you think an additional KVMPPC_SECURE_INIT_* value needed here?
May be not as long as UV can handle the unexpected uv_unregister_mem_slot() calls, we are good.
Regards,
Bharata.
Re: [PATCH 2/2] KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN
In reply to Laurent DufourQuoted 2 lines · Laurent Dufour
When the call to
UV_REGISTER_MEM_SLOTis failing, for instance because there is not enough free secured memory, the Hypervisor (HV) has to call
secure memory,
Quoted 11 lines
UV_RETURNto report the error to the Ultravisor (UV). Then the UV will callH_SVM_INIT_ABORTto abort the securing phase and go back to the calling VM.If the
kvm->arch.secure_guestis not set, in the return path rfid is called but there is no valid context to get back to the SVM since the Hcall has been routed by the Ultravisor.Move the setting of
kvm->arch.secure_guestearlier inkvmppc_h_svm_init_start()so in the return path,UV_RETURNwill be called instead of rfid.
Quoted 5 lines
Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>
Re: [PATCH 2/2] KVM: PPC: Book3S HV: H_SVM_INIT_START must call UV_RETURN
In reply to Laurent DufourQuoted 13 lines · Laurent Dufour
When the call to
UV_REGISTER_MEM_SLOTis failing, for instance because there is not enough free secured memory, the Hypervisor (HV) has to callUV_RETURNto report the error to the Ultravisor (UV). Then the UV will callH_SVM_INIT_ABORTto abort the securing phase and go back to the calling VM.If the
kvm->arch.secure_guestis not set, in the return path rfid is called but there is no valid context to get back to the SVM since the Hcall has been routed by the Ultravisor.Move the setting of
kvm->arch.secure_guestearlier inkvmppc_h_svm_init_start()so in the return path,UV_RETURNwill be called instead of rfid.
Quoted 6 lines
Cc Bharata B Rao <[email protected]>Cc Paul Mackerras <[email protected]>Cc Benjamin Herrenschmidt <[email protected]>Cc Michael Ellerman <[email protected]>Signed-off-by Laurent Dufour <[email protected]>---
I tested this along with the code that rejects the secure transition when it is not enabled in KVM.
I have also forced a KVM_PPC_SVM_OFF (via system_reset) right after setting kvm->arch.secure_guest and nothing bad happened.
Quoted 24 lines
arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)arch/powerpc/kvm/book3s_hv_uvmem.c +2
-1@@ -209,6 +209,8 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm)int ret =H_SUCCESS ;int srcu_idx; + kvm->arch.secure_guest =KVMPPC_SECURE_INIT_START ; +if (!kvmppc_uvmem_bitmap)return H_UNSUPPORTED ;@@ -233,7 +235,6 @@ unsigned long kvmppc_h_svm_init_start(struct kvm *kvm) goto out; } } - kvm->arch.secure_guest |=KVMPPC_SECURE_INIT_START ; out:srcu_read_unlock (&kvm->srcu, srcu_idx);return ret;
Quoted 27 lines · Laurent Dufour
This series is fixing a SVM hang occurring when starting a SVM requiring more secure memory than available. The hang happens in the SVM when calling
UV_ESM.The following is happening:
1. SVM calls
UV_ESM
2. Ultravisor (UV) callsH_SVM_INIT_START
3. Hypervisor (HV) callsUV_REGISTER_MEM_SLOT
4. UV returns error because there is not enough free secure memory
5. HV enter the error path inkvmppc_h_svm_init_start()
6. In the return path, sincekvm->arch.secure_guestis not yet set hrfid is called
7. As the HV doesn't know the SVM calling context hrfid is jumping to unknown address in the SVM leading to various expections.This series fixes the setting of
kvm->arch.secure_guestinkvmppc_h_svm_init_start()to ensure thatUV_RETURNis called on the return path to get back to the UV.In addition to ensure that a malicious VM will not call UV reserved Hcall, a check of the Secure bit in the calling MSR is addded to reject such a call.
It is assumed that the UV will filtered out such Hcalls made by a malicious SVM.
Quoted 7 lines
Laurent Dufour (2):
KVM: PPC: Book3S HV: check caller of H_SVM_* Hcalls
KVM: PPC: Book3S HV:H_SVM_INIT_STARTmust callUV_RETURNarch/powerpc/kvm/book3s_hv.c | 32 ++++++++++++++++++++---------- arch/powerpc/kvm/book3s_hv_uvmem.c | 3 ++- 2 files changed, 23 insertions(+), 12 deletions(-)
Thanks, series applied to my kvm-ppc-next branch.
Paul.