[PATCH 25/30] x86emul: support APX+MOVBE insn

[Jan Beulich <jbeulich@xxxxxxxx>]

Besides adding table entries and case labels there's very little to do
here; generic map 4 handling takes care of things.

Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>

--- a/tools/tests/x86_emulator/evex-disp8.c
+++ b/tools/tests/x86_emulator/evex-disp8.c
@@ -1159,6 +1159,13 @@ static const struct test lzcnt[] = {
     INSN(lzcntw,    66, map4, f5,    el,    w, no),
 };
 
+static const struct test movbe[] = {
+    INSN(movbe,       , map4, 60,    el,   dq, no),
+    INSN(movbe,       , map4, 61,    el,   dq, no),
+    INSN(movbew,    66, map4, 60,    el,    w, no),
+    INSN(movbew,    66, map4, 61,    el,    w, no),
+};
+
 static const struct test popcnt[] = {
     INSN(popcnt,      , map4, 88,    el,   dq, no),
     INSN(popcntw,   66, map4, 88,    el,    w, no),
@@ -1756,6 +1763,7 @@ void evex_disp8_test(void *instr, struct
         RUN(bmi2);
         RUN(cmpccxadd);
         RUN(lzcnt);
+        RUN(movbe);
         RUN(popcnt);
 
 #undef RUN
--- a/tools/tests/x86_emulator/predicates.c
+++ b/tools/tests/x86_emulator/predicates.c
@@ -2202,6 +2202,10 @@ static const struct evex {
     CND(0x40,   2, T, R, pfx_no, Wn, L0, ND ), /* {,cf}cmov<cc> */ \
     CND(0x40,   2, T, R, pfx_66, Wn, L0, ND ), /* {,cf}cmov<cc> */ \
     CND(0x40,   2, T, W, pfx_f2, WIG, L0 ), /* set<cc> */
+    { { 0x60 }, 2, T, R, pfx_no, Wn, L0 }, /* movbe */
+    { { 0x60 }, 2, T, R, pfx_66, Wn, L0 }, /* movbe */
+    { { 0x61 }, 2, T, W, pfx_no, Wn, L0 }, /* movbe */
+    { { 0x61 }, 2, T, W, pfx_66, Wn, L0 }, /* movbe */
     { { 0x66 }, 2, T, R, pfx_66, Wn, L0 }, /* adcx */
     { { 0x66 }, 2, T, R, pfx_f3, Wn, L0 }, /* adox */
     { { 0x69 }, 6, T, R, pfx_no, Wn, L0 }, /* imul */
--- a/tools/tests/x86_emulator/test_x86_emulator.c
+++ b/tools/tests/x86_emulator/test_x86_emulator.c
@@ -2475,6 +2475,35 @@ int main(int argc, char **argv)
         goto fail;
     printf("okay\n");
 
+#ifdef __x86_64__
+    printf("%-40s", "Testing movbe %rax,%rcx...");
+    instr[0] = 0x62; instr[1] = 0xf4; instr[2] = 0xfc; instr[3] = 0x08;
+    instr[4] = 0x60; instr[5] = 0xc8;
+    regs.rip = (unsigned long)&instr[0];
+    rc = x86_emulate(&ctxt, &emulops);
+    if ( (rc != X86EMUL_OKAY) ||
+         (regs.rcx != 0x1234567800000000UL) ||
+         (regs.rip != (unsigned long)&instr[6]) )
+        goto fail;
+    printf("okay\n");
+
+    if ( cpu_has_apx_f )
+    {
+        printf("%-40s", "Testing movbe %r30w,2(%r31)...");
+        instr[1] = 0x4c; instr[2] = 0x7d;
+        instr[4] = 0x61; instr[5] = 0x77; instr[6] = 0x02;
+        regs.rip = (unsigned long)&instr[0];
+        load_egpr(30, 0x12345678);
+        load_egpr(31, (unsigned long)res);
+        rc = x86_emulate(&ctxt, &emulops);
+        if ( (rc != X86EMUL_OKAY) ||
+             (*res != 0x78561234) ||
+             (regs.rip != (unsigned long)&instr[7]) )
+            goto fail;
+        printf("okay\n");
+    }
+#endif
+
     printf("%-40s", "Testing popcnt (%edx),%cx...");
     if ( cpu_has_popcnt )
     {
--- a/tools/tests/x86_emulator/x86-emulate.h
+++ b/tools/tests/x86_emulator/x86-emulate.h
@@ -168,6 +168,7 @@ void wrpkru(unsigned int val);
 #define cpu_has_fma                 (cpu_policy.basic.fma && xcr0_mask(6))
 #define cpu_has_sse4_1               cpu_policy.basic.sse4_1
 #define cpu_has_sse4_2               cpu_policy.basic.sse4_2
+#define cpu_has_movbe                cpu_policy.basic.movbe
 #define cpu_has_popcnt               cpu_policy.basic.popcnt
 #define cpu_has_aesni                cpu_policy.basic.aesni
 #define cpu_has_avx                 (cpu_policy.basic.avx  && xcr0_mask(6))
--- a/xen/arch/x86/x86_emulate/decode.c
+++ b/xen/arch/x86/x86_emulate/decode.c
@@ -691,6 +691,14 @@ static const struct map4_table {
         [vex_f2] = { .to_mem = true, .byte = true, .nd = true,
                      .nci_ndd_nf = true },
     },
+    [0x60] = { /* MOVBE ({,d,q}word) */
+        [vex_np] = { },
+        [vex_66] = { },
+    },
+    [0x61] = { /* MOVBE ({,d,q}word) */
+        [vex_np] = { .to_mem = true },
+        [vex_66] = { .to_mem = true },
+    },
     [0x66] = {
         [vex_66] = { .nd = true }, /* ADCX */
         [vex_f3] = { .nd = true }, /* ADOX */
--- a/xen/arch/x86/x86_emulate/x86_emulate.c
+++ b/xen/arch/x86/x86_emulate/x86_emulate.c
@@ -7345,6 +7345,11 @@ x86_emulate(
     case X86EMUL_OPC(0x0f38, 0xf0): /* movbe m,r */
     case X86EMUL_OPC(0x0f38, 0xf1): /* movbe r,m */
         generate_exception_if(ea.type != OP_MEM, X86_EXC_UD);
+        fallthrough;
+    case X86EMUL_OPC_EVEX   (4, 0x60): /* movbe r/m,reg */
+    case X86EMUL_OPC_EVEX_66(4, 0x60): /* movbe r/m,r16 */
+    case X86EMUL_OPC_EVEX   (4, 0x61): /* movbe reg,r/m */
+    case X86EMUL_OPC_EVEX_66(4, 0x61): /* movbe r16,r/m */
         vcpu_must_have(movbe);
         switch ( op_bytes )
         {