Add wrapper to read GPR of INVPCID, INVVPID, and INVEPT
v3 cover Add wrapper to read GPR of INVPCID, INVVPID, and INVEPT
Hello,
v3 is similar to v2 except that the commit message of "PATCH v3 2/2" is now clearer and detailed.
VMX code to handle INVPCID, INVVPID, and INVEPT read the same GPR index in VM exit info. This patch series improves that handling by adding a common wrapper function for them.
Patch 2 makes a sublte change in INVPCID type check. Unlike INVVPID and INVEPT, INVPCID is not explicitly documented to check the type before reading the operand from memory. So, this patch moves INVPCID type check to the common switch statement instead of VMX and SVM validating it individually.
Changes in v3:
- Patch 2's commit message is more detailed now.
Changes in v2:
- Keeping the register read visible in the functions.
- Removed INVPCID type check hardcoding and moved error condition to common function.
[v2] https://lore.kernel.org/lkml/[email protected]/
[v1] https://lore.kernel.org/lkml/[email protected]/
Vipin Sharma (2):
KVM: VMX: Add a wrapper to read index of GPR for INVPCID, INVVPID, and
INVEPT
KVM: Move INVPCID type check from vmx and svm to the common
kvm_handle_invpcid()
arch/x86/kvm/svm/svm.c | 5 ----- arch/x86/kvm/vmx/nested.c | 10 ++++++---- arch/x86/kvm/vmx/vmx.c | 9 +++------ arch/x86/kvm/vmx/vmx.h | 5 +++++ arch/x86/kvm/x86.c | 3 ++- 5 files changed, 16 insertions(+), 16 deletions(-)
Signature · 1 line
2.33.1.1089.g2158813163f-goog
PATCH v3 1/2 KVM: VMX: Add a wrapper to read index of GPR for INVPCID, INVVPID, and INVEPT
handle_invept(), handle_invvpid(), handle_invpcid() read the same reg2 on VM exit. Move them to a common wrapper function.
--- arch/x86/kvm/vmx/nested.c | 10 ++++++---- arch/x86/kvm/vmx/vmx.c | 4 +++- arch/x86/kvm/vmx/vmx.h | 5 +++++ 3 files changed, 14 insertions(+), 5 deletions(-)
arch/x86/kvm/vmx/nested.c +6 -4
@@ -5379,7 +5379,7 @@ static int handle_invept(struct kvm_vcpu *vcpu)struct {u64 eptp, gpa; } operand; -int i, r; +int i, r, gpr_index; if (!(vmx->nested.msrs.secondary_ctls_high & SECONDARY_EXEC_ENABLE_EPT) ||
@@ -5392,7 +5392,8 @@ static int handle_invept(struct kvm_vcpu *vcpu)return 1 ; vmx_instruction_info =vmcs_read32 (VMX_INSTRUCTION_INFO ); - type =kvm_register_read (vcpu, (vmx_instruction_info >>28 ) &0xf ); + gpr_index =vmx_get_instr_info_reg2 (vmx_instruction_info); + type =kvm_register_read (vcpu, gpr_index); types = (vmx->nested.msrs.ept_caps >>VMX_EPT_EXTENT_SHIFT ) &6 ;
@@ -5459,7 +5460,7 @@ static int handle_invvpid(struct kvm_vcpu *vcpu)u64 gla; } operand;u16 vpid02; -int r; +int r, gpr_index; if (!(vmx->nested.msrs.secondary_ctls_high & SECONDARY_EXEC_ENABLE_VPID) ||
@@ -5472,7 +5473,8 @@ static int handle_invvpid(struct kvm_vcpu *vcpu)return 1 ; vmx_instruction_info =vmcs_read32 (VMX_INSTRUCTION_INFO ); - type =kvm_register_read (vcpu, (vmx_instruction_info >>28 ) &0xf ); + gpr_index =vmx_get_instr_info_reg2 (vmx_instruction_info); + type =kvm_register_read (vcpu, gpr_index); types = (vmx->nested.msrs.vpid_caps &VMX_VPID_EXTENT_SUPPORTED_MASK ) >>8 ;
arch/x86/kvm/vmx/vmx.c +3 -1
@@ -5494,6 +5494,7 @@ static int handle_invpcid(struct kvm_vcpu *vcpu)u64 pcid;u64 gla; } operand; +int gpr_index;if (!guest_cpuid_has (vcpu,X86_FEATURE_INVPCID )) {kvm_queue_exception (vcpu,UD_VECTOR );
@@ -5501,7 +5502,8 @@ static int handle_invpcid(struct kvm_vcpu *vcpu) } vmx_instruction_info =vmcs_read32 (VMX_INSTRUCTION_INFO ); - type =kvm_register_read (vcpu, (vmx_instruction_info >>28 ) &0xf ); + gpr_index =vmx_get_instr_info_reg2 (vmx_instruction_info); + type =kvm_register_read (vcpu, gpr_index);if (type >3 ) {kvm_inject_gp (vcpu,0 );
arch/x86/kvm/vmx/vmx.h +5 -0
@@ -522,4 +522,9 @@ static inline bool vmx_guest_state_valid(struct kvm_vcpu *vcpu)void dump_vmcs (struct kvm_vcpu *vcpu); +static inline int vmx_get_instr_info_reg2 (u32 vmx_instr_info) +{ +return (vmx_instr_info >>28 ) &0xf ; +} +#endif /* __KVM_X86_VMX_H */
Signature · 1 line
2.33.1.1089.g2158813163f-goog
PATCH v3 2/2 KVM: Move INVPCID type check from vmx and svm to the common kvm_handle_invpcid()
In reply to Vipin SharmaHandle #GP on INVPCID due to an invalid type in the common switch statement instead of relying on the callers (VMX and SVM) to manually validate the type.
Unlike INVVPID and INVEPT, INVPCID is not explicitly documented to check the type before reading the operand from memory, so deferring the type validity check until after that point is architecturally allowed.
--- arch/x86/kvm/svm/svm.c | 5 ----- arch/x86/kvm/vmx/vmx.c | 5 ----- arch/x86/kvm/x86.c | 3 ++- 3 files changed, 2 insertions(+), 11 deletions(-)
arch/x86/kvm/svm/svm.c +0 -5
@@ -3119,11 +3119,6 @@ static int invpcid_interception(struct kvm_vcpu *vcpu) type = svm->vmcb->control.exit_info_2; gva = svm->vmcb->control.exit_info_1; -if (type >3 ) { -kvm_inject_gp (vcpu,0 ); -return 1 ; - } -return kvm_handle_invpcid (vcpu, type, gva); }
arch/x86/kvm/vmx/vmx.c +0 -5
@@ -5505,11 +5505,6 @@ static int handle_invpcid(struct kvm_vcpu *vcpu) gpr_index =vmx_get_instr_info_reg2 (vmx_instruction_info); type =kvm_register_read (vcpu, gpr_index); -if (type >3 ) { -kvm_inject_gp (vcpu,0 ); -return 1 ; - } -/* According to the Intel instruction reference, the memory operand * is read even if it isn't needed (e.g., for type==all) */
arch/x86/kvm/x86.c +2 -1
@@ -12443,7 +12443,8 @@ int kvm_handle_invpcid(struct kvm_vcpu *vcpu, unsigned long type, gva_t gva)return kvm_skip_emulated_instruction (vcpu);default : -BUG ();/* We have already checked above that type <= 3 */ +kvm_inject_gp (vcpu,0 ); +return 1 ; } }EXPORT_SYMBOL_GPL (kvm_handle_invpcid);
Signature · 1 line
2.33.1.1089.g2158813163f-goog
Re: [PATCH v3 2/2] KVM: Move INVPCID type check from vmx and svm to the common kvm_handle_invpcid()
Quoted 10 lines · Vipin Sharma
Handle #GP on INVPCID due to an invalid type in the common switch statement instead of relying on the callers (VMX and SVM) to manually validate the type.
Unlike INVVPID and INVEPT, INVPCID is not explicitly documented to check the type before reading the operand from memory, so deferring the type validity check until after that point is architecturally allowed.
Signed-off-by Vipin Sharma <[email protected]>---
For future reference, a R-b that comes with qualifiers can be carried so long as the issues raised by the reviewer are addressed. Obviously it can be somewhat subjective, but common sense usually goes a long ways, and most reviewers won't be too grumpy about mistakes so long as you had good intentions and remedy any mistakes. And if you're in doubt, you can always add a blurb in the cover letter or ignored part of the patch to explicitly confirm that it was ok to add the tag, e.g. "Sean, I added your Reviewed-by in patch 02 after fixing the changelog, let me know if that's not what you intended".
Thanks!
Re: [PATCH v3 2/2] KVM: Move INVPCID type check from vmx and svm to the common kvm_handle_invpcid()
Quoted 22 lines · Sean Christopherson
Quoted 10 lines · Vipin Sharma
Handle #GP on INVPCID due to an invalid type in the common switch statement instead of relying on the callers (VMX and SVM) to manually validate the type.
Unlike INVVPID and INVEPT, INVPCID is not explicitly documented to check the type before reading the operand from memory, so deferring the type validity check until after that point is architecturally allowed.
Signed-off-by Vipin Sharma <[email protected]>---
For future reference, a R-b that comes with qualifiers can be carried so long as the issues raised by the reviewer are addressed. Obviously it can be somewhat subjective, but common sense usually goes a long ways, and most reviewers won't be too grumpy about mistakes so long as you had good intentions and remedy any mistakes. And if you're in doubt, you can always add a blurb in the cover letter or ignored part of the patch to explicitly confirm that it was ok to add the tag, e.g. "Sean, I added your Reviewed-by in patch 02 after fixing the changelog, let me know if that's not what you intended".
Quoted 3 lines
Thanks!
Reviewed-by Sean Christopherson <[email protected]>
I was not sure if I can add R-b as it was only for the code and not changelog. Good to know that I can ask such things in the cover letter or the ignored part of the patch.
Thanks
Vipin
Sean Christopherson
maintainerRe: [PATCH v3 2/2] KVM: Move INVPCID type check from vmx and svm to the common kvm_handle_invpcid()
Quoted 23 lines · Vipin Sharma
Quoted 22 lines · Sean Christopherson
Quoted 10 lines · Vipin Sharma
Handle #GP on INVPCID due to an invalid type in the common switch statement instead of relying on the callers (VMX and SVM) to manually validate the type.
Unlike INVVPID and INVEPT, INVPCID is not explicitly documented to check the type before reading the operand from memory, so deferring the type validity check until after that point is architecturally allowed.
Signed-off-by Vipin Sharma <[email protected]>---
For future reference, a R-b that comes with qualifiers can be carried so long as the issues raised by the reviewer are addressed. Obviously it can be somewhat subjective, but common sense usually goes a long ways, and most reviewers won't be too grumpy about mistakes so long as you had good intentions and remedy any mistakes. And if you're in doubt, you can always add a blurb in the cover letter or ignored part of the patch to explicitly confirm that it was ok to add the tag, e.g. "Sean, I added your Reviewed-by in patch 02 after fixing the changelog, let me know if that's not what you intended".
Quoted 7 lines
Quoted 3 lines
Thanks!
Reviewed-by Sean Christopherson <[email protected]>I was not sure if I can add R-b as it was only for the code and not changelog. Good to know that I can ask such things in the cover letter or the ignored part of the patch.
Ah, that's my bad, that was indeed a very confusing way to phrase my contingent review.
Sean Christopherson
maintainerQuoted 4 lines · Vipin Sharma
Hello,
v3 is similar to v2 except that the commit message of "PATCH v3 2/2" is now clearer and detailed.
Heh, please wait at _minimum_ one day before spinning a new version. I know it's a bit weird/silly for such a small series, but even in this case I replied to the previous version because I circled back to the "series" while waiting for a build to complete. For small series and/or single patches, unless there's a reason for urgency, it's polite to wait a few days between versions to give folks a reasonable chance to weigh in before getting hit with a new version.
Quoted 7 lines · Sean Christopherson
Quoted 4 lines · Vipin Sharma
Hello,
v3 is similar to v2 except that the commit message of "PATCH v3 2/2" is now clearer and detailed.
Quoted 6 lines
Heh, please wait at _minimum_ one day before spinning a new version. I know it's a bit weird/silly for such a small series, but even in this case I replied to the previous version because I circled back to the "series" while waiting for a build to complete. For small series and/or single patches, unless there's a reason for urgency, it's polite to wait a few days between versions to give folks a reasonable chance to weigh in before getting hit with a new version.
I am sorry about this. I will keep this in mind for any future work. Thanks for the advice, I appreciate it.