x86/KVM/svm: get rid of hardcoded instructions lengths
cover x86/KVM/svm: get rid of hardcoded instructions lengths
Jim rightfully complains that hardcoding instuctions lengths is not always
correct: additional (redundant) prefixes can be used. Luckily, the ugliness is mostly harmless: modern AMD CPUs support NRIP_SAVE feature but I'd like to clean things up and sacrifice speed in favor of correctness.
Early RFC. Unfortunately, I got distracted by some other problems so sending it out half-baked.
TODO:
- Get rid of hardcoded '+ 3' in vmrun_interception().
- Test.
P.S. If you'd like to test the series you'll have to have a CPU without NRIP_SAVE feature or forcefully disable it, something like:
index 8d4e50428b68..93c7eaad7915 100644
arch/x86/kernel/cpu/amd.c +3 -0
@@ -922,6 +922,9 @@ static void init_amd(struct cpuinfo_x86 *c)/* AMD CPUs don't reset SS attributes on SYSRET, Xen does. */ if (!cpu_has (c,X86_FEATURE_XENPV ))set_cpu_bug (c,X86_BUG_SYSRET_SS_ATTRS ); + +/* No nrips */ +clear_cpu_cap (c,X86_FEATURE_NRIPS ); }#ifdef CONFIG_X86_32
arch/x86/kvm/svm.c +2 -2
@@ -773,11 +773,11 @@ static void skip_emulated_instruction(struct kvm_vcpu *vcpu) {struct vcpu_svm *svm =to_svm (vcpu); -if (svm->vmcb->control.next_rip !=0 ) { +/* if (svm->vmcb->control.next_rip != 0) { WARN_ON_ONCE(!static_cpu_has(X86_FEATURE_NRIPS)); svm->next_rip = svm->vmcb->control.next_rip; } - +*/ if (!svm->next_rip) {if (kvm_emulate_instruction (vcpu,EMULTYPE_SKIP ) !=EMULATE_DONE )
Vitaly Kuznetsov (5):
x86: KVM: svm: don't pretend to advance RIP in case
wrmsr_interception() results in #GP
x86: KVM: svm: avoid flooding logs when skip_emulated_instruction()
fails
x86: KVM: svm: clear interrupt shadow on all paths in
skip_emulated_instruction()
x86: KVM: add xsetbv to the emulator
x86: KVM: svm: remove hardcoded instruction length from intercepts
arch/x86/include/asm/kvm_emulate.h | 1 + arch/x86/kvm/emulate.c | 9 ++++++++- arch/x86/kvm/svm.c | 19 ++++++------------- 3 files changed, 15 insertions(+), 14 deletions(-)
Signature · 1 line
2.20.1
RFC PATCH 1/5 x86: KVM: svm: don't pretend to advance RIP in case wrmsr_interception() results in #GP
svm->next_rip is only used by skip_emulated_instruction() and in case kvm_set_msr() fails we rightfully don't do that. Move svm->next_rip advancement to 'else' branch to avoid creating false impression that it's always advanced.
By the way, rdmsr_interception() has it right already.
--- arch/x86/kvm/svm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kvm/svm.c +1 -1
@@ -4430,13 +4430,13 @@ static int wrmsr_interception(struct vcpu_svm *svm) msr.index = ecx; msr.host_initiated = false; - svm->next_rip =kvm_rip_read (&svm->vcpu) +2 ;if (kvm_set_msr (&svm->vcpu, &msr)) {trace_kvm_msr_write_ex (ecx, data);kvm_inject_gp (&svm->vcpu,0 );return 1 ; }else {trace_kvm_msr_write (ecx, data); + svm->next_rip =kvm_rip_read (&svm->vcpu) +2 ;return kvm_skip_emulated_instruction (&svm->vcpu); } }
Signature · 1 line
2.20.1
Re: [PATCH RFC 1/5] x86: KVM: svm: don't pretend to advance RIP in case wrmsr_interception() results in #GP
Quoted 7 lines · Vitaly Kuznetsov
svm->next_ripis only used byskip_emulated_instruction()and in casekvm_set_msr()fails we rightfully don't do that. Movesvm->next_ripadvancement to 'else' branch to avoid creating false impression that it's always advanced.By the way,
rdmsr_interception()has it right already.
I think I actually prefer the current placement, because this allows the code that's common to both kvm-amd.ko and kvm-intel.ko to be hoisted into the vendor-agnostic kvm module. Also, this hard-coded '2' should be going away, right?
Re: [PATCH RFC 1/5] x86: KVM: svm: don't pretend to advance RIP in case wrmsr_interception() results in #GP
Quoted 7 lines · Jim Mattson
Quoted 6 lines · Vitaly Kuznetsov
svm->next_ripis only used byskip_emulated_instruction()and in casekvm_set_msr()fails we rightfully don't do that. Movesvm->next_ripadvancement to 'else' branch to avoid creating false impression that it's always advanced.
Quoted 6 lines
Quoted 1 line
By the way,
rdmsr_interception()has it right already.I think I actually prefer the current placement, because this allows the code that's common to both kvm-amd.ko and kvm-intel.ko to be hoisted into the vendor-agnostic kvm module. Also, this hard-coded '2' should be going away, right?
This whole change goes away in PATCH5 (with hardcoded '+2'), I added this patch just to make it clear that RIP advancement we're doing here is only being used by kvm_skip_emulated_instruction() and kvm_inject_gp() branch is not affected.
We can throw this patch away from the series.
Signature · 1 line
Vitaly
RFC PATCH 2/5 x86: KVM: svm: avoid flooding logs when skip_emulated_instruction() fails
In reply to Vitaly KuznetsovWhen we're unable to skip instruction with kvm_emulate_instruction() we will not advance RIP and most likely the guest will get stuck as consequitive attempts to execute the same instruction will likely result in the same behavior.
As we're not supposed to see these messages under normal conditions, switch to pr_err_once().
--- arch/x86/kvm/svm.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
arch/x86/kvm/svm.c +2 -1
@@ -781,7 +781,8 @@ static void skip_emulated_instruction(struct kvm_vcpu *vcpu)if (!svm->next_rip) {if (kvm_emulate_instruction (vcpu,EMULTYPE_SKIP ) !=EMULATE_DONE ) -printk (KERN_DEBUG "%s: NOP\n" , __func__); +pr_err_once ("KVM: %s: unable to skip instruction\n" , + __func__);return ; }if (svm->next_rip -kvm_rip_read (vcpu) >MAX_INST_SIZE )
Signature · 1 line
2.20.1
Re: [PATCH RFC 2/5] x86: KVM: svm: avoid flooding logs when skip_emulated_instruction() fails
Quoted 10 lines · Vitaly Kuznetsov
When we're unable to skip instruction with
kvm_emulate_instruction()we will not advance RIP and most likely the guest will get stuck as consequitive attempts to execute the same instruction will likely result in the same behavior.As we're not supposed to see these messages under normal conditions, switch to
pr_err_once().Signed-off-by Vitaly Kuznetsov <[email protected]>
RFC PATCH 3/5 x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
In reply to Vitaly KuznetsovRegardless of the way how we skip instruction, interrupt shadow needs to be cleared.
--- arch/x86/kvm/svm.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
arch/x86/kvm/svm.c +3 -1
@@ -783,13 +783,15 @@ static void skip_emulated_instruction(struct kvm_vcpu *vcpu) EMULATE_DONE) pr_err_once("KVM: %s: unable to skip instruction\n", __func__); -return ; + goto clear_int_shadow; }if (svm->next_rip -kvm_rip_read (vcpu) >MAX_INST_SIZE )printk (KERN_ERR "%s: ip 0x%lx next 0x%llx\n" , __func__,kvm_rip_read (vcpu), svm->next_rip);kvm_rip_write (vcpu, svm->next_rip); + +clear_int_shadow:svm_set_interrupt_shadow (vcpu,0 ); }
Signature · 1 line
2.20.1
Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 3 lines · Vitaly Kuznetsov
Regardless of the way how we skip instruction, interrupt shadow needs to be cleared.
This change is definitely an improvement, but the existing code seems to assume that we never call skip_emulated_instruction on a POP-SS/MOV-to-SS/STI. Is that enforced anywhere?
Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 8 lines · Jim Mattson
Quoted 3 lines · Vitaly Kuznetsov
Regardless of the way how we skip instruction, interrupt shadow needs to be cleared.
This change is definitely an improvement, but the existing code seems to assume that we never call
skip_emulated_instructionon a POP-SS/MOV-to-SS/STI. Is that enforced anywhere?
Hm, good question. I'll try to check before v1.
Signature · 1 line
Vitaly
Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
In reply to Jim MattsonQuoted 8 lines · Jim Mattson
Quoted 3 lines · Vitaly Kuznetsov
Regardless of the way how we skip instruction, interrupt shadow needs to be cleared.
This change is definitely an improvement, but the existing code seems to assume that we never call
skip_emulated_instructionon a POP-SS/MOV-to-SS/STI. Is that enforced anywhere?
(before I send v1 of the series) I looked at the current code and I don't think it is enforced, however, VMX version does the same and honestly I can't think of a situation when we would be doing 'skip' for such an instruction.... and there's nothing we can easily enforce from skip_emulated_instruction() as we have no idea what the instruction is...
I can of course be totally wrong and would appreciate if someone more knowledgeable chimes in :-)
Signature · 1 line
Vitaly
Paolo Bonzini
maintainerRe: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 11 lines · Vitaly Kuznetsov
Quoted 8 lines · Jim Mattson
Quoted 3 lines · Vitaly Kuznetsov
Regardless of the way how we skip instruction, interrupt shadow needs to be cleared.
This change is definitely an improvement, but the existing code seems to assume that we never call
skip_emulated_instructionon a POP-SS/MOV-to-SS/STI. Is that enforced anywhere?
Quoted 6 lines
(before I send v1 of the series) I looked at the current code and I don't think it is enforced, however, VMX version does the same and honestly I can't think of a situation when we would be doing 'skip' for such an instruction.... and there's nothing we can easily enforce from
skip_emulated_instruction()as we have no idea what the instruction is...
I agree, I think a comment is worthwhile but we can live with the limitation.
Paolo
Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 13 lines · Paolo Bonzini
Quoted 11 lines · Vitaly Kuznetsov
Quoted 8 lines · Jim Mattson
Quoted 3 lines · Vitaly Kuznetsov
Regardless of the way how we skip instruction, interrupt shadow needs to be cleared.
This change is definitely an improvement, but the existing code seems to assume that we never call
skip_emulated_instructionon a POP-SS/MOV-to-SS/STI. Is that enforced anywhere?
Quoted 6 lines
Quoted 6 lines
(before I send v1 of the series) I looked at the current code and I don't think it is enforced, however, VMX version does the same and honestly I can't think of a situation when we would be doing 'skip' for such an instruction.... and there's nothing we can easily enforce from
skip_emulated_instruction()as we have no idea what the instruction is...
Can't we still coerce kvm into emulating any instruction by leveraging a stale ITLB entry? The 'emulator' kvm-unit-test did this before the KVM forced emulation prefix was introduced, but I haven't checked to see if the original (admittedly fragile) approach still works. Also, for POP-SS, you could always force emulation by mapping the %rsp address beyond guest physical memory. The hypervisor would then have to emulate the instruction to provide bus-error semantics.
Quoted 2 lines
I agree, I think a comment is worthwhile but we can live with the limitation.
I think we can live with the limitation, but I'd really prefer to see a KVM exit with KVM_INTERNAL_ERROR_EMULATION for an instruction that kvm doesn't emulate properly. That seems better than just a comment that the virtual CPU doesn't behave as architected. (I realize that I am probably in the minority here.)
Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 29 lines · Jim Mattson
Quoted 13 lines · Paolo Bonzini
Quoted 11 lines · Vitaly Kuznetsov
Quoted 8 lines · Jim Mattson
Quoted 3 lines · Vitaly Kuznetsov
Regardless of the way how we skip instruction, interrupt shadow needs to be cleared.
This change is definitely an improvement, but the existing code seems to assume that we never call
skip_emulated_instructionon a POP-SS/MOV-to-SS/STI. Is that enforced anywhere?Quoted 6 lines
Quoted 6 lines
(before I send v1 of the series) I looked at the current code and I don't think it is enforced, however, VMX version does the same and honestly I can't think of a situation when we would be doing 'skip' for such an instruction.... and there's nothing we can easily enforce from
skip_emulated_instruction()as we have no idea what the instruction is...Can't we still coerce kvm into emulating any instruction by leveraging a stale ITLB entry? The 'emulator' kvm-unit-test did this before the KVM forced emulation prefix was introduced, but I haven't checked to see if the original (admittedly fragile) approach still works. Also, for POP-SS, you could always force emulation by mapping the %rsp address beyond guest physical memory. The hypervisor would then have to emulate the instruction to provide bus-error semantics.
Quoted 8 lines
Quoted 2 lines
I agree, I think a comment is worthwhile but we can live with the limitation.
I think we can live with the limitation, but I'd really prefer to see a KVM exit with
KVM_INTERNAL_ERROR_EMULATIONfor an instruction that kvm doesn't emulate properly. That seems better than just a comment that the virtual CPU doesn't behave as architected. (I realize that I am probably in the minority here.)
At a glance, the full emulator models behavior correctly, e.g. see toggle_interruptibility() and setters of ctxt->interruptibility.
I'm pretty sure that leaves the EPT misconfig MMIO and APIC access EOI fast paths as the only (VMX) path that would incorrectly handle a MOV/POP SS. Reading the guest's instruction stream to detect MOV/POP SS would defeat the whole "fast path" thing, not to mention both paths aren't exactly architecturally compliant in the first place.
Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 8 lines · Sean Christopherson
At a glance, the full emulator models behavior correctly, e.g. see
toggle_interruptibility()and setters ofctxt->interruptibility.I'm pretty sure that leaves the EPT misconfig MMIO and APIC access EOI fast paths as the only (VMX) path that would incorrectly handle a MOV/POP SS. Reading the guest's instruction stream to detect MOV/POP SS would defeat the whole "fast path" thing, not to mention both paths aren't exactly architecturally compliant in the first place.
The proposed patch clears the interrupt shadow in the VMCB on all paths through svm's skip_emulated_instruction. If this happens at the tail end of emulation, it doesn't matter if the full emulator does the right thing.
Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 12 lines · Jim Mattson
Quoted 8 lines · Sean Christopherson
At a glance, the full emulator models behavior correctly, e.g. see
toggle_interruptibility()and setters ofctxt->interruptibility.I'm pretty sure that leaves the EPT misconfig MMIO and APIC access EOI fast paths as the only (VMX) path that would incorrectly handle a MOV/POP SS. Reading the guest's instruction stream to detect MOV/POP SS would defeat the whole "fast path" thing, not to mention both paths aren't exactly architecturally compliant in the first place.
Quoted 4 lines
The proposed patch clears the interrupt shadow in the VMCB on all paths through svm's
skip_emulated_instruction. If this happens at the tail end of emulation, it doesn't matter if the full emulator does the right thing.
Unless I'm missing something, skip_emulated_instruction() isn't called in the emulation case, x86_emulate_instruction() updates %rip directly, e.g.:
if (writeback) {
unsigned long rflags = kvm_x86_ops->get_rflags(vcpu);
toggle_interruptibility(vcpu, ctxt->interruptibility);
vcpu->arch.emulate_regs_need_sync_to_vcpu = false;
kvm_rip_write(vcpu, ctxt->eip);
if (r == EMULATE_DONE && ctxt->tf)
kvm_vcpu_do_singlestep(vcpu, &r);
if (!ctxt->have_exception ||
exception_type(ctxt->exception.vector) == EXCPT_TRAP)
__kvm_set_rflags(vcpu, ctxt->eflags);
/*
* For STI, interrupts are shadowed; so KVM_REQ_EVENT will
* do nothing, and it will be requested again as soon as
* the shadow expires. But we still need to check here,
* because POPF has no interrupt shadow.
*/
if (unlikely((ctxt->eflags & ~rflags) & X86_EFLAGS_IF))
kvm_make_request(KVM_REQ_EVENT, vcpu);
}Paolo Bonzini
maintainerRe: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 13 lines · Sean Christopherson
Quoted 12 lines · Jim Mattson
Quoted 8 lines · Sean Christopherson
At a glance, the full emulator models behavior correctly, e.g. see
toggle_interruptibility()and setters ofctxt->interruptibility.I'm pretty sure that leaves the EPT misconfig MMIO and APIC access EOI fast paths as the only (VMX) path that would incorrectly handle a MOV/POP SS. Reading the guest's instruction stream to detect MOV/POP SS would defeat the whole "fast path" thing, not to mention both paths aren't exactly architecturally compliant in the first place.
Quoted 7 lines
Quoted 4 lines
The proposed patch clears the interrupt shadow in the VMCB on all paths through svm's
skip_emulated_instruction. If this happens at the tail end of emulation, it doesn't matter if the full emulator does the right thing.Unless I'm missing something,
skip_emulated_instruction()isn't called in the emulation case,x86_emulate_instruction()updates %rip directly, e.g.:
Indeed. skip_emulated_instruction() is only used when the vmexit code takes care of emulation directly.
Paolo
Quoted 21 lines
if (writeback) {
unsigned long rflags = kvm_x86_ops->get_rflags(vcpu);
toggle_interruptibility(vcpu, ctxt->interruptibility);
vcpu->arch.emulate_regs_need_sync_to_vcpu = false;
kvm_rip_write(vcpu, ctxt->eip);
if (r == EMULATE_DONE && ctxt->tf)
kvm_vcpu_do_singlestep(vcpu, &r);
if (!ctxt->have_exception ||
exception_type(ctxt->exception.vector) == EXCPT_TRAP)
__kvm_set_rflags(vcpu, ctxt->eflags);
/*
* For STI, interrupts are shadowed; so KVM_REQ_EVENT will
* do nothing, and it will be requested again as soon as
* the shadow expires. But we still need to check here,
* because POPF has no interrupt shadow.
*/
if (unlikely((ctxt->eflags & ~rflags) & X86_EFLAGS_IF))
kvm_make_request(KVM_REQ_EVENT, vcpu);
}Re: [PATCH RFC 3/5] x86: KVM: svm: clear interrupt shadow on all paths in skip_emulated_instruction()
Quoted 20 lines · Paolo Bonzini
Quoted 18 lines · Sean Christopherson
Quoted 16 lines · Jim Mattson
Quoted 8 lines · Sean Christopherson
At a glance, the full emulator models behavior correctly, e.g. see
toggle_interruptibility()and setters ofctxt->interruptibility.I'm pretty sure that leaves the EPT misconfig MMIO and APIC access EOI fast paths as the only (VMX) path that would incorrectly handle a MOV/POP SS. Reading the guest's instruction stream to detect MOV/POP SS would defeat the whole "fast path" thing, not to mention both paths aren't exactly architecturally compliant in the first place.
The proposed patch clears the interrupt shadow in the VMCB on all paths through svm's
skip_emulated_instruction. If this happens at the tail end of emulation, it doesn't matter if the full emulator does the right thing.
Quoted 5 lines
Quoted 2 lines
Unless I'm missing something,
skip_emulated_instruction()isn't called in the emulation case,x86_emulate_instruction()updates %rip directly, e.g.:Indeed.
skip_emulated_instruction()is only used when the vmexit code takes care of emulation directly.
Mea culpa. I had incorrectly assumed that "skip_emulated_instruction" was used when an instruction was emulated. I retract my objection. Having now been twice bitten by misleading function names, I'll be more careful in the future.
RFC PATCH 4/5 x86: KVM: add xsetbv to the emulator
In reply to Vitaly KuznetsovTo avoid hardcoding xsetbv length to '3' we need to support decoding it in the emulator.
--- arch/x86/include/asm/kvm_emulate.h | 1 + arch/x86/kvm/emulate.c | 9 ++++++++- arch/x86/kvm/svm.c | 1 + 3 files changed, 10 insertions(+), 1 deletion(-)
arch/x86/include/asm/kvm_emulate.h +1 -0
@@ -429,6 +429,7 @@ enum x86_intercept { x86_intercept_ins, x86_intercept_out, x86_intercept_outs, + x86_intercept_xsetbv, nr_x86_intercepts };
arch/x86/kvm/emulate.c +8 -1
@@ -4393,6 +4393,12 @@ static const struct opcode group7_rm1[] = { N, N, N, N, N, N, }; +static const struct opcode group7_rm2[] = { +N , +DI (SrcNone | Priv, xsetbv), +N ,N ,N ,N ,N ,N , +}; +static const struct opcode group7_rm3[] = {DIP (SrcNone | Prot | Priv, vmrun, check_svme_pa),II (SrcNone | Prot | EmulateOnUD, em_hypercall, vmmcall),
@@ -4482,7 +4488,8 @@ static const struct group_dual group7 = { { }, {EXT (0 , group7_rm0),EXT (0 , group7_rm1), -N ,EXT (0 , group7_rm3), +EXT (0 , group7_rm2), +EXT (0 , group7_rm3),II (SrcNone | DstMem | Mov, em_smsw, smsw),N ,II (SrcMem16 | Mov | Priv, em_lmsw, lmsw), EXT(0, group7_rm7),
arch/x86/kvm/svm.c +1 -0
@@ -6041,6 +6041,7 @@ static const struct __x86_intercept { [x86_intercept_ins] = POST_EX(SVM_EXIT_IOIO), [x86_intercept_out] = POST_EX(SVM_EXIT_IOIO), [x86_intercept_outs] = POST_EX(SVM_EXIT_IOIO), + [x86_intercept_xsetbv] = PRE_EX(SVM_EXIT_XSETBV), };#undef PRE_EX
Signature · 1 line
2.20.1
Paolo Bonzini
maintainerRe: [PATCH RFC 4/5] x86: KVM: add xsetbv to the emulator
Quoted 4 lines · Vitaly Kuznetsov
To avoid hardcoding xsetbv length to '3' we need to support decoding it in the emulator.
Signed-off-by Vitaly Kuznetsov <[email protected]>
Can you also emulate it properly? The code from QEMU's target/i386/fpu_helper.c can help. :)
Paolo
Quoted 58 lines
--- arch/x86/include/asm/kvm_emulate.h | 1 + arch/x86/kvm/emulate.c | 9 ++++++++- arch/x86/kvm/svm.c | 1 + 3 files changed, 10 insertions(+), 1 deletion(-)arch/x86/include/asm/kvm_emulate.h +1
-0@@ -429,6 +429,7 @@ enum x86_intercept { x86_intercept_ins, x86_intercept_out, x86_intercept_outs, + x86_intercept_xsetbv, nr_x86_intercepts };arch/x86/kvm/emulate.c +8
-1@@ -4393,6 +4393,12 @@ static const struct opcode group7_rm1[] = { N, N, N, N, N, N, }; +static const struct opcode group7_rm2[] = { +N , +DI (SrcNone | Priv, xsetbv), +N ,N ,N ,N ,N ,N , +}; +static const struct opcode group7_rm3[] = {DIP (SrcNone | Prot | Priv, vmrun, check_svme_pa),II (SrcNone | Prot | EmulateOnUD, em_hypercall, vmmcall),@@ -4482,7 +4488,8 @@ static const struct group_dual group7 = { { }, {EXT (0 , group7_rm0),EXT (0 , group7_rm1), -N ,EXT (0 , group7_rm3), +EXT (0 , group7_rm2), +EXT (0 , group7_rm3),II (SrcNone | DstMem | Mov, em_smsw, smsw),N ,II (SrcMem16 | Mov | Priv, em_lmsw, lmsw), EXT(0, group7_rm7),arch/x86/kvm/svm.c +1
-0@@ -6041,6 +6041,7 @@ static const struct __x86_intercept { [x86_intercept_ins] = POST_EX(SVM_EXIT_IOIO), [x86_intercept_out] = POST_EX(SVM_EXIT_IOIO), [x86_intercept_outs] = POST_EX(SVM_EXIT_IOIO), + [x86_intercept_xsetbv] = PRE_EX(SVM_EXIT_XSETBV), };#undef PRE_EX
Re: [PATCH RFC 4/5] x86: KVM: add xsetbv to the emulator
Quoted 9 lines · Paolo Bonzini
Quoted 4 lines · Vitaly Kuznetsov
To avoid hardcoding xsetbv length to '3' we need to support decoding it in the emulator.
Signed-off-by Vitaly Kuznetsov <[email protected]>Can you also emulate it properly? The code from QEMU's
target/i386/fpu_helper.ccan help. :)
(Had a chance to get back to this just now, sorry)
Assuming __kvm_set_xcr() is also a correct implementation, would the code below do the job? (Just trying to figure out why you suggested me to take a look at QEMU's variant):
arch/x86/include/asm/kvm_emulate.h +2 -1
@@ -229,7 +229,7 @@ struct x86_emulate_ops {int (*pre_leave_smm)(struct x86_emulate_ctxt *ctxt,const char *smstate);void (*post_leave_smm)(struct x86_emulate_ctxt *ctxt); - +int (*set_xcr)(struct x86_emulate_ctxt *ctxt,u32 index,u64 xcr); }; typedef u32 __attribute__((vector_size(16))) sse128_t;
@@ -429,6 +429,7 @@ enum x86_intercept { x86_intercept_ins, x86_intercept_out, x86_intercept_outs, + x86_intercept_xsetbv, nr_x86_intercepts };
arch/x86/kvm/emulate.c +22 -1
@@ -4156,6 +4156,20 @@ static int em_fxrstor(struct x86_emulate_ctxt *ctxt)return rc; } +static int em_xsetbv (struct x86_emulate_ctxt *ctxt) +{ +u32 eax, ecx, edx; + + eax =reg_read (ctxt,VCPU_REGS_RAX ); + edx =reg_read (ctxt,VCPU_REGS_RDX ); + ecx =reg_read (ctxt,VCPU_REGS_RCX ); + +if (ctxt->ops->set_xcr (ctxt, ecx, ((u64 )edx <<32 ) | eax)) +return emulate_gp (ctxt,0 ); + +return X86EMUL_CONTINUE ; +} +static bool valid_cr (int nr) {switch (nr) {
@@ -4409,6 +4423,12 @@ static const struct opcode group7_rm1[] = { N, N, N, N, N, N, }; +static const struct opcode group7_rm2[] = { +N , +II (ImplicitOps | Priv, em_xsetbv, xsetbv), +N ,N ,N ,N ,N ,N , +}; +static const struct opcode group7_rm3[] = {DIP (SrcNone | Prot | Priv, vmrun, check_svme_pa),II (SrcNone | Prot | EmulateOnUD, em_hypercall, vmmcall),
@@ -4498,7 +4518,8 @@ static const struct group_dual group7 = { { }, {EXT (0 , group7_rm0),EXT (0 , group7_rm1), -N ,EXT (0 , group7_rm3), +EXT (0 , group7_rm2), +EXT (0 , group7_rm3),II (SrcNone | DstMem | Mov, em_smsw, smsw),N ,II (SrcMem16 | Mov | Priv, em_lmsw, lmsw), EXT(0, group7_rm7),
arch/x86/kvm/x86.c +6 -0
@@ -6068,6 +6068,11 @@ static void emulator_post_leave_smm(struct x86_emulate_ctxt *ctxt)kvm_smm_changed (emul_to_vcpu (ctxt)); } +static int emulator_set_xcr (struct x86_emulate_ctxt *ctxt,u32 index,u64 xcr) +{ +return __kvm_set_xcr (emul_to_vcpu (ctxt), index, xcr); +} +static const struct x86_emulate_ops emulate_ops = { .read_gpr = emulator_read_gpr, .write_gpr = emulator_write_gpr,
@@ -6109,6 +6114,7 @@ static const struct x86_emulate_ops emulate_ops = { .set_hflags = emulator_set_hflags, .pre_leave_smm = emulator_pre_leave_smm, .post_leave_smm = emulator_post_leave_smm, + .set_xcr = emulator_set_xcr, };static void toggle_interruptibility (struct kvm_vcpu *vcpu,u32 mask)
Signature · 1 line
Vitaly
Paolo Bonzini
maintainerRe: [PATCH RFC 4/5] x86: KVM: add xsetbv to the emulator
Quoted 12 lines · Vitaly Kuznetsov
Quoted 9 lines · Paolo Bonzini
Quoted 4 lines · Vitaly Kuznetsov
To avoid hardcoding xsetbv length to '3' we need to support decoding it in the emulator.
Signed-off-by Vitaly Kuznetsov <[email protected]>Can you also emulate it properly? The code from QEMU's
target/i386/fpu_helper.ccan help. :)
Quoted 6 lines
(Had a chance to get back to this just now, sorry)
Assuming
__kvm_set_xcr()is also a correct implementation, would the code below do the job? (Just trying to figure out why you suggested me to take a look at QEMU's variant):
Yes, I didn't remember __kvm_set_xcr.
Paolo
Quoted 94 lines
arch/x86/include/asm/kvm_emulate.h +2
-1@@ -229,7 +229,7 @@ struct x86_emulate_ops {int (*pre_leave_smm)(struct x86_emulate_ctxt *ctxt,const char *smstate);void (*post_leave_smm)(struct x86_emulate_ctxt *ctxt); - +int (*set_xcr)(struct x86_emulate_ctxt *ctxt,u32 index,u64 xcr); }; typedef u32 __attribute__((vector_size(16))) sse128_t;@@ -429,6 +429,7 @@ enum x86_intercept { x86_intercept_ins, x86_intercept_out, x86_intercept_outs, + x86_intercept_xsetbv, nr_x86_intercepts };arch/x86/kvm/emulate.c +22
-1@@ -4156,6 +4156,20 @@ static int em_fxrstor(struct x86_emulate_ctxt *ctxt)return rc; } +static int em_xsetbv (struct x86_emulate_ctxt *ctxt) +{ +u32 eax, ecx, edx; + + eax =reg_read (ctxt,VCPU_REGS_RAX ); + edx =reg_read (ctxt,VCPU_REGS_RDX ); + ecx =reg_read (ctxt,VCPU_REGS_RCX ); + +if (ctxt->ops->set_xcr (ctxt, ecx, ((u64 )edx <<32 ) | eax)) +return emulate_gp (ctxt,0 ); + +return X86EMUL_CONTINUE ; +} +static bool valid_cr (int nr) {switch (nr) {@@ -4409,6 +4423,12 @@ static const struct opcode group7_rm1[] = { N, N, N, N, N, N, }; +static const struct opcode group7_rm2[] = { +N , +II (ImplicitOps | Priv, em_xsetbv, xsetbv), +N ,N ,N ,N ,N ,N , +}; +static const struct opcode group7_rm3[] = {DIP (SrcNone | Prot | Priv, vmrun, check_svme_pa),II (SrcNone | Prot | EmulateOnUD, em_hypercall, vmmcall),@@ -4498,7 +4518,8 @@ static const struct group_dual group7 = { { }, {EXT (0 , group7_rm0),EXT (0 , group7_rm1), -N ,EXT (0 , group7_rm3), +EXT (0 , group7_rm2), +EXT (0 , group7_rm3),II (SrcNone | DstMem | Mov, em_smsw, smsw),N ,II (SrcMem16 | Mov | Priv, em_lmsw, lmsw), EXT(0, group7_rm7),arch/x86/kvm/x86.c +6
-0@@ -6068,6 +6068,11 @@ static void emulator_post_leave_smm(struct x86_emulate_ctxt *ctxt)kvm_smm_changed (emul_to_vcpu (ctxt)); } +static int emulator_set_xcr (struct x86_emulate_ctxt *ctxt,u32 index,u64 xcr) +{ +return __kvm_set_xcr (emul_to_vcpu (ctxt), index, xcr); +} +static const struct x86_emulate_ops emulate_ops = { .read_gpr = emulator_read_gpr, .write_gpr = emulator_write_gpr,@@ -6109,6 +6114,7 @@ static const struct x86_emulate_ops emulate_ops = { .set_hflags = emulator_set_hflags, .pre_leave_smm = emulator_pre_leave_smm, .post_leave_smm = emulator_post_leave_smm, + .set_xcr = emulator_set_xcr, };static void toggle_interruptibility (struct kvm_vcpu *vcpu,u32 mask)
RFC PATCH 5/5 x86: KVM: svm: remove hardcoded instruction length from intercepts
In reply to Vitaly KuznetsovVarious intercepts hard-code the respective instruction lengths to optimize skip_emulated_instruction(): when next_rip is pre-set we skip kvm_emulate_instruction(vcpu, EMULTYPE_SKIP). The optimization is, however,
incorrect: different (redundant) prefixes could be used to enlarge the instruction. We can't really avoid decoding.
svm->next_rip is not used when CPU supports 'nrips' (X86_FEATURE_NRIPS)
feature: next RIP is provided in VMCB. The feature is not really new (Opteron G3s had it already) and the change should have zero affect.
Remove manual svm->next_rip setting with hard-coded instruction lengths. The only case where we now use svm->next_rip is EXIT_IOIO: the instruction length is provided to us by hardware.
--- arch/x86/kvm/svm.c | 11 ----------- 1 file changed, 11 deletions(-)
arch/x86/kvm/svm.c +0 -11
@@ -2890,13 +2890,11 @@ static int nop_on_interception(struct vcpu_svm *svm)static int halt_interception (struct vcpu_svm *svm) { - svm->next_rip =kvm_rip_read (&svm->vcpu) +1 ;return kvm_emulate_halt (&svm->vcpu); }static int vmmcall_interception (struct vcpu_svm *svm) { - svm->next_rip =kvm_rip_read (&svm->vcpu) +3 ;return kvm_emulate_hypercall (&svm->vcpu); }
@@ -3684,7 +3682,6 @@ static int vmload_interception(struct vcpu_svm *svm) nested_vmcb = map.hva; - svm->next_rip =kvm_rip_read (&svm->vcpu) +3 ; ret =kvm_skip_emulated_instruction (&svm->vcpu);nested_svm_vmloadsave (nested_vmcb, svm->vmcb);
@@ -3711,7 +3708,6 @@ static int vmsave_interception(struct vcpu_svm *svm) nested_vmcb = map.hva; - svm->next_rip =kvm_rip_read (&svm->vcpu) +3 ; ret =kvm_skip_emulated_instruction (&svm->vcpu);nested_svm_vmloadsave (svm->vmcb, nested_vmcb);
@@ -3762,7 +3758,6 @@ static int stgi_interception(struct vcpu_svm *svm)if (vgif_enabled (svm))clr_intercept (svm,INTERCEPT_STGI ); - svm->next_rip =kvm_rip_read (&svm->vcpu) +3 ; ret =kvm_skip_emulated_instruction (&svm->vcpu);kvm_make_request (KVM_REQ_EVENT , &svm->vcpu);
@@ -3778,7 +3773,6 @@ static int clgi_interception(struct vcpu_svm *svm)if (nested_svm_check_permissions (svm))return 1 ; - svm->next_rip =kvm_rip_read (&svm->vcpu) +3 ; ret =kvm_skip_emulated_instruction (&svm->vcpu);disable_gif (svm);
@@ -3803,7 +3797,6 @@ static int invlpga_interception(struct vcpu_svm *svm)/* Let's treat INVLPGA the same as INVLPG (can be optimized!) */ kvm_mmu_invlpg (vcpu,kvm_rax_read (&svm->vcpu)); - svm->next_rip =kvm_rip_read (&svm->vcpu) +3 ;return kvm_skip_emulated_instruction (&svm->vcpu); }
@@ -3826,7 +3819,6 @@ static int xsetbv_interception(struct vcpu_svm *svm)u32 index =kvm_rcx_read (&svm->vcpu);if (kvm_set_xcr (&svm->vcpu, index, new_bv) ==0 ) { - svm->next_rip =kvm_rip_read (&svm->vcpu) +3 ;return kvm_skip_emulated_instruction (&svm->vcpu); }
@@ -3903,7 +3895,6 @@ static int task_switch_interception(struct vcpu_svm *svm)static int cpuid_interception (struct vcpu_svm *svm) { - svm->next_rip =kvm_rip_read (&svm->vcpu) +2 ;return kvm_emulate_cpuid (&svm->vcpu); }
@@ -4233,7 +4224,6 @@ static int rdmsr_interception(struct vcpu_svm *svm)kvm_rax_write (&svm->vcpu, msr_info.data &0xffffffff );kvm_rdx_write (&svm->vcpu, msr_info.data >>32 ); - svm->next_rip =kvm_rip_read (&svm->vcpu) +2 ;return kvm_skip_emulated_instruction (&svm->vcpu); } }
@@ -4439,7 +4429,6 @@ static int wrmsr_interception(struct vcpu_svm *svm)return 1 ; }else {trace_kvm_msr_write (ecx, data); - svm->next_rip =kvm_rip_read (&svm->vcpu) +2 ;return kvm_skip_emulated_instruction (&svm->vcpu); } }
Signature · 1 line
2.20.1
Re: [PATCH RFC 5/5] x86: KVM: svm: remove hardcoded instruction length from intercepts
Quoted 11 lines · Vitaly Kuznetsov
Various intercepts hard-code the respective instruction lengths to optimize
skip_emulated_instruction(): whennext_ripis pre-set we skipkvm_emulate_instruction(vcpu, EMULTYPE_SKIP). The optimization is, however,
incorrect: different (redundant) prefixes could be used to enlarge the instruction. We can't really avoid decoding.
svm->next_ripis not used when CPU supports 'nrips' (X86_FEATURE_NRIPS)
feature: next RIP is provided in VMCB. The feature is not really new (Opteron G3s had it already) and the change should have zero affect.
Quoted 6 lines
Remove manual
svm->next_ripsetting with hard-coded instruction lengths. The only case where we now usesvm->next_ripisEXIT_IOIO: the instruction length is provided to us by hardware.Reported-by Jim Mattson <[email protected]>Signed-off-by Vitaly Kuznetsov <[email protected]>
Paolo Bonzini
maintainerQuoted 8 lines · Vitaly Kuznetsov
P.S. If you'd like to test the series you'll have to have a CPU withoutNRIP_SAVEfeature or forcefully disable it, something like:index
8d4e50428b68..93c7eaad7915100644
arch/x86/kernel/cpu/amd.c static void init_amd(struct cpuinfo_x86 *c) · line 930
Quoted 9 lines
arch/x86/kernel/cpu/amd.c +3
-0@@ -922,6 +922,9 @@ static void init_amd(struct cpuinfo_x86 *c)/* AMD CPUs don't reset SS attributes on SYSRET, Xen does. */ if (!cpu_has (c,X86_FEATURE_XENPV ))set_cpu_bug (c,X86_BUG_SYSRET_SS_ATTRS ); + +/* No nrips */ +clear_cpu_cap (c,X86_FEATURE_NRIPS ); }#ifdef CONFIG_X86_32
Let's add a module parameter instead. Patch sent (forgot to Cc you).
Paolo
Quoted 17 lines · Paolo Bonzini
Quoted 16 lines · Vitaly Kuznetsov
P.S. If you'd like to test the series you'll have to have a CPU withoutNRIP_SAVEfeature or forcefully disable it, something like:index
8d4e50428b68..93c7eaad7915100644arch/x86/kernel/cpu/amd.c +3
-0@@ -922,6 +922,9 @@ static void init_amd(struct cpuinfo_x86 *c)/* AMD CPUs don't reset SS attributes on SYSRET, Xen does. */ if (!cpu_has (c,X86_FEATURE_XENPV ))set_cpu_bug (c,X86_BUG_SYSRET_SS_ATTRS ); + +/* No nrips */ +clear_cpu_cap (c,X86_FEATURE_NRIPS ); }
Quoted 4 lines
arch/x86/kernel/cpu/amd.c static void init_amd(struct cpuinfo_x86 *c) · line 930
Quoted 1 line
arch/x86/kernel/cpu/amd.c +0
-0#ifdef CONFIG_X86_32 Let's add a module parameter instead. Patch sent (forgot to Cc you).
Sure, I thought I'm the only interested person around but if there's hope for more this definitely sounds like a good idea)
Signature · 1 line
Vitaly