[PATCH v6 0/2] fallback for emulation errors

[PATCH v6 0/2] fallback for emulation errors

[Aaron Lewis]

This patchset allows userspace to be a fallback for handling emulation errors.

v1 -> v2:

 - Added additional documentation for KVM_CAP_EXIT_ON_EMULATION_FAILURE.
 - In prepare_emulation_failure_exit():
   - Created a local variable for vcpu->run.
   - Cleared the flags, emulation_failure.flags.
   - Or'd the instruction bytes flag on to emulation_failure.flags.
 - Updated the comment for KVM_INTERNAL_ERROR_EMULATION flags on how they are
   to be used.
 - Updated the comment for struct emulation_failure.

v2 -> v3:

 - Update documentation for KVM_CAP_EXIT_ON_EMULATION_FAILURE.
 - Fix spacing in prepare_emulation_failure_exit().

v3 -> v4:

 - In prepare_emulation_failure_exit():
   - Clear instruction bytes to 0x90.
   - Copy over insn_size bytes rather than sizeof(ctxt->fetch.data).
 - set_page_table_entry() takes a pte rather than mask.
 - In _vm_get_page_table_entry():
   - Removed check for page aligned addresses only.
   - Added canonical check.
   - Added a check to make sure no reserved bits are set along the walk except
     for the final pte (the pte cannot have the reserved bits checked otherwise
     the test would fail).
   - Added check to ensure superpage bits are clear.
 - Added check in test for 'allow_smaller_maxphyaddr' module parameter.
 - If the is_flds() check fails, only look at the first byte.
 - Don't use labels to increment the RIP.  Decode the instruction well enough to
   ensure it is only 2-bytes.

v4 -> v5:

 - Switch 'insn_size' to u32.
 - Add documentation for how the flags are used.
 - Remove 'max_insn_size' and use 'sizeof(run->emulation_failure.insn_bytes)' instead.
 - Fix typos.
 - Fix canonical check.
 - Add reserved check for bit-7 of PML4E.
 - Add reserved check for bit-63 of all page table levels if EFER.NXE = 0.
 - Remove opcode check (it might be a prefix).
 - Remove labels.
 - Remove detritus (rogue cpuid entry in the test).

v5 -> v6

 - Fix documentation.

Aaron Lewis (2):
  kvm: x86: Allow userspace to handle emulation errors
  selftests: kvm: Allows userspace to handle emulation errors.

 Documentation/virt/kvm/api.rst                |  19 ++
 arch/x86/include/asm/kvm_host.h               |   6 +
 arch/x86/kvm/x86.c                            |  37 ++-
 include/uapi/linux/kvm.h                      |  23 ++
 tools/include/uapi/linux/kvm.h                |  23 ++
 tools/testing/selftests/kvm/.gitignore        |   1 +
 tools/testing/selftests/kvm/Makefile          |   1 +
 .../selftests/kvm/include/x86_64/processor.h  |   4 +
 .../selftests/kvm/lib/x86_64/processor.c      |  94 ++++++++
 .../kvm/x86_64/emulator_error_test.c          | 219 ++++++++++++++++++
 10 files changed, 423 insertions(+), 4 deletions(-)
 create mode 100644 tools/testing/selftests/kvm/x86_64/emulator_error_test.c

-- 
2.31.1.607.g51e8a6a459-goog

[PATCH v6 1/2] kvm: x86: Allow userspace to handle emulation errors

[Aaron Lewis]

Add a fallback mechanism to the in-kernel instruction emulator that
allows userspace the opportunity to process an instruction the emulator
was unable to.  When the in-kernel instruction emulator fails to process
an instruction it will either inject a #UD into the guest or exit to
userspace with exit reason KVM_INTERNAL_ERROR.  This is because it does
not know how to proceed in an appropriate manner.  This feature lets
userspace get involved to see if it can figure out a better path
forward.

Signed-off-by: Aaron Lewis <aaronlewis@google.com>
Reviewed-by: David Edmondson <david.edmondson@oracle.com>
---
 Documentation/virt/kvm/api.rst  | 19 +++++++++++++++++
 arch/x86/include/asm/kvm_host.h |  6 ++++++
 arch/x86/kvm/x86.c              | 37 +++++++++++++++++++++++++++++----
 include/uapi/linux/kvm.h        | 23 ++++++++++++++++++++
 tools/include/uapi/linux/kvm.h  | 23 ++++++++++++++++++++
 5 files changed, 104 insertions(+), 4 deletions(-)

diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
index 307f2fcf1b02..17668d3e1cb7 100644
--- a/Documentation/virt/kvm/api.rst
+++ b/Documentation/virt/kvm/api.rst
@@ -6233,6 +6233,25 @@ KVM_RUN_BUS_LOCK flag is used to distinguish between them.
 This capability can be used to check / enable 2nd DAWR feature provided
 by POWER10 processor.
 
+7.24 KVM_CAP_EXIT_ON_EMULATION_FAILURE
+--------------------------------------
+
+:Architectures: x86
+:Parameters: args[0] whether the feature should be enabled or not
+
+When this capability is enabled, an emulation failure will result in an exit
+to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
+to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
+to 15 instruction bytes for any exit to userspace resulting from an emulation
+failure.  When these exits to userspace occur use the emulation_failure struct
+instead of the internal struct.  They both have the same layout, but the
+emulation_failure struct matches the content better.  It also explicitly
+defines the 'flags' field which is used to describe the fields in the struct
+that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
+set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
+in them.)
+
+
 8. Other capabilities.
 ======================
 
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 3768819693e5..07235d08e976 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -1049,6 +1049,12 @@ struct kvm_arch {
 	bool exception_payload_enabled;
 
 	bool bus_lock_detection_enabled;
+	/*
+	 * If exit_on_emulation_error is set, and the in-kernel instruction
+	 * emulator fails to emulate an instruction, allow userspace
+	 * the opportunity to look at it.
+	 */
+	bool exit_on_emulation_error;
 
 	/* Deflect RDMSR and WRMSR to user space when they trigger a #GP */
 	u32 user_space_msr_mask;
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index eca63625aee4..703bcc93b129 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3771,6 +3771,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_X86_USER_SPACE_MSR:
 	case KVM_CAP_X86_MSR_FILTER:
 	case KVM_CAP_ENFORCE_PV_FEATURE_CPUID:
+	case KVM_CAP_EXIT_ON_EMULATION_FAILURE:
 		r = 1;
 		break;
 #ifdef CONFIG_KVM_XEN
@@ -5357,6 +5358,10 @@ int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
 			kvm->arch.bus_lock_detection_enabled = true;
 		r = 0;
 		break;
+	case KVM_CAP_EXIT_ON_EMULATION_FAILURE:
+		kvm->arch.exit_on_emulation_error = cap->args[0];
+		r = 0;
+		break;
 	default:
 		r = -EINVAL;
 		break;
@@ -7119,8 +7124,33 @@ void kvm_inject_realmode_interrupt(struct kvm_vcpu *vcpu, int irq, int inc_eip)
 }
 EXPORT_SYMBOL_GPL(kvm_inject_realmode_interrupt);
 
+static void prepare_emulation_failure_exit(struct kvm_vcpu *vcpu)
+{
+	struct x86_emulate_ctxt *ctxt = vcpu->arch.emulate_ctxt;
+	u32 insn_size = ctxt->fetch.end - ctxt->fetch.data;
+	struct kvm_run *run = vcpu->run;
+
+	run->exit_reason = KVM_EXIT_INTERNAL_ERROR;
+	run->emulation_failure.suberror = KVM_INTERNAL_ERROR_EMULATION;
+	run->emulation_failure.ndata = 0;
+	run->emulation_failure.flags = 0;
+
+	if (insn_size) {
+		run->emulation_failure.ndata = 3;
+		run->emulation_failure.flags |=
+			KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES;
+		run->emulation_failure.insn_size = insn_size;
+		memset(run->emulation_failure.insn_bytes, 0x90,
+		       sizeof(run->emulation_failure.insn_bytes));
+		memcpy(run->emulation_failure.insn_bytes,
+		       ctxt->fetch.data, insn_size);
+	}
+}
+
 static int handle_emulation_failure(struct kvm_vcpu *vcpu, int emulation_type)
 {
+	struct kvm *kvm = vcpu->kvm;
+
 	++vcpu->stat.insn_emulation_fail;
 	trace_kvm_emulate_insn_failed(vcpu);
 
@@ -7129,10 +7159,9 @@ static int handle_emulation_failure(struct kvm_vcpu *vcpu, int emulation_type)
 		return 1;
 	}
 
-	if (emulation_type & EMULTYPE_SKIP) {
-		vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR;
-		vcpu->run->internal.suberror = KVM_INTERNAL_ERROR_EMULATION;
-		vcpu->run->internal.ndata = 0;
+	if (kvm->arch.exit_on_emulation_error ||
+	    (emulation_type & EMULTYPE_SKIP)) {
+		prepare_emulation_failure_exit(vcpu);
 		return 0;
 	}
 
diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h
index f6afee209620..1bca5d066e3c 100644
--- a/include/uapi/linux/kvm.h
+++ b/include/uapi/linux/kvm.h
@@ -279,6 +279,9 @@ struct kvm_xen_exit {
 /* Encounter unexpected vm-exit reason */
 #define KVM_INTERNAL_ERROR_UNEXPECTED_EXIT_REASON	4
 
+/* Flags that describe what fields in emulation_failure hold valid data. */
+#define KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES (1ULL << 0)
+
 /* for KVM_RUN, returned by mmap(vcpu_fd, offset=0) */
 struct kvm_run {
 	/* in */
@@ -382,6 +385,25 @@ struct kvm_run {
 			__u32 ndata;
 			__u64 data[16];
 		} internal;
+		/*
+		 * KVM_INTERNAL_ERROR_EMULATION
+		 *
+		 * "struct emulation_failure" is an overlay of "struct internal"
+		 * that is used for the KVM_INTERNAL_ERROR_EMULATION sub-type of
+		 * KVM_EXIT_INTERNAL_ERROR.  Note, unlike other internal error
+		 * sub-types, this struct is ABI!  It also needs to be backwards
+		 * compatible with "struct internal".  Take special care that
+		 * "ndata" is correct, that new fields are enumerated in "flags",
+		 * and that each flag enumerates fields that are 64-bit aligned
+		 * and sized (so that ndata+internal.data[] is valid/accurate).
+		 */
+		struct {
+			__u32 suberror;
+			__u32 ndata;
+			__u64 flags;
+			__u8  insn_size;
+			__u8  insn_bytes[15];
+		} emulation_failure;
 		/* KVM_EXIT_OSI */
 		struct {
 			__u64 gprs[32];
@@ -1078,6 +1100,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_DIRTY_LOG_RING 192
 #define KVM_CAP_X86_BUS_LOCK_EXIT 193
 #define KVM_CAP_PPC_DAWR1 194
+#define KVM_CAP_EXIT_ON_EMULATION_FAILURE 195
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
diff --git a/tools/include/uapi/linux/kvm.h b/tools/include/uapi/linux/kvm.h
index f6afee209620..1bca5d066e3c 100644
--- a/tools/include/uapi/linux/kvm.h
+++ b/tools/include/uapi/linux/kvm.h
@@ -279,6 +279,9 @@ struct kvm_xen_exit {
 /* Encounter unexpected vm-exit reason */
 #define KVM_INTERNAL_ERROR_UNEXPECTED_EXIT_REASON	4
 
+/* Flags that describe what fields in emulation_failure hold valid data. */
+#define KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES (1ULL << 0)
+
 /* for KVM_RUN, returned by mmap(vcpu_fd, offset=0) */
 struct kvm_run {
 	/* in */
@@ -382,6 +385,25 @@ struct kvm_run {
 			__u32 ndata;
 			__u64 data[16];
 		} internal;
+		/*
+		 * KVM_INTERNAL_ERROR_EMULATION
+		 *
+		 * "struct emulation_failure" is an overlay of "struct internal"
+		 * that is used for the KVM_INTERNAL_ERROR_EMULATION sub-type of
+		 * KVM_EXIT_INTERNAL_ERROR.  Note, unlike other internal error
+		 * sub-types, this struct is ABI!  It also needs to be backwards
+		 * compatible with "struct internal".  Take special care that
+		 * "ndata" is correct, that new fields are enumerated in "flags",
+		 * and that each flag enumerates fields that are 64-bit aligned
+		 * and sized (so that ndata+internal.data[] is valid/accurate).
+		 */
+		struct {
+			__u32 suberror;
+			__u32 ndata;
+			__u64 flags;
+			__u8  insn_size;
+			__u8  insn_bytes[15];
+		} emulation_failure;
 		/* KVM_EXIT_OSI */
 		struct {
 			__u64 gprs[32];
@@ -1078,6 +1100,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_DIRTY_LOG_RING 192
 #define KVM_CAP_X86_BUS_LOCK_EXIT 193
 #define KVM_CAP_PPC_DAWR1 194
+#define KVM_CAP_EXIT_ON_EMULATION_FAILURE 195
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
-- 
2.31.1.607.g51e8a6a459-goog

Re: [PATCH v6 1/2] kvm: x86: Allow userspace to handle emulation errors

[Jim Mattson]

On Mon, May 10, 2021 at 7:48 AM Aaron Lewis <aaronlewis@google.com> wrote:
>
> Add a fallback mechanism to the in-kernel instruction emulator that
> allows userspace the opportunity to process an instruction the emulator
> was unable to.  When the in-kernel instruction emulator fails to process
> an instruction it will either inject a #UD into the guest or exit to
> userspace with exit reason KVM_INTERNAL_ERROR.  This is because it does
> not know how to proceed in an appropriate manner.  This feature lets
> userspace get involved to see if it can figure out a better path
> forward.
>
> Signed-off-by: Aaron Lewis <aaronlewis@google.com>
> Reviewed-by: David Edmondson <david.edmondson@oracle.com>
Reviewed-by: Jim Mattson <jmattson@google.com>

Re: [PATCH v6 1/2] kvm: x86: Allow userspace to handle emulation errors

[Aaron Lewis]

On Thu, Jun 3, 2021 at 1:35 PM Jim Mattson <jmattson@google.com> wrote:
>
> On Mon, May 10, 2021 at 7:48 AM Aaron Lewis <aaronlewis@google.com> wrote:
> >
> > Add a fallback mechanism to the in-kernel instruction emulator that
> > allows userspace the opportunity to process an instruction the emulator
> > was unable to.  When the in-kernel instruction emulator fails to process
> > an instruction it will either inject a #UD into the guest or exit to
> > userspace with exit reason KVM_INTERNAL_ERROR.  This is because it does
> > not know how to proceed in an appropriate manner.  This feature lets
> > userspace get involved to see if it can figure out a better path
> > forward.
> >
> > Signed-off-by: Aaron Lewis <aaronlewis@google.com>
> > Reviewed-by: David Edmondson <david.edmondson@oracle.com>
> Reviewed-by: Jim Mattson <jmattson@google.com>

Hi Paolo,

Does this change look okay to you?  Can I get it queued?

Thanks,
Aaron

Re: [PATCH v6 1/2] kvm: x86: Allow userspace to handle emulation errors

[Paolo Bonzini]

On 23/06/21 19:44, Aaron Lewis wrote:
> On Thu, Jun 3, 2021 at 1:35 PM Jim Mattson <jmattson@google.com> wrote:
>>
>> On Mon, May 10, 2021 at 7:48 AM Aaron Lewis <aaronlewis@google.com> wrote:
>>>
>>> Add a fallback mechanism to the in-kernel instruction emulator that
>>> allows userspace the opportunity to process an instruction the emulator
>>> was unable to.  When the in-kernel instruction emulator fails to process
>>> an instruction it will either inject a #UD into the guest or exit to
>>> userspace with exit reason KVM_INTERNAL_ERROR.  This is because it does
>>> not know how to proceed in an appropriate manner.  This feature lets
>>> userspace get involved to see if it can figure out a better path
>>> forward.
>>>
>>> Signed-off-by: Aaron Lewis <aaronlewis@google.com>
>>> Reviewed-by: David Edmondson <david.edmondson@oracle.com>
>> Reviewed-by: Jim Mattson <jmattson@google.com>
> 
> Hi Paolo,
> 
> Does this change look okay to you?  Can I get it queued?
> 
> Thanks,
> Aaron
> 

Sorry I missed it.  Looking now.

Paolo

[PATCH v6 2/2] selftests: kvm: Allows userspace to handle emulation errors.

[Aaron Lewis]

This test exercises the feature KVM_CAP_EXIT_ON_EMULATION_FAILURE.  When
enabled, errors in the in-kernel instruction emulator are forwarded to
userspace with the instruction bytes stored in the exit struct for
KVM_EXIT_INTERNAL_ERROR.  So, when the guest attempts to emulate an
'flds' instruction, which isn't able to be emulated in KVM, instead
of failing, KVM sends the instruction to userspace to handle.

For this test to work properly the module parameter
'allow_smaller_maxphyaddr' has to be set.

Signed-off-by: Aaron Lewis <aaronlewis@google.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
---
 tools/testing/selftests/kvm/.gitignore        |   1 +
 tools/testing/selftests/kvm/Makefile          |   1 +
 .../selftests/kvm/include/x86_64/processor.h  |   4 +
 .../selftests/kvm/lib/x86_64/processor.c      |  94 ++++++++
 .../kvm/x86_64/emulator_error_test.c          | 219 ++++++++++++++++++
 5 files changed, 319 insertions(+)
 create mode 100644 tools/testing/selftests/kvm/x86_64/emulator_error_test.c

diff --git a/tools/testing/selftests/kvm/.gitignore b/tools/testing/selftests/kvm/.gitignore
index 7bd7e776c266..ec9e20a2f752 100644
--- a/tools/testing/selftests/kvm/.gitignore
+++ b/tools/testing/selftests/kvm/.gitignore
@@ -7,6 +7,7 @@
 /x86_64/cr4_cpuid_sync_test
 /x86_64/debug_regs
 /x86_64/evmcs_test
+/x86_64/emulator_error_test
 /x86_64/get_cpuid_test
 /x86_64/get_msr_index_features
 /x86_64/kvm_pv_test
diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
index 67eebb53235f..5ff705d92d02 100644
--- a/tools/testing/selftests/kvm/Makefile
+++ b/tools/testing/selftests/kvm/Makefile
@@ -41,6 +41,7 @@ LIBKVM_s390x = lib/s390x/processor.c lib/s390x/ucall.c lib/s390x/diag318_test_ha
 TEST_GEN_PROGS_x86_64 = x86_64/cr4_cpuid_sync_test
 TEST_GEN_PROGS_x86_64 += x86_64/get_msr_index_features
 TEST_GEN_PROGS_x86_64 += x86_64/evmcs_test
+TEST_GEN_PROGS_x86_64 += x86_64/emulator_error_test
 TEST_GEN_PROGS_x86_64 += x86_64/get_cpuid_test
 TEST_GEN_PROGS_x86_64 += x86_64/hyperv_clock
 TEST_GEN_PROGS_x86_64 += x86_64/hyperv_cpuid
diff --git a/tools/testing/selftests/kvm/include/x86_64/processor.h b/tools/testing/selftests/kvm/include/x86_64/processor.h
index 0b30b4e15c38..8fdd21710aae 100644
--- a/tools/testing/selftests/kvm/include/x86_64/processor.h
+++ b/tools/testing/selftests/kvm/include/x86_64/processor.h
@@ -394,6 +394,10 @@ void vcpu_init_descriptor_tables(struct kvm_vm *vm, uint32_t vcpuid);
 void vm_handle_exception(struct kvm_vm *vm, int vector,
 			void (*handler)(struct ex_regs *));
 
+uint64_t vm_get_page_table_entry(struct kvm_vm *vm, int vcpuid, uint64_t vaddr);
+void vm_set_page_table_entry(struct kvm_vm *vm, int vcpuid, uint64_t vaddr,
+			     uint64_t pte);
+
 /*
  * set_cpuid() - overwrites a matching cpuid entry with the provided value.
  *		 matches based on ent->function && ent->index. returns true
diff --git a/tools/testing/selftests/kvm/lib/x86_64/processor.c b/tools/testing/selftests/kvm/lib/x86_64/processor.c
index a8906e60a108..78f5109e30da 100644
--- a/tools/testing/selftests/kvm/lib/x86_64/processor.c
+++ b/tools/testing/selftests/kvm/lib/x86_64/processor.c
@@ -292,6 +292,100 @@ void virt_pg_map(struct kvm_vm *vm, uint64_t vaddr, uint64_t paddr,
 	pte[index[0]].present = 1;
 }
 
+static struct pageTableEntry *_vm_get_page_table_entry(struct kvm_vm *vm, int vcpuid,
+						       uint64_t vaddr)
+{
+	uint16_t index[4];
+	struct pageMapL4Entry *pml4e;
+	struct pageDirectoryPointerEntry *pdpe;
+	struct pageDirectoryEntry *pde;
+	struct pageTableEntry *pte;
+	struct kvm_cpuid_entry2 *entry;
+	struct kvm_sregs sregs;
+	int max_phy_addr;
+	/* Set the bottom 52 bits. */
+	uint64_t rsvd_mask = 0x000fffffffffffff;
+
+	entry = kvm_get_supported_cpuid_index(0x80000008, 0);
+	max_phy_addr = entry->eax & 0x000000ff;
+	/* Clear the bottom bits of the reserved mask. */
+	rsvd_mask = (rsvd_mask >> max_phy_addr) << max_phy_addr;
+
+	/*
+	 * SDM vol 3, fig 4-11 "Formats of CR3 and Paging-Structure Entries
+	 * with 4-Level Paging and 5-Level Paging".
+	 * If IA32_EFER.NXE = 0 and the P flag of a paging-structure entry is 1,
+	 * the XD flag (bit 63) is reserved.
+	 */
+	vcpu_sregs_get(vm, vcpuid, &sregs);
+	if ((sregs.efer & EFER_NX) == 0) {
+		rsvd_mask |= (1ull << 63);
+	}
+
+	TEST_ASSERT(vm->mode == VM_MODE_PXXV48_4K, "Attempt to use "
+		"unknown or unsupported guest mode, mode: 0x%x", vm->mode);
+	TEST_ASSERT(sparsebit_is_set(vm->vpages_valid,
+		(vaddr >> vm->page_shift)),
+		"Invalid virtual address, vaddr: 0x%lx",
+		vaddr);
+	/*
+	 * Based on the mode check above there are 48 bits in the vaddr, so
+	 * shift 16 to sign extend the last bit (bit-47),
+	 */
+	TEST_ASSERT(vaddr == (((int64_t)vaddr << 16) >> 16),
+		"Canonical check failed.  The virtual address is invalid.");
+
+	index[0] = (vaddr >> 12) & 0x1ffu;
+	index[1] = (vaddr >> 21) & 0x1ffu;
+	index[2] = (vaddr >> 30) & 0x1ffu;
+	index[3] = (vaddr >> 39) & 0x1ffu;
+
+	pml4e = addr_gpa2hva(vm, vm->pgd);
+	TEST_ASSERT(pml4e[index[3]].present,
+		"Expected pml4e to be present for gva: 0x%08lx", vaddr);
+	TEST_ASSERT((*(uint64_t*)(&pml4e[index[3]]) &
+		(rsvd_mask | (1ull << 7))) == 0,
+		"Unexpected reserved bits set.");
+
+	pdpe = addr_gpa2hva(vm, pml4e[index[3]].address * vm->page_size);
+	TEST_ASSERT(pdpe[index[2]].present,
+		"Expected pdpe to be present for gva: 0x%08lx", vaddr);
+	TEST_ASSERT(pdpe[index[2]].page_size == 0,
+		"Expected pdpe to map a pde not a 1-GByte page.");
+	TEST_ASSERT((*(uint64_t*)(&pdpe[index[2]]) & rsvd_mask) == 0,
+		"Unexpected reserved bits set.");
+
+	pde = addr_gpa2hva(vm, pdpe[index[2]].address * vm->page_size);
+	TEST_ASSERT(pde[index[1]].present,
+		"Expected pde to be present for gva: 0x%08lx", vaddr);
+	TEST_ASSERT(pde[index[1]].page_size == 0,
+		"Expected pde to map a pte not a 2-MByte page.");
+	TEST_ASSERT((*(uint64_t*)(&pde[index[1]]) & rsvd_mask) == 0,
+		"Unexpected reserved bits set.");
+
+	pte = addr_gpa2hva(vm, pde[index[1]].address * vm->page_size);
+	TEST_ASSERT(pte[index[0]].present,
+		"Expected pte to be present for gva: 0x%08lx", vaddr);
+
+	return &pte[index[0]];
+}
+
+uint64_t vm_get_page_table_entry(struct kvm_vm *vm, int vcpuid, uint64_t vaddr)
+{
+	struct pageTableEntry *pte = _vm_get_page_table_entry(vm, vcpuid, vaddr);
+
+	return *(uint64_t *)pte;
+}
+
+void vm_set_page_table_entry(struct kvm_vm *vm, int vcpuid, uint64_t vaddr,
+			     uint64_t pte)
+{
+	struct pageTableEntry *new_pte = _vm_get_page_table_entry(vm, vcpuid,
+								  vaddr);
+
+	*(uint64_t *)new_pte = pte;
+}
+
 void virt_dump(FILE *stream, struct kvm_vm *vm, uint8_t indent)
 {
 	struct pageMapL4Entry *pml4e, *pml4e_start;
diff --git a/tools/testing/selftests/kvm/x86_64/emulator_error_test.c b/tools/testing/selftests/kvm/x86_64/emulator_error_test.c
new file mode 100644
index 000000000000..054b39e49413
--- /dev/null
+++ b/tools/testing/selftests/kvm/x86_64/emulator_error_test.c
@@ -0,0 +1,219 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020, Google LLC.
+ *
+ * Tests for KVM_CAP_EXIT_ON_EMULATION_FAILURE capability.
+ */
+
+#define _GNU_SOURCE /* for program_invocation_short_name */
+
+#include "test_util.h"
+#include "kvm_util.h"
+#include "vmx.h"
+
+#define VCPU_ID	   1
+#define PAGE_SIZE  4096
+#define MAXPHYADDR 36
+
+#define MEM_REGION_GVA	0x0000123456789000
+#define MEM_REGION_GPA	0x0000000700000000
+#define MEM_REGION_SLOT	10
+#define MEM_REGION_SIZE PAGE_SIZE
+
+static void guest_code(void)
+{
+	__asm__ __volatile__("flds (%[addr])"
+			     :: [addr]"r"(MEM_REGION_GVA));
+
+	GUEST_DONE();
+}
+
+static void run_guest(struct kvm_vm *vm)
+{
+	int rc;
+
+	rc = _vcpu_run(vm, VCPU_ID);
+	TEST_ASSERT(rc == 0, "vcpu_run failed: %d\n", rc);
+}
+
+/*
+ * Accessors to get R/M, REG, and Mod bits described in the SDM vol 2,
+ * figure 2-2 "Table Interpretation of ModR/M Byte (C8H)".
+ */
+#define GET_RM(insn_byte) (insn_byte & 0x7)
+#define GET_REG(insn_byte) ((insn_byte & 0x38) >> 3)
+#define GET_MOD(insn_byte) ((insn_byte & 0xc) >> 6)
+
+/* Ensure we are dealing with a simple 2-byte flds instruction. */
+static bool is_flds(uint8_t *insn_bytes, uint8_t insn_size)
+{
+	return insn_size >= 2 &&
+	       insn_bytes[0] == 0xd9 &&
+	       GET_REG(insn_bytes[1]) == 0x0 &&
+	       GET_MOD(insn_bytes[1]) == 0x0 &&
+	       /* Ensure there is no SIB byte. */
+	       GET_RM(insn_bytes[1]) != 0x4 &&
+	       /* Ensure there is no displacement byte. */
+	       GET_RM(insn_bytes[1]) != 0x5;
+}
+
+static void process_exit_on_emulation_error(struct kvm_vm *vm)
+{
+	struct kvm_run *run = vcpu_state(vm, VCPU_ID);
+	struct kvm_regs regs;
+	uint8_t *insn_bytes;
+	uint8_t insn_size;
+	uint64_t flags;
+
+	TEST_ASSERT(run->exit_reason == KVM_EXIT_INTERNAL_ERROR,
+		    "Unexpected exit reason: %u (%s)",
+		    run->exit_reason,
+		    exit_reason_str(run->exit_reason));
+
+	TEST_ASSERT(run->emulation_failure.suberror == KVM_INTERNAL_ERROR_EMULATION,
+		    "Unexpected suberror: %u",
+		    run->emulation_failure.suberror);
+
+	if (run->emulation_failure.ndata >= 1) {
+		flags = run->emulation_failure.flags;
+		if ((flags & KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES) &&
+		    run->emulation_failure.ndata >= 3) {
+			insn_size = run->emulation_failure.insn_size;
+			insn_bytes = run->emulation_failure.insn_bytes;
+
+			TEST_ASSERT(insn_size <= 15 && insn_size > 0,
+				    "Unexpected instruction size: %u",
+				    insn_size);
+
+			TEST_ASSERT(is_flds(insn_bytes, insn_size),
+				    "Unexpected instruction.  Expected 'flds' (0xd9 /0)");
+
+			/*
+			 * If is_flds() succeeded then the instruction bytes
+			 * contained an flds instruction that is 2-bytes in
+			 * length (ie: no prefix, no SIB, no displacement).
+			 */
+			vcpu_regs_get(vm, VCPU_ID, &regs);
+			regs.rip += 2;
+			vcpu_regs_set(vm, VCPU_ID, &regs);
+		}
+	}
+}
+
+static void do_guest_assert(struct kvm_vm *vm, struct ucall *uc)
+{
+	TEST_FAIL("%s at %s:%ld", (const char *)uc->args[0], __FILE__,
+		  uc->args[1]);
+}
+
+static void check_for_guest_assert(struct kvm_vm *vm)
+{
+	struct kvm_run *run = vcpu_state(vm, VCPU_ID);
+	struct ucall uc;
+
+	if (run->exit_reason == KVM_EXIT_IO &&
+	    get_ucall(vm, VCPU_ID, &uc) == UCALL_ABORT) {
+		do_guest_assert(vm, &uc);
+	}
+}
+
+static void process_ucall_done(struct kvm_vm *vm)
+{
+	struct kvm_run *run = vcpu_state(vm, VCPU_ID);
+	struct ucall uc;
+
+	check_for_guest_assert(vm);
+
+	TEST_ASSERT(run->exit_reason == KVM_EXIT_IO,
+		    "Unexpected exit reason: %u (%s)",
+		    run->exit_reason,
+		    exit_reason_str(run->exit_reason));
+
+	TEST_ASSERT(get_ucall(vm, VCPU_ID, &uc) == UCALL_DONE,
+		    "Unexpected ucall command: %lu, expected UCALL_DONE (%d)",
+		    uc.cmd, UCALL_DONE);
+}
+
+static uint64_t process_ucall(struct kvm_vm *vm)
+{
+	struct kvm_run *run = vcpu_state(vm, VCPU_ID);
+	struct ucall uc;
+
+	TEST_ASSERT(run->exit_reason == KVM_EXIT_IO,
+		    "Unexpected exit reason: %u (%s)",
+		    run->exit_reason,
+		    exit_reason_str(run->exit_reason));
+
+	switch (get_ucall(vm, VCPU_ID, &uc)) {
+	case UCALL_SYNC:
+		break;
+	case UCALL_ABORT:
+		do_guest_assert(vm, &uc);
+		break;
+	case UCALL_DONE:
+		process_ucall_done(vm);
+		break;
+	default:
+		TEST_ASSERT(false, "Unexpected ucall");
+	}
+
+	return uc.cmd;
+}
+
+int main(int argc, char *argv[])
+{
+	struct kvm_enable_cap emul_failure_cap = {
+		.cap = KVM_CAP_EXIT_ON_EMULATION_FAILURE,
+		.args[0] = 1,
+	};
+	struct kvm_cpuid_entry2 *entry;
+	struct kvm_cpuid2 *cpuid;
+	struct kvm_vm *vm;
+	uint64_t gpa, pte;
+	uint64_t *hva;
+	int rc;
+
+	/* Tell stdout not to buffer its content */
+	setbuf(stdout, NULL);
+
+	vm = vm_create_default(VCPU_ID, 0, guest_code);
+
+	if (!kvm_check_cap(KVM_CAP_SMALLER_MAXPHYADDR)) {
+		printf("module parameter 'allow_smaller_maxphyaddr' is not set.  Skipping test.\n");
+		return 0;
+	}
+
+	cpuid = kvm_get_supported_cpuid();
+
+	entry = kvm_get_supported_cpuid_index(0x80000008, 0);
+	entry->eax = (entry->eax & 0xffffff00) | MAXPHYADDR;
+	set_cpuid(cpuid, entry);
+
+	vcpu_set_cpuid(vm, VCPU_ID, cpuid);
+
+	rc = kvm_check_cap(KVM_CAP_EXIT_ON_EMULATION_FAILURE);
+	TEST_ASSERT(rc, "KVM_CAP_EXIT_ON_EMULATION_FAILURE is unavailable");
+	vm_enable_cap(vm, &emul_failure_cap);
+
+	vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS,
+				    MEM_REGION_GPA, MEM_REGION_SLOT,
+				    MEM_REGION_SIZE / PAGE_SIZE, 0);
+	gpa = vm_phy_pages_alloc(vm, MEM_REGION_SIZE / PAGE_SIZE,
+				 MEM_REGION_GPA, MEM_REGION_SLOT);
+	TEST_ASSERT(gpa == MEM_REGION_GPA, "Failed vm_phy_pages_alloc\n");
+	virt_map(vm, MEM_REGION_GVA, MEM_REGION_GPA, 1, 0);
+	hva = addr_gpa2hva(vm, MEM_REGION_GPA);
+	memset(hva, 0, PAGE_SIZE);
+	pte = vm_get_page_table_entry(vm, VCPU_ID, MEM_REGION_GVA);
+	vm_set_page_table_entry(vm, VCPU_ID, MEM_REGION_GVA, pte | (1ull << 36));
+
+	run_guest(vm);
+	process_exit_on_emulation_error(vm);
+	run_guest(vm);
+
+	TEST_ASSERT(process_ucall(vm) == UCALL_DONE, "Expected UCALL_DONE");
+
+	kvm_vm_free(vm);
+
+	return 0;
+}
-- 
2.31.1.607.g51e8a6a459-goog

Re: [PATCH v6 0/2] fallback for emulation errors

[Paolo Bonzini]

On 10/05/21 16:48, Aaron Lewis wrote:
> This patchset allows userspace to be a fallback for handling emulation errors.
> 
> v1 -> v2:
> 
>   - Added additional documentation for KVM_CAP_EXIT_ON_EMULATION_FAILURE.
>   - In prepare_emulation_failure_exit():
>     - Created a local variable for vcpu->run.
>     - Cleared the flags, emulation_failure.flags.
>     - Or'd the instruction bytes flag on to emulation_failure.flags.
>   - Updated the comment for KVM_INTERNAL_ERROR_EMULATION flags on how they are
>     to be used.
>   - Updated the comment for struct emulation_failure.
> 
> v2 -> v3:
> 
>   - Update documentation for KVM_CAP_EXIT_ON_EMULATION_FAILURE.
>   - Fix spacing in prepare_emulation_failure_exit().
> 
> v3 -> v4:
> 
>   - In prepare_emulation_failure_exit():
>     - Clear instruction bytes to 0x90.
>     - Copy over insn_size bytes rather than sizeof(ctxt->fetch.data).
>   - set_page_table_entry() takes a pte rather than mask.
>   - In _vm_get_page_table_entry():
>     - Removed check for page aligned addresses only.
>     - Added canonical check.
>     - Added a check to make sure no reserved bits are set along the walk except
>       for the final pte (the pte cannot have the reserved bits checked otherwise
>       the test would fail).
>     - Added check to ensure superpage bits are clear.
>   - Added check in test for 'allow_smaller_maxphyaddr' module parameter.
>   - If the is_flds() check fails, only look at the first byte.
>   - Don't use labels to increment the RIP.  Decode the instruction well enough to
>     ensure it is only 2-bytes.
> 
> v4 -> v5:
> 
>   - Switch 'insn_size' to u32.
>   - Add documentation for how the flags are used.
>   - Remove 'max_insn_size' and use 'sizeof(run->emulation_failure.insn_bytes)' instead.
>   - Fix typos.
>   - Fix canonical check.
>   - Add reserved check for bit-7 of PML4E.
>   - Add reserved check for bit-63 of all page table levels if EFER.NXE = 0.
>   - Remove opcode check (it might be a prefix).
>   - Remove labels.
>   - Remove detritus (rogue cpuid entry in the test).
> 
> v5 -> v6
> 
>   - Fix documentation.
> 
> Aaron Lewis (2):
>    kvm: x86: Allow userspace to handle emulation errors
>    selftests: kvm: Allows userspace to handle emulation errors.
> 
>   Documentation/virt/kvm/api.rst                |  19 ++
>   arch/x86/include/asm/kvm_host.h               |   6 +
>   arch/x86/kvm/x86.c                            |  37 ++-
>   include/uapi/linux/kvm.h                      |  23 ++
>   tools/include/uapi/linux/kvm.h                |  23 ++
>   tools/testing/selftests/kvm/.gitignore        |   1 +
>   tools/testing/selftests/kvm/Makefile          |   1 +
>   .../selftests/kvm/include/x86_64/processor.h  |   4 +
>   .../selftests/kvm/lib/x86_64/processor.c      |  94 ++++++++
>   .../kvm/x86_64/emulator_error_test.c          | 219 ++++++++++++++++++
>   10 files changed, 423 insertions(+), 4 deletions(-)
>   create mode 100644 tools/testing/selftests/kvm/x86_64/emulator_error_test.c
> 

Queued, thanks (not yet tested, but still).

Paolo