[PATCH v8 0/6] Enable CET support for guest

[PATCH v8 0/6] Enable CET support for guest

[Yang Weijiang]

Control-flow Enforcement Technology (CET) provides protection against
Return/Jump-Oriented Programming (ROP/JOP). It includes two features:
Shadow Stack(SHSTK) and Indirect Branch Tracking(IBT).
This patch series is to enable CET related CPUID report, XSAVES/XRSTORS
support and MSR access etc. for guest.

Change in v8:
- Extended xsave_area_size() to accommodate compacted format size calculation.
- Added CPUID(0xD,1).EBX assigment per maintain's feedback.
- Changed XSS field check and added more comments to make things clearer.
- Other ajustment per maintainer's review feedback.
- Rebased to 6.0.0.

v7 patch:
https://lore.kernel.org/kvm/20210226022058.24562-1-weijiang.yang@intel.com

CET KVM patches:
https://git.kernel.org/pub/scm/virt/kvm/kvm.git/log/?h=intel

CET kernel patches:
https://lkml.kernel.org/r/20210427204315.24153-1-yu-cheng.yu@intel.com


Yang Weijiang (6):
  target/i386: Change XSAVE related feature-word names
  target/i386: Enable XSS feature CPUID enumeration
  target/i386: Enable XSAVES support for CET states
  target/i386: Add user-space MSR access interface for CET
  target/i386: Add CET state support for guest migration
  target/i386: Advise CET bits in CPU/MSR feature words

 target/i386/cpu.c     | 138 +++++++++++++++++++++++++++++-------
 target/i386/cpu.h     |  52 +++++++++++++-
 target/i386/kvm/kvm.c |  72 +++++++++++++++++++
 target/i386/machine.c | 161 ++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 395 insertions(+), 28 deletions(-)

-- 
2.26.2

[PATCH v8 1/6] target/i386: Change XSAVE related feature-word names

[Yang Weijiang]

Rename XSAVE related feature-words for introducing XSAVES related
feature-words.

Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
---
 target/i386/cpu.c | 24 ++++++++++++------------
 target/i386/cpu.h |  4 ++--
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index ad99cad0e7..5c76186883 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -1077,7 +1077,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
         .cpuid = { .eax = 6, .reg = R_EAX, },
         .tcg_features = TCG_6_EAX_FEATURES,
     },
-    [FEAT_XSAVE_COMP_LO] = {
+    [FEAT_XSAVE_XCR0_LO] = {
         .type = CPUID_FEATURE_WORD,
         .cpuid = {
             .eax = 0xD,
@@ -1090,7 +1090,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
             XSTATE_OPMASK_MASK | XSTATE_ZMM_Hi256_MASK | XSTATE_Hi16_ZMM_MASK |
             XSTATE_PKRU_MASK,
     },
-    [FEAT_XSAVE_COMP_HI] = {
+    [FEAT_XSAVE_XCR0_HI] = {
         .type = CPUID_FEATURE_WORD,
         .cpuid = {
             .eax = 0xD,
@@ -1519,8 +1519,8 @@ static inline bool accel_uses_host_cpuid(void)
 
 static inline uint64_t x86_cpu_xsave_components(X86CPU *cpu)
 {
-    return ((uint64_t)cpu->env.features[FEAT_XSAVE_COMP_HI]) << 32 |
-           cpu->env.features[FEAT_XSAVE_COMP_LO];
+    return ((uint64_t)cpu->env.features[FEAT_XSAVE_XCR0_HI]) << 32 |
+           cpu->env.features[FEAT_XSAVE_XCR0_LO];
 }
 
 /* Return name of 32-bit register, from a R_* constant */
@@ -4811,8 +4811,8 @@ static const char *x86_cpu_feature_name(FeatureWord w, int bitnr)
     /* XSAVE components are automatically enabled by other features,
      * so return the original feature name instead
      */
-    if (w == FEAT_XSAVE_COMP_LO || w == FEAT_XSAVE_COMP_HI) {
-        int comp = (w == FEAT_XSAVE_COMP_HI) ? bitnr + 32 : bitnr;
+    if (w == FEAT_XSAVE_XCR0_LO || w == FEAT_XSAVE_XCR0_HI) {
+        int comp = (w == FEAT_XSAVE_XCR0_HI) ? bitnr + 32 : bitnr;
 
         if (comp < ARRAY_SIZE(x86_ext_save_areas) &&
             x86_ext_save_areas[comp].bits) {
@@ -5860,8 +5860,8 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
 
         if (count == 0) {
             *ecx = xsave_area_size(x86_cpu_xsave_components(cpu));
-            *eax = env->features[FEAT_XSAVE_COMP_LO];
-            *edx = env->features[FEAT_XSAVE_COMP_HI];
+            *eax = env->features[FEAT_XSAVE_XCR0_LO];
+            *edx = env->features[FEAT_XSAVE_XCR0_HI];
             /*
              * The initial value of xcr0 and ebx == 0, On host without kvm
              * commit 412a3c41(e.g., CentOS 6), the ebx's value always == 0
@@ -6431,8 +6431,8 @@ static void x86_cpu_enable_xsave_components(X86CPU *cpu)
     uint64_t mask;
 
     if (!(env->features[FEAT_1_ECX] & CPUID_EXT_XSAVE)) {
-        env->features[FEAT_XSAVE_COMP_LO] = 0;
-        env->features[FEAT_XSAVE_COMP_HI] = 0;
+        env->features[FEAT_XSAVE_XCR0_LO] = 0;
+        env->features[FEAT_XSAVE_XCR0_HI] = 0;
         return;
     }
 
@@ -6444,8 +6444,8 @@ static void x86_cpu_enable_xsave_components(X86CPU *cpu)
         }
     }
 
-    env->features[FEAT_XSAVE_COMP_LO] = mask;
-    env->features[FEAT_XSAVE_COMP_HI] = mask >> 32;
+    env->features[FEAT_XSAVE_XCR0_LO] = mask;
+    env->features[FEAT_XSAVE_XCR0_HI] = mask >> 32;
 }
 
 /***** Steps involved on loading and filtering CPUID data
diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index 570f916878..84cb6adcaa 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -526,8 +526,8 @@ typedef enum FeatureWord {
     FEAT_SVM,           /* CPUID[8000_000A].EDX */
     FEAT_XSAVE,         /* CPUID[EAX=0xd,ECX=1].EAX */
     FEAT_6_EAX,         /* CPUID[6].EAX */
-    FEAT_XSAVE_COMP_LO, /* CPUID[EAX=0xd,ECX=0].EAX */
-    FEAT_XSAVE_COMP_HI, /* CPUID[EAX=0xd,ECX=0].EDX */
+    FEAT_XSAVE_XCR0_LO, /* CPUID[EAX=0xd,ECX=0].EAX */
+    FEAT_XSAVE_XCR0_HI, /* CPUID[EAX=0xd,ECX=0].EDX */
     FEAT_ARCH_CAPABILITIES,
     FEAT_CORE_CAPABILITY,
     FEAT_PERF_CAPABILITIES,
-- 
2.26.2

[PATCH v8 2/6] target/i386: Enable XSS feature CPUID enumeration

[Yang Weijiang]

Currently, CPUID.(EAX=0DH,ECX=01H) doesn't enumerate features in XSS
properly, so enable the support. XCR0 bits indicate user-mode XSAVE
components, and XSS bits indicate supervisor-mode XSAVE components.

Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
---
 target/i386/cpu.c | 68 +++++++++++++++++++++++++++++++++++++++--------
 target/i386/cpu.h |  9 +++++++
 2 files changed, 66 insertions(+), 11 deletions(-)

diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index 5c76186883..d74d68e319 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -1062,6 +1062,24 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
         },
         .tcg_features = TCG_XSAVE_FEATURES,
     },
+    [FEAT_XSAVE_XSS_LO] = {
+        .type = CPUID_FEATURE_WORD,
+        .cpuid = {
+            .eax = 0xD,
+            .needs_ecx = true,
+            .ecx = 1,
+            .reg = R_ECX,
+        },
+    },
+    [FEAT_XSAVE_XSS_HI] = {
+        .type = CPUID_FEATURE_WORD,
+        .cpuid = {
+            .eax = 0xD,
+            .needs_ecx = true,
+            .ecx = 1,
+            .reg = R_EDX
+        },
+    },
     [FEAT_6_EAX] = {
         .type = CPUID_FEATURE_WORD,
         .feat_names = {
@@ -1453,6 +1471,9 @@ typedef struct ExtSaveArea {
     uint32_t offset, size;
 } ExtSaveArea;
 
+/* CPUID feature bits available in XSS */
+#define CPUID_XSTATE_XSS_MASK    0
+
 static const ExtSaveArea x86_ext_save_areas[] = {
     [XSTATE_FP_BIT] = {
         /* x87 FP state component is always enabled if XSAVE is supported */
@@ -1498,15 +1519,18 @@ static const ExtSaveArea x86_ext_save_areas[] = {
             .size = sizeof(XSavePKRU) },
 };
 
-static uint32_t xsave_area_size(uint64_t mask)
+static uint32_t xsave_area_size(uint64_t mask, bool compacted)
 {
+    uint64_t ret = x86_ext_save_areas[0].size;
+    const ExtSaveArea *esa;
+    uint32_t offset = 0;
     int i;
-    uint64_t ret = 0;
 
-    for (i = 0; i < ARRAY_SIZE(x86_ext_save_areas); i++) {
-        const ExtSaveArea *esa = &x86_ext_save_areas[i];
+    for (i = 2; i < ARRAY_SIZE(x86_ext_save_areas); i++) {
+        esa = &x86_ext_save_areas[i];
         if ((mask >> i) & 1) {
-            ret = MAX(ret, esa->offset + esa->size);
+            offset = compacted ? ret : esa->offset;
+            ret = MAX(ret, offset + esa->size);
         }
     }
     return ret;
@@ -1517,7 +1541,7 @@ static inline bool accel_uses_host_cpuid(void)
     return kvm_enabled() || hvf_enabled();
 }
 
-static inline uint64_t x86_cpu_xsave_components(X86CPU *cpu)
+static inline uint64_t x86_cpu_xsave_xcr0_components(X86CPU *cpu)
 {
     return ((uint64_t)cpu->env.features[FEAT_XSAVE_XCR0_HI]) << 32 |
            cpu->env.features[FEAT_XSAVE_XCR0_LO];
@@ -1532,6 +1556,12 @@ static const char *get_register_name_32(unsigned int reg)
     return x86_reg_info_32[reg].name;
 }
 
+static inline uint64_t x86_cpu_xsave_xss_components(X86CPU *cpu)
+{
+    return ((uint64_t)cpu->env.features[FEAT_XSAVE_XSS_HI]) << 32 |
+           cpu->env.features[FEAT_XSAVE_XSS_LO];
+}
+
 /*
  * Returns the set of feature flags that are supported and migratable by
  * QEMU, for a given FeatureWord.
@@ -5859,7 +5889,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
         }
 
         if (count == 0) {
-            *ecx = xsave_area_size(x86_cpu_xsave_components(cpu));
+            *ecx = xsave_area_size(x86_cpu_xsave_xcr0_components(cpu), false);
             *eax = env->features[FEAT_XSAVE_XCR0_LO];
             *edx = env->features[FEAT_XSAVE_XCR0_HI];
             /*
@@ -5868,14 +5898,25 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
              * even through guest update xcr0, this will crash some legacy guest
              * (e.g., CentOS 6), So set ebx == ecx to workaroud it.
              */
-            *ebx = kvm_enabled() ? *ecx : xsave_area_size(env->xcr0);
+            *ebx = kvm_enabled() ? *ecx : xsave_area_size(env->xcr0, false);
         } else if (count == 1) {
+            uint64_t xstate = x86_cpu_xsave_xcr0_components(cpu) |
+                              x86_cpu_xsave_xss_components(cpu);
+
             *eax = env->features[FEAT_XSAVE];
+            *ebx = xsave_area_size(xstate, true);
+            *ecx = env->features[FEAT_XSAVE_XSS_LO];
+            *edx = env->features[FEAT_XSAVE_XSS_HI];
         } else if (count < ARRAY_SIZE(x86_ext_save_areas)) {
-            if ((x86_cpu_xsave_components(cpu) >> count) & 1) {
-                const ExtSaveArea *esa = &x86_ext_save_areas[count];
+            const ExtSaveArea *esa = &x86_ext_save_areas[count];
+
+            if ((x86_cpu_xsave_xcr0_components(cpu) >> count) & 1) {
                 *eax = esa->size;
                 *ebx = esa->offset;
+            } else if ((x86_cpu_xsave_xss_components(cpu) >> count) & 1) {
+                *eax = esa->size;
+                *ebx = 0;
+                *ecx = 1;
             }
         }
         break;
@@ -6206,6 +6247,9 @@ static void x86_cpu_reset(DeviceState *dev)
     }
     for (i = 2; i < ARRAY_SIZE(x86_ext_save_areas); i++) {
         const ExtSaveArea *esa = &x86_ext_save_areas[i];
+        if (!((1 << i) & CPUID_XSTATE_XCR0_MASK)) {
+            continue;
+        }
         if (env->features[esa->feature] & esa->bits) {
             xcr0 |= 1ull << i;
         }
@@ -6444,8 +6488,10 @@ static void x86_cpu_enable_xsave_components(X86CPU *cpu)
         }
     }
 
-    env->features[FEAT_XSAVE_XCR0_LO] = mask;
+    env->features[FEAT_XSAVE_XCR0_LO] = mask & CPUID_XSTATE_XCR0_MASK;
     env->features[FEAT_XSAVE_XCR0_HI] = mask >> 32;
+    env->features[FEAT_XSAVE_XSS_LO] = mask & CPUID_XSTATE_XSS_MASK;
+    env->features[FEAT_XSAVE_XSS_HI] = mask >> 32;
 }
 
 /***** Steps involved on loading and filtering CPUID data
diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index 84cb6adcaa..42f835d455 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -503,6 +503,13 @@ typedef enum X86Seg {
 #define XSTATE_Hi16_ZMM_MASK            (1ULL << XSTATE_Hi16_ZMM_BIT)
 #define XSTATE_PKRU_MASK                (1ULL << XSTATE_PKRU_BIT)
 
+/* CPUID feature bits available in XCR0 */
+#define CPUID_XSTATE_XCR0_MASK  (XSTATE_FP_MASK | XSTATE_SSE_MASK | \
+                                 XSTATE_YMM_MASK | XSTATE_BNDREGS_MASK | \
+                                 XSTATE_BNDCSR_MASK | XSTATE_OPMASK_MASK | \
+                                 XSTATE_ZMM_Hi256_MASK | \
+                                 XSTATE_Hi16_ZMM_MASK | XSTATE_PKRU_MASK)
+
 /* CPUID feature words */
 typedef enum FeatureWord {
     FEAT_1_EDX,         /* CPUID[1].EDX */
@@ -541,6 +548,8 @@ typedef enum FeatureWord {
     FEAT_VMX_BASIC,
     FEAT_VMX_VMFUNC,
     FEAT_14_0_ECX,
+    FEAT_XSAVE_XSS_LO,     /* CPUID[EAX=0xd,ECX=1].ECX */
+    FEAT_XSAVE_XSS_HI,     /* CPUID[EAX=0xd,ECX=1].EDX */
     FEATURE_WORDS,
 } FeatureWord;
 
-- 
2.26.2

[PATCH v8 3/6] target/i386: Enable XSAVES support for CET states

[Yang Weijiang]

CET Shadow Stack(SHSTK) and Indirect Branch Tracking(IBT) are enumerated
via CPUID.(EAX=07H,ECX=0H):ECX[bit 7] and EDX[bit 20] respectively.
Two CET bits (bit 11 and 12) are defined in MSR_IA32_XSS for XSAVES.
They correspond to CET states in user and supervisor mode respectively.

Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
---
 target/i386/cpu.c | 42 +++++++++++++++++++++++++++++++++++++++++-
 target/i386/cpu.h | 21 +++++++++++++++++++++
 2 files changed, 62 insertions(+), 1 deletion(-)

diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index d74d68e319..bae827c8d5 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -1064,6 +1064,16 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
     },
     [FEAT_XSAVE_XSS_LO] = {
         .type = CPUID_FEATURE_WORD,
+        .feat_names = {
+            NULL, NULL, NULL, NULL,
+            NULL, NULL, NULL, NULL,
+            NULL, NULL, NULL, "cet-u",
+            "cet-s", NULL, NULL, NULL,
+            NULL, NULL, NULL, NULL,
+            NULL, NULL, NULL, NULL,
+            NULL, NULL, NULL, NULL,
+            NULL, NULL, NULL, NULL,
+        },
         .cpuid = {
             .eax = 0xD,
             .needs_ecx = true,
@@ -1472,7 +1482,7 @@ typedef struct ExtSaveArea {
 } ExtSaveArea;
 
 /* CPUID feature bits available in XSS */
-#define CPUID_XSTATE_XSS_MASK    0
+#define CPUID_XSTATE_XSS_MASK    (XSTATE_CET_U_MASK)
 
 static const ExtSaveArea x86_ext_save_areas[] = {
     [XSTATE_FP_BIT] = {
@@ -1517,6 +1527,19 @@ static const ExtSaveArea x86_ext_save_areas[] = {
           { .feature = FEAT_7_0_ECX, .bits = CPUID_7_0_ECX_PKU,
             .offset = offsetof(X86XSaveArea, pkru_state),
             .size = sizeof(XSavePKRU) },
+    [XSTATE_CET_U_BIT] = {
+            .feature = FEAT_7_0_ECX, .bits = CPUID_7_0_ECX_CET_SHSTK,
+        /*
+         * The features enabled in XSS MSR always use compacted format
+         * to store the data, in this case .offset == 0. And CET bits
+         * fall into this category.
+         */
+            .offset = 0,
+            .size = sizeof(XSavesCETU) },
+    [XSTATE_CET_S_BIT] = {
+            .feature = FEAT_7_0_ECX, .bits = CPUID_7_0_ECX_CET_SHSTK,
+            .offset = 0,
+            .size = sizeof(XSavesCETS) },
 };
 
 static uint32_t xsave_area_size(uint64_t mask, bool compacted)
@@ -6486,6 +6509,23 @@ static void x86_cpu_enable_xsave_components(X86CPU *cpu)
         if (env->features[esa->feature] & esa->bits) {
             mask |= (1ULL << i);
         }
+
+        /*
+         * Both CET SHSTK and IBT feature requires XSAVES support, but two
+         * features can be controlled independently by kernel, and we only
+         * have one correlated bit set in x86_ext_save_areas, so if either
+         * of two features is enabled, we set the XSAVES support bit to make
+         * the enabled feature work.
+         */
+        if (i == XSTATE_CET_U_BIT || i == XSTATE_CET_S_BIT) {
+            uint64_t ecx = env->features[FEAT_7_0_ECX];
+            uint64_t edx = env->features[FEAT_7_0_EDX];
+
+            if ((ecx & CPUID_7_0_ECX_CET_SHSTK) ||
+                (edx & CPUID_7_0_EDX_CET_IBT)) {
+                mask |= (1ULL << i);
+            }
+        }
     }
 
     env->features[FEAT_XSAVE_XCR0_LO] = mask & CPUID_XSTATE_XCR0_MASK;
diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index 42f835d455..593a2d6823 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -492,6 +492,8 @@ typedef enum X86Seg {
 #define XSTATE_ZMM_Hi256_BIT            6
 #define XSTATE_Hi16_ZMM_BIT             7
 #define XSTATE_PKRU_BIT                 9
+#define XSTATE_CET_U_BIT                11
+#define XSTATE_CET_S_BIT                12
 
 #define XSTATE_FP_MASK                  (1ULL << XSTATE_FP_BIT)
 #define XSTATE_SSE_MASK                 (1ULL << XSTATE_SSE_BIT)
@@ -502,6 +504,8 @@ typedef enum X86Seg {
 #define XSTATE_ZMM_Hi256_MASK           (1ULL << XSTATE_ZMM_Hi256_BIT)
 #define XSTATE_Hi16_ZMM_MASK            (1ULL << XSTATE_Hi16_ZMM_BIT)
 #define XSTATE_PKRU_MASK                (1ULL << XSTATE_PKRU_BIT)
+#define XSTATE_CET_U_MASK               (1ULL << XSTATE_CET_U_BIT)
+#define XSTATE_CET_S_MASK               (1ULL << XSTATE_CET_S_BIT)
 
 /* CPUID feature bits available in XCR0 */
 #define CPUID_XSTATE_XCR0_MASK  (XSTATE_FP_MASK | XSTATE_SSE_MASK | \
@@ -761,6 +765,8 @@ typedef uint64_t FeatureWordArray[FEATURE_WORDS];
 #define CPUID_7_0_ECX_WAITPKG           (1U << 5)
 /* Additional AVX-512 Vector Byte Manipulation Instruction */
 #define CPUID_7_0_ECX_AVX512_VBMI2      (1U << 6)
+/* CET SHSTK feature */
+#define CPUID_7_0_ECX_CET_SHSTK         (1U << 7)
 /* Galois Field New Instructions */
 #define CPUID_7_0_ECX_GFNI              (1U << 8)
 /* Vector AES Instructions */
@@ -800,6 +806,8 @@ typedef uint64_t FeatureWordArray[FEATURE_WORDS];
 #define CPUID_7_0_EDX_SERIALIZE         (1U << 14)
 /* TSX Suspend Load Address Tracking instruction */
 #define CPUID_7_0_EDX_TSX_LDTRK         (1U << 16)
+/* CET IBT feature */
+#define CPUID_7_0_EDX_CET_IBT           (1U << 20)
 /* AVX512_FP16 instruction */
 #define CPUID_7_0_EDX_AVX512_FP16       (1U << 23)
 /* Speculation Control */
@@ -1301,6 +1309,19 @@ typedef struct XSavePKRU {
     uint32_t padding;
 } XSavePKRU;
 
+/* Ext. save area 11: User mode CET state */
+typedef struct XSavesCETU {
+    uint64_t u_cet;
+    uint64_t user_ssp;
+} XSavesCETU;
+
+/* Ext. save area 12: Supervisor mode CET state */
+typedef struct XSavesCETS {
+    uint64_t kernel_ssp;
+    uint64_t pl1_ssp;
+    uint64_t pl2_ssp;
+} XSavesCETS;
+
 typedef struct X86XSaveArea {
     X86LegacyXSaveArea legacy;
     X86XSaveHeader header;
-- 
2.26.2

[PATCH v8 4/6] target/i386: Add user-space MSR access interface for CET

[Yang Weijiang]

CET states are divided into user-mode and supervisor-mode states,
MSR_KVM_GUEST_SSP holds current SHSTK pointer in use, MSR_IA32_U_CET and
MSR_IA32_PL3_SSP are for user-mode states, others are for supervisor-mode
states. Expose access according to current CET supported bits in CPUID
and XSS.

Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
---
 target/i386/cpu.h     | 18 +++++++++++
 target/i386/kvm/kvm.c | 72 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 90 insertions(+)

diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index 593a2d6823..a5bb049d4e 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -483,6 +483,15 @@ typedef enum X86Seg {
 #define MSR_IA32_VMX_TRUE_ENTRY_CTLS     0x00000490
 #define MSR_IA32_VMX_VMFUNC             0x00000491
 
+#define MSR_IA32_U_CET                  0x000006a0
+#define MSR_IA32_S_CET                  0x000006a2
+#define MSR_IA32_PL0_SSP                0x000006a4
+#define MSR_IA32_PL1_SSP                0x000006a5
+#define MSR_IA32_PL2_SSP                0x000006a6
+#define MSR_IA32_PL3_SSP                0x000006a7
+#define MSR_IA32_SSP_TBL                0x000006a8
+#define MSR_KVM_GUEST_SSP               0x4b564d08
+
 #define XSTATE_FP_BIT                   0
 #define XSTATE_SSE_BIT                  1
 #define XSTATE_YMM_BIT                  2
@@ -1601,6 +1610,15 @@ typedef struct CPUX86State {
 
     uintptr_t retaddr;
 
+    uint64_t u_cet;
+    uint64_t s_cet;
+    uint64_t pl0_ssp;
+    uint64_t pl1_ssp;
+    uint64_t pl2_ssp;
+    uint64_t pl3_ssp;
+    uint64_t ssp_tbl;
+    uint64_t guest_ssp;
+
     /* Fields up to this point are cleared by a CPU reset */
     struct {} end_reset_fields;
 
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index 7fe9f52710..c36ff7ffec 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -3047,6 +3047,30 @@ static int kvm_put_msrs(X86CPU *cpu, int level)
         }
     }
 
+    if (((env->features[FEAT_7_0_ECX] & CPUID_7_0_ECX_CET_SHSTK) ||
+        (env->features[FEAT_7_0_EDX] & CPUID_7_0_EDX_CET_IBT)) &&
+        (env->features[FEAT_XSAVE_XSS_LO] & XSTATE_CET_U_MASK)) {
+        kvm_msr_entry_add(cpu, MSR_IA32_U_CET, env->u_cet);
+        kvm_msr_entry_add(cpu, MSR_IA32_PL3_SSP, env->pl3_ssp);
+    }
+
+    if (env->features[FEAT_XSAVE_XSS_LO] & XSTATE_CET_S_MASK) {
+        if (env->features[FEAT_7_0_ECX] & CPUID_7_0_ECX_CET_SHSTK) {
+            kvm_msr_entry_add(cpu, MSR_IA32_PL0_SSP, env->pl0_ssp);
+            kvm_msr_entry_add(cpu, MSR_IA32_PL1_SSP, env->pl1_ssp);
+            kvm_msr_entry_add(cpu, MSR_IA32_PL2_SSP, env->pl2_ssp);
+            kvm_msr_entry_add(cpu, MSR_IA32_SSP_TBL, env->ssp_tbl);
+        }
+
+        kvm_msr_entry_add(cpu, MSR_IA32_S_CET, env->s_cet);
+    }
+
+    if ((env->features[FEAT_7_0_ECX] & CPUID_7_0_ECX_CET_SHSTK) &&
+        (env->features[FEAT_XSAVE_XSS_LO] & (XSTATE_CET_U_MASK |
+        XSTATE_CET_S_MASK))) {
+        kvm_msr_entry_add(cpu, MSR_KVM_GUEST_SSP, env->guest_ssp);
+    }
+
     return kvm_buf_set_msrs(cpu);
 }
 
@@ -3369,6 +3393,30 @@ static int kvm_get_msrs(X86CPU *cpu)
         }
     }
 
+    if (((env->features[FEAT_7_0_ECX] & CPUID_7_0_ECX_CET_SHSTK) ||
+        (env->features[FEAT_7_0_EDX] & CPUID_7_0_EDX_CET_IBT)) &&
+        (env->features[FEAT_XSAVE_XSS_LO] & XSTATE_CET_U_MASK)) {
+        kvm_msr_entry_add(cpu, MSR_IA32_U_CET, 0);
+        kvm_msr_entry_add(cpu, MSR_IA32_PL3_SSP, 0);
+    }
+
+    if (env->features[FEAT_XSAVE_XSS_LO] & XSTATE_CET_S_MASK) {
+        if (env->features[FEAT_7_0_ECX] & CPUID_7_0_ECX_CET_SHSTK) {
+            kvm_msr_entry_add(cpu, MSR_IA32_PL0_SSP, 0);
+            kvm_msr_entry_add(cpu, MSR_IA32_PL1_SSP, 0);
+            kvm_msr_entry_add(cpu, MSR_IA32_PL2_SSP, 0);
+            kvm_msr_entry_add(cpu, MSR_IA32_SSP_TBL, 0);
+        }
+
+        kvm_msr_entry_add(cpu, MSR_IA32_S_CET, 0);
+    }
+
+    if ((env->features[FEAT_7_0_ECX] & CPUID_7_0_ECX_CET_SHSTK) &&
+        (env->features[FEAT_XSAVE_XSS_LO] & (XSTATE_CET_U_MASK |
+        XSTATE_CET_S_MASK))) {
+        kvm_msr_entry_add(cpu, MSR_KVM_GUEST_SSP, 0);
+    }
+
     ret = kvm_vcpu_ioctl(CPU(cpu), KVM_GET_MSRS, cpu->kvm_msr_buf);
     if (ret < 0) {
         return ret;
@@ -3658,6 +3706,30 @@ static int kvm_get_msrs(X86CPU *cpu)
         case MSR_IA32_RTIT_ADDR0_A ... MSR_IA32_RTIT_ADDR3_B:
             env->msr_rtit_addrs[index - MSR_IA32_RTIT_ADDR0_A] = msrs[i].data;
             break;
+        case MSR_IA32_U_CET:
+            env->u_cet = msrs[i].data;
+            break;
+        case MSR_IA32_S_CET:
+            env->s_cet = msrs[i].data;
+            break;
+        case MSR_IA32_PL0_SSP:
+            env->pl0_ssp = msrs[i].data;
+            break;
+        case MSR_IA32_PL1_SSP:
+            env->pl1_ssp = msrs[i].data;
+            break;
+        case MSR_IA32_PL2_SSP:
+            env->pl2_ssp = msrs[i].data;
+            break;
+        case MSR_IA32_PL3_SSP:
+            env->pl3_ssp = msrs[i].data;
+            break;
+        case MSR_IA32_SSP_TBL:
+            env->ssp_tbl = msrs[i].data;
+            break;
+        case MSR_KVM_GUEST_SSP:
+            env->guest_ssp = msrs[i].data;
+            break;
         }
     }
 
-- 
2.26.2

[PATCH v8 5/6] target/i386: Add CET state support for guest migration

[Yang Weijiang]

Save the MSRs being used on source machine and restore them
on destination machine.

Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
---
 target/i386/machine.c | 161 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 161 insertions(+)

diff --git a/target/i386/machine.c b/target/i386/machine.c
index 137604ddb8..4d63340931 100644
--- a/target/i386/machine.c
+++ b/target/i386/machine.c
@@ -1000,6 +1000,159 @@ static const VMStateDescription vmstate_umwait = {
     }
 };
 
+static bool u_cet_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->u_cet != 0;
+}
+
+static const VMStateDescription vmstate_u_cet = {
+    .name = "cpu/u_cet",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = u_cet_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.u_cet, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+static bool s_cet_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->s_cet != 0;
+}
+
+static const VMStateDescription vmstate_s_cet = {
+    .name = "cpu/s_cet",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = s_cet_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.s_cet, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+static bool pl0_ssp_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->pl0_ssp != 0;
+}
+
+static const VMStateDescription vmstate_pl0_ssp = {
+    .name = "cpu/pl0_ssp",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = pl0_ssp_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.pl0_ssp, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+static bool pl1_ssp_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->pl1_ssp != 0;
+}
+
+static const VMStateDescription vmstate_pl1_ssp = {
+    .name = "cpu/pl1_ssp",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = pl1_ssp_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.pl1_ssp, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+static bool pl2_ssp_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->pl2_ssp != 0;
+}
+
+static const VMStateDescription vmstate_pl2_ssp = {
+    .name = "cpu/pl2_ssp",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = pl2_ssp_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.pl2_ssp, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+
+static bool pl3_ssp_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->pl3_ssp != 0;
+}
+
+static const VMStateDescription vmstate_pl3_ssp = {
+    .name = "cpu/pl3_ssp",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = pl3_ssp_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.pl3_ssp, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+static bool ssp_tbl_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->ssp_tbl != 0;
+}
+
+static const VMStateDescription vmstate_ssp_tbl = {
+    .name = "cpu/ssp_tbl",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = ssp_tbl_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.ssp_tbl, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
+static bool guest_ssp_needed(void *opaque)
+{
+    X86CPU *cpu = opaque;
+    CPUX86State *env = &cpu->env;
+
+    return env->guest_ssp != 0;
+}
+
+static const VMStateDescription vmstate_guest_ssp = {
+    .name = "cpu/guest_ssp",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = guest_ssp_needed,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(env.guest_ssp, X86CPU),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
 static bool pkru_needed(void *opaque)
 {
     X86CPU *cpu = opaque;
@@ -1531,6 +1684,14 @@ VMStateDescription vmstate_x86_cpu = {
         &vmstate_nested_state,
 #endif
         &vmstate_msr_tsx_ctrl,
+        &vmstate_u_cet,
+        &vmstate_s_cet,
+        &vmstate_pl0_ssp,
+        &vmstate_pl1_ssp,
+        &vmstate_pl2_ssp,
+        &vmstate_pl3_ssp,
+        &vmstate_ssp_tbl,
+        &vmstate_guest_ssp,
         NULL
     }
 };
-- 
2.26.2

[PATCH v8 6/6] target/i386: Advise CET bits in CPU/MSR feature words

[Yang Weijiang]

CET SHSTK and IBT feature are enumerated via CPUID.(EAX=07H,ECX=0H):ECX[bit 7]
and EDX[bit 20]. CET state load/restore at vmentry/vmexit are enabled via
VMX_ENTRY_CTLS[bit 20] and VMX_EXIT_CTLS[bit 28].

Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
---
 target/i386/cpu.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index bae827c8d5..b432b681d8 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -958,7 +958,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
         .type = CPUID_FEATURE_WORD,
         .feat_names = {
             NULL, "avx512vbmi", "umip", "pku",
-            NULL /* ospke */, "waitpkg", "avx512vbmi2", NULL,
+            NULL /* ospke */, "waitpkg", "avx512vbmi2", "shstk",
             "gfni", "vaes", "vpclmulqdq", "avx512vnni",
             "avx512bitalg", NULL, "avx512-vpopcntdq", NULL,
             "la57", NULL, NULL, NULL,
@@ -981,7 +981,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
             "avx512-vp2intersect", NULL, "md-clear", NULL,
             NULL, NULL, "serialize", NULL,
             "tsx-ldtrk", NULL, NULL /* pconfig */, NULL,
-            NULL, NULL, NULL, "avx512-fp16",
+            "ibt", NULL, NULL, "avx512-fp16",
             NULL, NULL, "spec-ctrl", "stibp",
             NULL, "arch-capabilities", "core-capability", "ssbd",
         },
@@ -1243,7 +1243,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
             "vmx-exit-save-efer", "vmx-exit-load-efer",
                 "vmx-exit-save-preemption-timer", "vmx-exit-clear-bndcfgs",
             NULL, "vmx-exit-clear-rtit-ctl", NULL, NULL,
-            NULL, "vmx-exit-load-pkrs", NULL, NULL,
+            "vmx-exit-save-cet-ctl", "vmx-exit-load-pkrs", NULL, NULL,
         },
         .msr = {
             .index = MSR_IA32_VMX_TRUE_EXIT_CTLS,
@@ -1258,7 +1258,7 @@ static FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
             NULL, "vmx-entry-ia32e-mode", NULL, NULL,
             NULL, "vmx-entry-load-perf-global-ctrl", "vmx-entry-load-pat", "vmx-entry-load-efer",
             "vmx-entry-load-bndcfgs", NULL, "vmx-entry-load-rtit-ctl", NULL,
-            NULL, NULL, "vmx-entry-load-pkrs", NULL,
+            "vmx-entry-load-cet-ctl", NULL, "vmx-entry-load-pkrs", NULL,
             NULL, NULL, NULL, NULL,
             NULL, NULL, NULL, NULL,
         },
-- 
2.26.2