Skip to contentSkip to conversation

KVM: x86/mmu: Page fault handling cleanups

PATCH Applied Sean Christopherson 2 participants 6 messages Sep 23, 2020 – Sep 25, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover KVM: x86/mmu: Page fault handling cleanups

Cleanups for page fault handling that were encountered during early TDX enabling, but are worthwhile on their own. Specifically, patch 4 fixes an issue where KVM doesn't detect a spurious page fault (due to the fault being fixed by a different pCPU+vCPU) and does the full gamut of writing the SPTE, updating stats, and prefetching SPTEs.

Sean Christopherson (4):
KVM: x86/mmu: Return -EIO if page fault returns RET_PF_INVALID
KVM: x86/mmu: Invert RET_PF_* check when falling through to emulation
KVM: x86/mmu: Return unique RET_PF_* values if the fault was fixed
KVM: x86/mmu: Bail early from final #PF handling on spurious faults

 arch/x86/kvm/mmu/mmu.c         | 70 +++++++++++++++++++++-------------
 arch/x86/kvm/mmu/mmutrace.h    | 13 +++----
 arch/x86/kvm/mmu/paging_tmpl.h |  3 ++
 3 files changed, 52 insertions(+), 34 deletions(-)
Signature · 1 line
2.28.0
#1 lore raw

PATCH 1/4 KVM: x86/mmu: Return -EIO if page fault returns RET_PF_INVALID

Exit to userspace with an error if the MMU is buggy and returns RET_PF_INVALID when servicing a page fault. This will allow a future patch to invert the emulation path, i.e. emulate only on RET_PF_EMULATE instead of emulating on anything but RET_PF_RETRY. This technically means that KVM will exit to userspace instead of emulating on RET_PF_INVALID, but practically speaking it's a nop as the MMU never returns RET_PF_INVALID.

---
 arch/x86/kvm/mmu/mmu.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/x86/kvm/mmu/mmu.c +2 -1
index 76c5826e29a2..4a2a8f67a8f7 100644
@@ -5450,7 +5450,8 @@ int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, u64 error_code,
 	if (r == RET_PF_INVALID) {
 		r = kvm_mmu_do_page_fault(vcpu, cr2_or_gpa,
 					  lower_32_bits(error_code), false);
-		WARN_ON(r == RET_PF_INVALID);
+		if (WARN_ON_ONCE(r == RET_PF_INVALID))
+			return -EIO;
 	}
 
 	if (r == RET_PF_RETRY)
Signature · 1 line
2.28.0
#2 lore raw

PATCH 2/4 KVM: x86/mmu: Invert RET_PF_* check when falling through to emulation

In reply to Sean Christopherson

Explicitly check for RET_PF_EMULATE instead of implicitly doing the same by checking for !RET_PF_RETRY (RET_PF_INVALID is handled earlier). This will adding new RET_PF_ types in future patches without breaking the emulation path.

No functional change intended.

---
 arch/x86/kvm/mmu/mmu.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kvm/mmu/mmu.c +2 -2
index 4a2a8f67a8f7..0b27c4caf5e6 100644
@@ -5454,10 +5454,10 @@ int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, u64 error_code,
 			return -EIO;
 	}
 
-	if (r == RET_PF_RETRY)
-		return 1;
 	if (r < 0)
 		return r;
+	if (r != RET_PF_EMULATE)
+		return 1;
 
 	/*
 	 * Before emulating the instruction, check if the error code
Signature · 1 line
2.28.0
#3 lore raw

PATCH 4/4 KVM: x86/mmu: Bail early from final #PF handling on spurious faults

In reply to Sean Christopherson

Detect spurious page faults, e.g. page faults that occur when multiple vCPUs simultaneously access a not-present page, and skip the SPTE write, prefetch, and stats update for spurious faults.

Note, the performance benefits of skipping the write and prefetch are likely negligible, and the false positive stats adjustment is probably lost in the noise. The primary motivation is to play nice with TDX's SEPT in the long term. SEAMCALLs (to program SEPT entries) are quite costly, e.g. thousands of cycles, and a spurious SEPT update will result in a SEAMCALL error (which KVM will ideally treat as fatal).

---
 arch/x86/kvm/mmu/mmu.c         | 17 ++++++++++++++++-
 arch/x86/kvm/mmu/paging_tmpl.h |  3 +++
 2 files changed, 19 insertions(+), 1 deletion(-)
arch/x86/kvm/mmu/mmu.c +16 -1
index 9253bcecbfe3..9bd657a8e78b 100644
@@ -2973,6 +2973,7 @@ static bool kvm_is_mmio_pfn(kvm_pfn_t pfn)
 /* Bits which may be returned by set_spte() */
 #define SET_SPTE_WRITE_PROTECTED_PT	BIT(0)
 #define SET_SPTE_NEED_REMOTE_TLB_FLUSH	BIT(1)
+#define SET_SPTE_SPURIOUS		BIT(2)
 
 static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
 		    unsigned int pte_access, int level,
@@ -3061,7 +3062,9 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
 		spte = mark_spte_for_access_track(spte);
 
 set_pte:
-	if (mmu_spte_update(sptep, spte))
+	if (*sptep == spte)
+		ret |= SET_SPTE_SPURIOUS;
+	else if (mmu_spte_update(sptep, spte))
 		ret |= SET_SPTE_NEED_REMOTE_TLB_FLUSH;
 	return ret;
 }
@@ -3116,6 +3119,15 @@ static int mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
 	if (unlikely(is_mmio_spte(*sptep)))
 		ret = RET_PF_EMULATE;
 
+	/*
+	 * The fault is fully spurious if and only if the new SPTE and old SPTE
+	 * are identical, and emulation is not required.
+	 */
+	if ((set_spte_ret & SET_SPTE_SPURIOUS) && ret == RET_PF_FIXED) {
+		WARN_ON_ONCE(!was_rmapped);
+		return RET_PF_SPURIOUS;
+	}
+
 	pgprintk("%s: setting spte %llx\n", __func__, *sptep);
 	trace_kvm_mmu_set_spte(level, gfn, sptep);
 	if (!was_rmapped && is_large_pte(*sptep))
@@ -3352,6 +3364,9 @@ static int __direct_map(struct kvm_vcpu *vcpu, gpa_t gpa, int write,
 	ret = mmu_set_spte(vcpu, it.sptep, ACC_ALL,
 			   write, level, base_gfn, pfn, prefault,
 			   map_writable);
+	if (ret == RET_PF_SPURIOUS)
+		return ret;
+
 	direct_pte_prefetch(vcpu, it.sptep);
 	++vcpu->stat.pf_fixed;
 	return ret;
arch/x86/kvm/mmu/paging_tmpl.h +3 -0
index 4dd6b1e5b8cf..dda09c27ff25 100644
@@ -711,6 +711,9 @@ static int FNAME(fetch)(struct kvm_vcpu *vcpu, gpa_t addr,
 
 	ret = mmu_set_spte(vcpu, it.sptep, gw->pte_access, write_fault,
 			   it.level, base_gfn, pfn, prefault, map_writable);
+	if (ret == RET_PF_SPURIOUS)
+		return ret;
+
 	FNAME(pte_prefetch)(vcpu, gw, it.sptep);
 	++vcpu->stat.pf_fixed;
 	return ret;
Signature · 1 line
2.28.0
maintainer Applied
#4 lore raw
In reply to Sean Christopherson
Quoted 12 lines · Sean Christopherson

Cleanups for page fault handling that were encountered during early TDX enabling, but are worthwhile on their own. Specifically, patch 4 fixes an issue where KVM doesn't detect a spurious page fault (due to the fault being fixed by a different pCPU+vCPU) and does the full gamut of writing the SPTE, updating stats, and prefetching SPTEs.

Sean Christopherson (4):
KVM: x86/mmu: Return -EIO if page fault returns RET_PF_INVALID
KVM: x86/mmu: Invert RET_PF_* check when falling through to emulation
KVM: x86/mmu: Return unique RET_PF_* values if the fault was fixed
KVM: x86/mmu: Bail early from final #PF handling on spurious faults

Quoted 5 lines
 arch/x86/kvm/mmu/mmu.c         | 70 +++++++++++++++++++++-------------
 arch/x86/kvm/mmu/mmutrace.h    | 13 +++----
 arch/x86/kvm/mmu/paging_tmpl.h |  3 ++
 3 files changed, 52 insertions(+), 34 deletions(-)

Queued, thanks. Looking at the KVM_BUG_ON now since patch 1 is somewhat related.

Paolo

#5 lore raw
Quoted 13 lines · Paolo Bonzini
Quoted 12 lines · Sean Christopherson

Cleanups for page fault handling that were encountered during early TDX enabling, but are worthwhile on their own. Specifically, patch 4 fixes an issue where KVM doesn't detect a spurious page fault (due to the fault being fixed by a different pCPU+vCPU) and does the full gamut of writing the SPTE, updating stats, and prefetching SPTEs.

Sean Christopherson (4):
KVM: x86/mmu: Return -EIO if page fault returns RET_PF_INVALID
KVM: x86/mmu: Invert RET_PF_* check when falling through to emulation
KVM: x86/mmu: Return unique RET_PF_* values if the fault was fixed
KVM: x86/mmu: Bail early from final #PF handling on spurious faults

Quoted 8 lines
Quoted 5 lines
 arch/x86/kvm/mmu/mmu.c         | 70 +++++++++++++++++++++-------------
 arch/x86/kvm/mmu/mmutrace.h    | 13 +++----
 arch/x86/kvm/mmu/paging_tmpl.h |  3 ++
 3 files changed, 52 insertions(+), 34 deletions(-)

Queued, thanks. Looking at the KVM_BUG_ON now since patch 1 is somewhat related.

Ha, very prescient of you, that's actually a KVM_BUG_ON() in my "kitchen sink" combo of everything :-)