* [PATCH 0/3 v3] kvm-unit-test: nVMX: Test Selector and Base Address fields of Guest Segment registers
@ 2020-09-21 8:10 Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 1/3 v3] KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it Krish Sadhukhan
` (2 more replies)
0 siblings, 3 replies; 5+ messages in thread
From: Krish Sadhukhan @ 2020-09-21 8:10 UTC (permalink / raw)
To: kvm; +Cc: pbonzini, jmattson, sean.j.christopherson
It's been a while since I sent out v2. This patch-series tests the Selector
and Base Address fields of the Guest Segement register according to section
"Checks on Guest Segment Registers" in SDM vol 3C. "Unrestricted guest"
VM-execution control is a condition in testing some parts of the Guest
Segement registers.
v2 -> v3
1. Patch# 1 has been enhanced to check the "unrestricted VM-execution
control" of the nested VMCS, along with the
"enabled_unrestricted_guest" global variable, in some of the code
paths where the VM-execution control from both vmcs01 and vmcs02
needs to be considered.
2. Patch# 3 is new. It adds a test for VMENTRY of an unrestricted guest
in unpaged protected mode.
[PATCH 1/3 v3] KVM: nVMX: KVM needs to unset "unrestricted guest"
[PATCH 2/3 v3] nVMX: Test Selector and Base Address fields of Guest Segment
[PATCH 3/3 v3] nVMX: Test vmentry of unrestricted (unpaged protected) nested
arch/x86/kvm/vmx/nested.c | 3 +++
arch/x86/kvm/vmx/vmx.c | 17 +++++++++--------
arch/x86/kvm/vmx/vmx.h | 7 +++++++
3 files changed, 19 insertions(+), 8 deletions(-)
Krish Sadhukhan (1):
KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it
lib/x86/processor.h | 1 +
x86/vmx_tests.c | 200 ++++++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 201 insertions(+)
Krish Sadhukhan (1):
nVMX: Test Selector and Base Address fields of Guest Segment Registers on vmentry of nested guests
x86/vmx.c | 2 +-
x86/vmx.h | 1 +
x86/vmx_tests.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 50 insertions(+), 1 deletion(-)
Krish Sadhukhan (1):
nVMX: Test vmentry of unrestricted (unpaged protected) nested guest
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH 1/3 v3] KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it
2020-09-21 8:10 [PATCH 0/3 v3] kvm-unit-test: nVMX: Test Selector and Base Address fields of Guest Segment registers Krish Sadhukhan
@ 2020-09-21 8:10 ` Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 2/3 v3] nVMX: Test Selector and Base Address fields of Guest Segment Registers on vmentry of nested guests Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 3/3 v3] nVMX: Test vmentry of unrestricted (unpaged protected) nested guest Krish Sadhukhan
2 siblings, 0 replies; 5+ messages in thread
From: Krish Sadhukhan @ 2020-09-21 8:10 UTC (permalink / raw)
To: kvm; +Cc: pbonzini, jmattson, sean.j.christopherson
Currently, prepare_vmcs02_early() does not check if the "unrestricted guest"
VM-execution control in vmcs12 is turned off and leaves the corresponding
bit on in vmcs02. Due to this setting, vmentry checks which are supposed to
render the nested guest state as invalid when this VM-execution control is
not set, are passing in hardware.
This patch turns off the "unrestricted guest" VM-execution control in vmcs02
if vmcs12 has turned it off.
Suggested-by: Jim Mattson <jmattson@google.com>
Suggested-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
---
arch/x86/kvm/vmx/nested.c | 3 +++
arch/x86/kvm/vmx/vmx.c | 17 +++++++++--------
arch/x86/kvm/vmx/vmx.h | 7 +++++++
3 files changed, 19 insertions(+), 8 deletions(-)
diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index 1bb6b31eb646..86fc044e5af9 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -2314,6 +2314,9 @@ static void prepare_vmcs02_early(struct vcpu_vmx *vmx, struct vmcs12 *vmcs12)
vmcs_write16(GUEST_INTR_STATUS,
vmcs12->guest_intr_status);
+ if (!nested_cpu_has2(vmcs12, SECONDARY_EXEC_UNRESTRICTED_GUEST))
+ exec_control &= ~SECONDARY_EXEC_UNRESTRICTED_GUEST;
+
secondary_exec_controls_set(vmx, exec_control);
}
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index 8646a797b7a8..0559c11d227c 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -1441,7 +1441,7 @@ void vmx_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
struct vcpu_vmx *vmx = to_vmx(vcpu);
unsigned long old_rflags;
- if (enable_unrestricted_guest) {
+ if (is_unrestricted_guest(vcpu)) {
kvm_register_mark_available(vcpu, VCPU_EXREG_RFLAGS);
vmx->rflags = rflags;
vmcs_writel(GUEST_RFLAGS, rflags);
@@ -2267,7 +2267,8 @@ static void vmx_cache_reg(struct kvm_vcpu *vcpu, enum kvm_reg reg)
vcpu->arch.cr0 |= vmcs_readl(GUEST_CR0) & guest_owned_bits;
break;
case VCPU_EXREG_CR3:
- if (enable_unrestricted_guest || (enable_ept && is_paging(vcpu)))
+ if (is_unrestricted_guest(vcpu) ||
+ (enable_ept && is_paging(vcpu)))
vcpu->arch.cr3 = vmcs_readl(GUEST_CR3);
break;
case VCPU_EXREG_CR4:
@@ -3033,7 +3034,7 @@ void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
unsigned long hw_cr0;
hw_cr0 = (cr0 & ~KVM_VM_CR0_ALWAYS_OFF);
- if (enable_unrestricted_guest)
+ if (is_unrestricted_guest(vcpu))
hw_cr0 |= KVM_VM_CR0_ALWAYS_ON_UNRESTRICTED_GUEST;
else {
hw_cr0 |= KVM_VM_CR0_ALWAYS_ON;
@@ -3054,7 +3055,7 @@ void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
}
#endif
- if (enable_ept && !enable_unrestricted_guest)
+ if (enable_ept && !is_unrestricted_guest(vcpu))
ept_update_paging_mode_cr0(&hw_cr0, cr0, vcpu);
vmcs_writel(CR0_READ_SHADOW, cr0);
@@ -3134,7 +3135,7 @@ int vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
unsigned long hw_cr4;
hw_cr4 = (cr4_read_shadow() & X86_CR4_MCE) | (cr4 & ~X86_CR4_MCE);
- if (enable_unrestricted_guest)
+ if (is_unrestricted_guest(vcpu))
hw_cr4 |= KVM_VM_CR4_ALWAYS_ON_UNRESTRICTED_GUEST;
else if (vmx->rmode.vm86_active)
hw_cr4 |= KVM_RMODE_VM_CR4_ALWAYS_ON;
@@ -3169,7 +3170,7 @@ int vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
vcpu->arch.cr4 = cr4;
kvm_register_mark_available(vcpu, VCPU_EXREG_CR4);
- if (!enable_unrestricted_guest) {
+ if (!is_unrestricted_guest(vcpu)) {
if (enable_ept) {
if (!is_paging(vcpu)) {
hw_cr4 &= ~X86_CR4_PAE;
@@ -3309,7 +3310,7 @@ void vmx_set_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg)
* tree. Newer qemu binaries with that qemu fix would not need this
* kvm hack.
*/
- if (enable_unrestricted_guest && (seg != VCPU_SREG_LDTR))
+ if (is_unrestricted_guest(vcpu) && (seg != VCPU_SREG_LDTR))
var->type |= 0x1; /* Accessed */
vmcs_write32(sf->ar_bytes, vmx_segment_access_rights(var));
@@ -3500,7 +3501,7 @@ static bool cs_ss_rpl_check(struct kvm_vcpu *vcpu)
*/
static bool guest_state_valid(struct kvm_vcpu *vcpu)
{
- if (enable_unrestricted_guest)
+ if (is_unrestricted_guest(vcpu))
return true;
/* real mode guest state checks */
diff --git a/arch/x86/kvm/vmx/vmx.h b/arch/x86/kvm/vmx/vmx.h
index a2f82127c170..0a39a7831ba9 100644
--- a/arch/x86/kvm/vmx/vmx.h
+++ b/arch/x86/kvm/vmx/vmx.h
@@ -555,6 +555,13 @@ static inline bool vmx_need_pf_intercept(struct kvm_vcpu *vcpu)
return !enable_ept || cpuid_maxphyaddr(vcpu) < boot_cpu_data.x86_phys_bits;
}
+static inline bool is_unrestricted_guest(struct kvm_vcpu *vcpu)
+{
+ return enable_unrestricted_guest && (!is_guest_mode(vcpu) ||
+ (secondary_exec_controls_get(to_vmx(vcpu)) &
+ SECONDARY_EXEC_UNRESTRICTED_GUEST));
+}
+
void dump_vmcs(void);
#endif /* __KVM_X86_VMX_H */
--
2.18.4
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH 2/3 v3] nVMX: Test Selector and Base Address fields of Guest Segment Registers on vmentry of nested guests
2020-09-21 8:10 [PATCH 0/3 v3] kvm-unit-test: nVMX: Test Selector and Base Address fields of Guest Segment registers Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 1/3 v3] KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it Krish Sadhukhan
@ 2020-09-21 8:10 ` Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 3/3 v3] nVMX: Test vmentry of unrestricted (unpaged protected) nested guest Krish Sadhukhan
2 siblings, 0 replies; 5+ messages in thread
From: Krish Sadhukhan @ 2020-09-21 8:10 UTC (permalink / raw)
To: kvm; +Cc: pbonzini, jmattson, sean.j.christopherson
According to section "Checks on Guest Segment Registers" in Intel SDM vol 3C,
the following checks are performed on the Guest Segment Registers on vmentry
of nested guests:
Selector fields:
— TR. The TI flag (bit 2) must be 0.
— LDTR. If LDTR is usable, the TI flag (bit 2) must be 0.
— SS. If the guest will not be virtual-8086 and the "unrestricted
guest" VM-execution control is 0, the RPL (bits 1:0) must equal
the RPL of the selector field for CS.1
Base-address fields:
— CS, SS, DS, ES, FS, GS. If the guest will be virtual-8086, the
address must be the selector field shifted left 4 bits (multiplied
by 16).
— The following checks are performed on processors that support Intel
64 architecture:
TR, FS, GS. The address must be canonical.
LDTR. If LDTR is usable, the address must be canonical.
CS. Bits 63:32 of the address must be zero.
SS, DS, ES. If the register is usable, bits 63:32 of the
address must be zero.
Signed-off-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
---
lib/x86/processor.h | 1 +
x86/vmx_tests.c | 200 ++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 201 insertions(+)
diff --git a/lib/x86/processor.h b/lib/x86/processor.h
index 74a2498..c2c487c 100644
--- a/lib/x86/processor.h
+++ b/lib/x86/processor.h
@@ -63,6 +63,7 @@
#define X86_EFLAGS_OF 0x00000800
#define X86_EFLAGS_IOPL 0x00003000
#define X86_EFLAGS_NT 0x00004000
+#define X86_EFLAGS_VM 0x00020000
#define X86_EFLAGS_AC 0x00040000
#define X86_EFLAGS_ALU (X86_EFLAGS_CF | X86_EFLAGS_PF | X86_EFLAGS_AF | \
diff --git a/x86/vmx_tests.c b/x86/vmx_tests.c
index 22f0c7b..2b6e47b 100644
--- a/x86/vmx_tests.c
+++ b/x86/vmx_tests.c
@@ -7980,6 +7980,203 @@ static void test_load_guest_bndcfgs(void)
vmcs_write(GUEST_BNDCFGS, bndcfgs_saved);
}
+#define GUEST_SEG_UNUSABLE_MASK (1u << 16)
+#define GUEST_SEG_SEL_TI_MASK (1u << 2)
+#define TEST_SEGMENT_SEL(xfail, sel, sel_name, val) \
+ vmcs_write(sel, val); \
+ test_guest_state("Test Guest Segment Selector", xfail, val, \
+ sel_name);
+
+/*
+ * The following checks are done on the Selector field of the Guest Segment
+ * Registers:
+ * — TR. The TI flag (bit 2) must be 0.
+ * — LDTR. If LDTR is usable, the TI flag (bit 2) must be 0.
+ * — SS. If the guest will not be virtual-8086 and the "unrestricted
+ * guest" VM-execution control is 0, the RPL (bits 1:0) must equal
+ * the RPL of the selector field for CS.
+ *
+ * [Intel SDM]
+ */
+static void test_guest_segment_sel_fields(void)
+{
+ u16 sel_saved;
+ u32 ar_saved;
+ u32 cpu_ctrl0_saved;
+ u32 cpu_ctrl1_saved;
+ u16 cs_rpl_bits;
+
+ /*
+ * Test for GUEST_SEL_TR
+ */
+ sel_saved = vmcs_read(GUEST_SEL_TR);
+ TEST_SEGMENT_SEL(true, GUEST_SEL_TR, "GUEST_SEL_TR",
+ sel_saved | GUEST_SEG_SEL_TI_MASK);
+ vmcs_write(GUEST_SEL_TR, sel_saved);
+
+ /*
+ * Test for GUEST_SEL_LDTR
+ */
+ sel_saved = vmcs_read(GUEST_SEL_LDTR);
+ ar_saved = vmcs_read(GUEST_AR_LDTR);
+ /* LDTR is set unusable */
+ vmcs_write(GUEST_AR_LDTR, ar_saved | GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_SEL(false, GUEST_SEL_LDTR, "GUEST_SEL_LDTR",
+ sel_saved | GUEST_SEG_SEL_TI_MASK);
+ TEST_SEGMENT_SEL(false, GUEST_SEL_LDTR, "GUEST_SEL_LDTR",
+ sel_saved & ~GUEST_SEG_SEL_TI_MASK);
+ /* LDTR is set usable */
+ vmcs_write(GUEST_AR_LDTR, ar_saved & ~GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_SEL(true, GUEST_SEL_LDTR, "GUEST_SEL_LDTR",
+ sel_saved | GUEST_SEG_SEL_TI_MASK);
+
+ TEST_SEGMENT_SEL(false, GUEST_SEL_LDTR, "GUEST_SEL_LDTR",
+ sel_saved & ~GUEST_SEG_SEL_TI_MASK);
+
+ vmcs_write(GUEST_AR_LDTR, ar_saved);
+ vmcs_write(GUEST_SEL_LDTR, sel_saved);
+
+ /*
+ * Test for GUEST_SEL_SS
+ */
+ cpu_ctrl0_saved = vmcs_read(CPU_EXEC_CTRL0);
+ cpu_ctrl1_saved = vmcs_read(CPU_EXEC_CTRL1);
+ ar_saved = vmcs_read(GUEST_AR_SS);
+ /* Turn off "unrestricted guest" vm-execution control */
+ vmcs_write(CPU_EXEC_CTRL1, cpu_ctrl1_saved & ~CPU_URG);
+ cs_rpl_bits = vmcs_read(GUEST_SEL_CS) & 0x3;
+ sel_saved = vmcs_read(GUEST_SEL_SS);
+ TEST_SEGMENT_SEL(true, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (~cs_rpl_bits & 0x3)));
+ TEST_SEGMENT_SEL(false, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (cs_rpl_bits & 0x3)));
+ /* Make SS usable if it's unusable or vice-versa */
+ if (ar_saved & GUEST_SEG_UNUSABLE_MASK)
+ vmcs_write(GUEST_AR_SS, ar_saved & ~GUEST_SEG_UNUSABLE_MASK);
+ else
+ vmcs_write(GUEST_AR_SS, ar_saved | GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_SEL(true, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (~cs_rpl_bits & 0x3)));
+ TEST_SEGMENT_SEL(false, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (cs_rpl_bits & 0x3)));
+
+ /* Turn on "unrestricted guest" vm-execution control */
+ vmcs_write(CPU_EXEC_CTRL0, cpu_ctrl0_saved | CPU_SECONDARY);
+ vmcs_write(CPU_EXEC_CTRL1, cpu_ctrl1_saved | CPU_URG);
+ /* EPT and EPTP must be setup when "unrestricted guest" is on */
+ setup_ept(false);
+ TEST_SEGMENT_SEL(false, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (~cs_rpl_bits & 0x3)));
+ TEST_SEGMENT_SEL(false, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (cs_rpl_bits & 0x3)));
+ /* Make SS usable if it's unusable or vice-versa */
+ if (vmcs_read(GUEST_AR_SS) & GUEST_SEG_UNUSABLE_MASK)
+ vmcs_write(GUEST_AR_SS, ar_saved & ~GUEST_SEG_UNUSABLE_MASK);
+ else
+ vmcs_write(GUEST_AR_SS, ar_saved | GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_SEL(false, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (~cs_rpl_bits & 0x3)));
+ TEST_SEGMENT_SEL(false, GUEST_SEL_SS, "GUEST_SEL_SS",
+ ((sel_saved & ~0x3) | (cs_rpl_bits & 0x3)));
+
+ vmcs_write(GUEST_AR_SS, ar_saved);
+ vmcs_write(GUEST_SEL_SS, sel_saved);
+ vmcs_write(CPU_EXEC_CTRL0, cpu_ctrl0_saved);
+ vmcs_write(CPU_EXEC_CTRL1, cpu_ctrl1_saved);
+}
+
+#define TEST_SEGMENT_BASE_ADDR_UPPER_BITS(xfail, seg_base, seg_base_name)\
+ addr_saved = vmcs_read(seg_base); \
+ for (i = 32; i < 63; i = i + 4) { \
+ addr = addr_saved | 1ull << i; \
+ vmcs_write(seg_base, addr); \
+ test_guest_state(seg_base_name, xfail, addr, \
+ seg_base_name); \
+ } \
+ vmcs_write(seg_base, addr_saved);
+
+#define TEST_SEGMENT_BASE_ADDR_CANONICAL(xfail, seg_base, seg_base_name)\
+ addr_saved = vmcs_read(seg_base); \
+ vmcs_write(seg_base, NONCANONICAL); \
+ test_guest_state(seg_base_name, xfail, NONCANONICAL, \
+ seg_base_name); \
+ vmcs_write(seg_base, addr_saved);
+
+/*
+ * The following checks are done on the Base Address field of the Guest
+ * Segment Registers on processors that support Intel 64 architecture:
+ * - TR, FS, GS : The address must be canonical.
+ * - LDTR : If LDTR is usable, the address must be canonical.
+ * - CS : Bits 63:32 of the address must be zero.
+ * - SS, DS, ES : If the register is usable, bits 63:32 of the address
+ * must be zero.
+ *
+ * [Intel SDM]
+ */
+static void test_guest_segment_base_addr_fields(void)
+{
+ u64 addr_saved;
+ u64 addr;
+ u32 ar_saved;
+ int i;
+
+ /*
+ * The address of TR, FS, GS and LDTR must be canonical.
+ */
+ TEST_SEGMENT_BASE_ADDR_CANONICAL(true, GUEST_BASE_TR, "GUEST_BASE_TR");
+ TEST_SEGMENT_BASE_ADDR_CANONICAL(true, GUEST_BASE_FS, "GUEST_BASE_FS");
+ TEST_SEGMENT_BASE_ADDR_CANONICAL(true, GUEST_BASE_GS, "GUEST_BASE_GS");
+ ar_saved = vmcs_read(GUEST_AR_LDTR);
+ /* Make LDTR unusable */
+ vmcs_write(GUEST_AR_LDTR, ar_saved | GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_CANONICAL(false, GUEST_BASE_LDTR,
+ "GUEST_BASE_LDTR");
+ /* Make LDTR usable */
+ vmcs_write(GUEST_AR_LDTR, ar_saved & ~GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_CANONICAL(true, GUEST_BASE_LDTR,
+ "GUEST_BASE_LDTR");
+
+ vmcs_write(GUEST_AR_LDTR, ar_saved);
+
+ /*
+ * Bits 63:32 in CS, SS, DS and ES base address must be zero
+ */
+ TEST_SEGMENT_BASE_ADDR_UPPER_BITS(true, GUEST_BASE_CS,
+ "GUEST_BASE_CS");
+ ar_saved = vmcs_read(GUEST_AR_SS);
+ /* Make SS unusable */
+ vmcs_write(GUEST_AR_SS, ar_saved | GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_UPPER_BITS(false, GUEST_BASE_SS,
+ "GUEST_BASE_SS");
+ /* Make SS usable */
+ vmcs_write(GUEST_AR_SS, ar_saved & ~GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_UPPER_BITS(true, GUEST_BASE_SS,
+ "GUEST_BASE_SS");
+ vmcs_write(GUEST_AR_SS, ar_saved);
+
+ ar_saved = vmcs_read(GUEST_AR_DS);
+ /* Make DS unusable */
+ vmcs_write(GUEST_AR_DS, ar_saved | GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_UPPER_BITS(false, GUEST_BASE_DS,
+ "GUEST_BASE_DS");
+ /* Make DS usable */
+ vmcs_write(GUEST_AR_DS, ar_saved & ~GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_UPPER_BITS(true, GUEST_BASE_DS,
+ "GUEST_BASE_DS");
+ vmcs_write(GUEST_AR_DS, ar_saved);
+
+ ar_saved = vmcs_read(GUEST_AR_ES);
+ /* Make ES unusable */
+ vmcs_write(GUEST_AR_ES, ar_saved | GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_UPPER_BITS(false, GUEST_BASE_ES,
+ "GUEST_BASE_ES");
+ /* Make ES usable */
+ vmcs_write(GUEST_AR_ES, ar_saved & ~GUEST_SEG_UNUSABLE_MASK);
+ TEST_SEGMENT_BASE_ADDR_UPPER_BITS(true, GUEST_BASE_ES,
+ "GUEST_BASE_ES");
+ vmcs_write(GUEST_AR_ES, ar_saved);
+}
+
/*
* Check that the virtual CPU checks the VMX Guest State Area as
* documented in the Intel SDM.
@@ -8002,6 +8199,9 @@ static void vmx_guest_state_area_test(void)
test_load_guest_perf_global_ctrl();
test_load_guest_bndcfgs();
+ test_guest_segment_sel_fields();
+ test_guest_segment_base_addr_fields();
+
test_canonical(GUEST_BASE_GDTR, "GUEST_BASE_GDTR", false);
test_canonical(GUEST_BASE_IDTR, "GUEST_BASE_IDTR", false);
--
2.18.4
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH 3/3 v3] nVMX: Test vmentry of unrestricted (unpaged protected) nested guest
2020-09-21 8:10 [PATCH 0/3 v3] kvm-unit-test: nVMX: Test Selector and Base Address fields of Guest Segment registers Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 1/3 v3] KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 2/3 v3] nVMX: Test Selector and Base Address fields of Guest Segment Registers on vmentry of nested guests Krish Sadhukhan
@ 2020-09-21 8:10 ` Krish Sadhukhan
2020-09-29 3:21 ` Sean Christopherson
2 siblings, 1 reply; 5+ messages in thread
From: Krish Sadhukhan @ 2020-09-21 8:10 UTC (permalink / raw)
To: kvm; +Cc: pbonzini, jmattson, sean.j.christopherson
According to section "UNRESTRICTED GUESTS" in SDM vol 3c, if the
"unrestricted guest" secondary VM-execution control is set, guests can run
in unpaged protected mode or in real mode. This patch tests vmetnry of an
unrestricted guest in unpaged protected mode.
Signed-off-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
---
x86/vmx.c | 2 +-
x86/vmx.h | 1 +
x86/vmx_tests.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 50 insertions(+), 1 deletion(-)
diff --git a/x86/vmx.c b/x86/vmx.c
index 07415b4..1a84a74 100644
--- a/x86/vmx.c
+++ b/x86/vmx.c
@@ -1699,7 +1699,7 @@ static void test_vmx_caps(void)
}
/* This function can only be called in guest */
-static void __attribute__((__used__)) hypercall(u32 hypercall_no)
+void __attribute__((__used__)) hypercall(u32 hypercall_no)
{
u64 val = 0;
val = (hypercall_no & HYPERCALL_MASK) | HYPERCALL_BIT;
diff --git a/x86/vmx.h b/x86/vmx.h
index d1c2436..e29301e 100644
--- a/x86/vmx.h
+++ b/x86/vmx.h
@@ -895,6 +895,7 @@ bool ept_ad_bits_supported(void);
void __enter_guest(u8 abort_flag, struct vmentry_result *result);
void enter_guest(void);
void enter_guest_with_bad_controls(void);
+void hypercall(u32 hypercall_no);
typedef void (*test_guest_func)(void);
typedef void (*test_teardown_func)(void *data);
diff --git a/x86/vmx_tests.c b/x86/vmx_tests.c
index 22f0c7b..1cadc56 100644
--- a/x86/vmx_tests.c
+++ b/x86/vmx_tests.c
@@ -8029,6 +8029,53 @@ static void vmx_guest_state_area_test(void)
enter_guest();
}
+extern void unrestricted_guest_main(void);
+asm (".code32\n"
+ "unrestricted_guest_main:\n"
+ "vmcall\n"
+ "nop\n"
+ "mov $1, %edi\n"
+ "call hypercall\n"
+ ".code64\n");
+
+static void setup_unrestricted_guest(void)
+{
+ vmcs_write(GUEST_CR0, vmcs_read(GUEST_CR0) & ~(X86_CR0_PG));
+ vmcs_write(ENT_CONTROLS, vmcs_read(ENT_CONTROLS) & ~ENT_GUEST_64);
+ vmcs_write(GUEST_EFER, vmcs_read(GUEST_EFER) & ~EFER_LMA);
+ vmcs_write(GUEST_RIP, virt_to_phys(unrestricted_guest_main));
+}
+
+static void unsetup_unrestricted_guest(void)
+{
+ vmcs_write(GUEST_CR0, vmcs_read(GUEST_CR0) | X86_CR0_PG);
+ vmcs_write(ENT_CONTROLS, vmcs_read(ENT_CONTROLS) | ENT_GUEST_64);
+ vmcs_write(GUEST_EFER, vmcs_read(GUEST_EFER) | EFER_LMA);
+ vmcs_write(GUEST_RIP, (u64) phys_to_virt(vmcs_read(GUEST_RIP)));
+ vmcs_write(GUEST_RSP, (u64) phys_to_virt(vmcs_read(GUEST_RSP)));
+}
+
+/*
+ * If "unrestricted guest" secondary VM-execution control is set, guests
+ * can run in unpaged protected mode.
+ */
+static void vmentry_unrestricted_guest_test(void)
+{
+ test_set_guest(unrestricted_guest_main);
+ setup_unrestricted_guest();
+ if (setup_ept(false))
+ test_skip("EPT not supported");
+ vmcs_write(CPU_EXEC_CTRL1, vmcs_read(CPU_EXEC_CTRL1) | CPU_URG);
+ test_guest_state("Unrestricted guest test", false, CPU_URG, "CPU_URG");
+
+ /*
+ * Let the guest finish execution as a regular guest
+ */
+ unsetup_unrestricted_guest();
+ vmcs_write(CPU_EXEC_CTRL1, vmcs_read(CPU_EXEC_CTRL1) & ~CPU_URG);
+ enter_guest();
+}
+
static bool valid_vmcs_for_vmentry(void)
{
struct vmcs *current_vmcs = NULL;
@@ -10234,6 +10281,7 @@ struct vmx_test vmx_tests[] = {
TEST(vmx_host_state_area_test),
TEST(vmx_guest_state_area_test),
TEST(vmentry_movss_shadow_test),
+ TEST(vmentry_unrestricted_guest_test),
/* APICv tests */
TEST(vmx_eoi_bitmap_ioapic_scan_test),
TEST(vmx_hlt_with_rvi_test),
--
2.18.4
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH 3/3 v3] nVMX: Test vmentry of unrestricted (unpaged protected) nested guest
2020-09-21 8:10 ` [PATCH 3/3 v3] nVMX: Test vmentry of unrestricted (unpaged protected) nested guest Krish Sadhukhan
@ 2020-09-29 3:21 ` Sean Christopherson
0 siblings, 0 replies; 5+ messages in thread
From: Sean Christopherson @ 2020-09-29 3:21 UTC (permalink / raw)
To: Krish Sadhukhan; +Cc: kvm, pbonzini, jmattson
On Mon, Sep 21, 2020 at 08:10:27AM +0000, Krish Sadhukhan wrote:
> According to section "UNRESTRICTED GUESTS" in SDM vol 3c, if the
> "unrestricted guest" secondary VM-execution control is set, guests can run
> in unpaged protected mode or in real mode. This patch tests vmetnry of an
> unrestricted guest in unpaged protected mode.
>
> Signed-off-by: Jim Mattson <jmattson@google.com>
> Signed-off-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
SOB chain is wrong. Missing a Co-developed-by? Or is Jim supposed to be
the author?
> ---
> x86/vmx.c | 2 +-
> x86/vmx.h | 1 +
> x86/vmx_tests.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 50 insertions(+), 1 deletion(-)
>
> diff --git a/x86/vmx.c b/x86/vmx.c
> index 07415b4..1a84a74 100644
> --- a/x86/vmx.c
> +++ b/x86/vmx.c
> @@ -1699,7 +1699,7 @@ static void test_vmx_caps(void)
> }
>
> /* This function can only be called in guest */
> -static void __attribute__((__used__)) hypercall(u32 hypercall_no)
> +void __attribute__((__used__)) hypercall(u32 hypercall_no)
> {
> u64 val = 0;
> val = (hypercall_no & HYPERCALL_MASK) | HYPERCALL_BIT;
> diff --git a/x86/vmx.h b/x86/vmx.h
> index d1c2436..e29301e 100644
> --- a/x86/vmx.h
> +++ b/x86/vmx.h
> @@ -895,6 +895,7 @@ bool ept_ad_bits_supported(void);
> void __enter_guest(u8 abort_flag, struct vmentry_result *result);
> void enter_guest(void);
> void enter_guest_with_bad_controls(void);
> +void hypercall(u32 hypercall_no);
>
> typedef void (*test_guest_func)(void);
> typedef void (*test_teardown_func)(void *data);
> diff --git a/x86/vmx_tests.c b/x86/vmx_tests.c
> index 22f0c7b..1cadc56 100644
> --- a/x86/vmx_tests.c
> +++ b/x86/vmx_tests.c
> @@ -8029,6 +8029,53 @@ static void vmx_guest_state_area_test(void)
> enter_guest();
> }
>
> +extern void unrestricted_guest_main(void);
> +asm (".code32\n"
> + "unrestricted_guest_main:\n"
> + "vmcall\n"
> + "nop\n"
> + "mov $1, %edi\n"
> + "call hypercall\n"
> + ".code64\n");
> +
> +static void setup_unrestricted_guest(void)
> +{
> + vmcs_write(GUEST_CR0, vmcs_read(GUEST_CR0) & ~(X86_CR0_PG));
> + vmcs_write(ENT_CONTROLS, vmcs_read(ENT_CONTROLS) & ~ENT_GUEST_64);
> + vmcs_write(GUEST_EFER, vmcs_read(GUEST_EFER) & ~EFER_LMA);
> + vmcs_write(GUEST_RIP, virt_to_phys(unrestricted_guest_main));
> +}
> +
> +static void unsetup_unrestricted_guest(void)
> +{
> + vmcs_write(GUEST_CR0, vmcs_read(GUEST_CR0) | X86_CR0_PG);
> + vmcs_write(ENT_CONTROLS, vmcs_read(ENT_CONTROLS) | ENT_GUEST_64);
> + vmcs_write(GUEST_EFER, vmcs_read(GUEST_EFER) | EFER_LMA);
> + vmcs_write(GUEST_RIP, (u64) phys_to_virt(vmcs_read(GUEST_RIP)));
> + vmcs_write(GUEST_RSP, (u64) phys_to_virt(vmcs_read(GUEST_RSP)));
> +}
> +
> +/*
> + * If "unrestricted guest" secondary VM-execution control is set, guests
> + * can run in unpaged protected mode.
> + */
> +static void vmentry_unrestricted_guest_test(void)
> +{
> + test_set_guest(unrestricted_guest_main);
> + setup_unrestricted_guest();
> + if (setup_ept(false))
> + test_skip("EPT not supported");
> + vmcs_write(CPU_EXEC_CTRL1, vmcs_read(CPU_EXEC_CTRL1) | CPU_URG);
> + test_guest_state("Unrestricted guest test", false, CPU_URG, "CPU_URG");
Indentation looks funky.
> +
> + /*
> + * Let the guest finish execution as a regular guest
> + */
> + unsetup_unrestricted_guest();
> + vmcs_write(CPU_EXEC_CTRL1, vmcs_read(CPU_EXEC_CTRL1) & ~CPU_URG);
> + enter_guest();
> +}
> +
> static bool valid_vmcs_for_vmentry(void)
> {
> struct vmcs *current_vmcs = NULL;
> @@ -10234,6 +10281,7 @@ struct vmx_test vmx_tests[] = {
> TEST(vmx_host_state_area_test),
> TEST(vmx_guest_state_area_test),
> TEST(vmentry_movss_shadow_test),
> + TEST(vmentry_unrestricted_guest_test),
> /* APICv tests */
> TEST(vmx_eoi_bitmap_ioapic_scan_test),
> TEST(vmx_hlt_with_rvi_test),
> --
> 2.18.4
>
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2020-09-29 3:21 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-09-21 8:10 [PATCH 0/3 v3] kvm-unit-test: nVMX: Test Selector and Base Address fields of Guest Segment registers Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 1/3 v3] KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 2/3 v3] nVMX: Test Selector and Base Address fields of Guest Segment Registers on vmentry of nested guests Krish Sadhukhan
2020-09-21 8:10 ` [PATCH 3/3 v3] nVMX: Test vmentry of unrestricted (unpaged protected) nested guest Krish Sadhukhan
2020-09-29 3:21 ` Sean Christopherson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).