[PATCH v4 02/16] x86emul: support AVX10 forms of SM4 insns

[Jan Beulich <jbeulich@xxxxxxxx>]

Simply clone the VEX-encoded handling to cover the EVEX forms.

Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
---
SDE: -dmr / -future
---
v4: Address v3 TODO. Move ahead in series, switching feature dependency to
    plain AVX10. Add call to avx512_vlen_check().
v3: New.

--- a/tools/tests/x86_emulator/evex-disp8.c
+++ b/tools/tests/x86_emulator/evex-disp8.c
@@ -727,6 +727,11 @@ static const struct test movrs_all[] = {
     INSN(movrsw, f2, map5, 6f, vl,    w, vl),
 };
 
+static const struct test sm4_all[] = {
+    INSN(sm4key4,  f3, 0f38, da, vl, d_nb, vl),
+    INSN(sm4rnds4, f2, 0f38, da, vl, d_nb, vl),
+};
+
 static const unsigned char vl_all[] = { VL_512, VL_128, VL_256 };
 static const unsigned char vl_128[] = { VL_128 };
 static const unsigned char vl_no128[] = { VL_512, VL_256 };
@@ -1124,5 +1129,6 @@ void evex_disp8_test(void *instr, struct
     if ( cpu_has_avx10 )
     {
         run(ctxt->addr_size == 64 && cpu_has_movrs, movrs, all);
+        run(cpu_has_sm4, sm4, all);
     }
 }
--- a/tools/tests/x86_emulator/predicates.c
+++ b/tools/tests/x86_emulator/predicates.c
@@ -2025,6 +2025,8 @@ static const struct evex {
     { { 0xcc }, 2, T, R, pfx_66, Wn, L2 }, /* vrsqrt28p{s,d} */
     { { 0xcd }, 2, T, R, pfx_66, Wn, LIG }, /* vrsqrt28s{s,d} */
     { { 0xcf }, 2, T, R, pfx_66, W0, Ln }, /* vgf2p8mulb */
+    { { 0xda }, 2, T, R, pfx_f3, W0, Ln }, /* vsm4key4 */
+    { { 0xda }, 2, T, R, pfx_f2, W0, Ln }, /* vsm4rnds4 */
     { { 0xdc }, 2, T, R, pfx_66, WIG, Ln }, /* vaesenc */
     { { 0xdd }, 2, T, R, pfx_66, WIG, Ln }, /* vaesenclast */
     { { 0xde }, 2, T, R, pfx_66, WIG, Ln }, /* vaesdec */
--- a/xen/arch/x86/x86_emulate/decode.c
+++ b/xen/arch/x86/x86_emulate/decode.c
@@ -438,7 +438,7 @@ static const struct ext0f38_table {
     [0xd3] = { .simd_size = simd_other },
     [0xd6] = { .simd_size = simd_other, .d8s = d8s_vl },
     [0xd7] = { .simd_size = simd_scalar_vexw, .d8s = d8s_dq },
-    [0xda] = { .simd_size = simd_other },
+    [0xda] = { .simd_size = simd_other, .d8s = d8s_vl },
     [0xdb] = { .simd_size = simd_packed_int, .two_op = 1 },
     [0xdc ... 0xdf] = { .simd_size = simd_packed_int, .d8s = d8s_vl },
     [0xe0 ... 0xef] = { .to_mem = 1 },
--- a/xen/arch/x86/x86_emulate/x86_emulate.c
+++ b/xen/arch/x86/x86_emulate/x86_emulate.c
@@ -6819,6 +6819,15 @@ x86_emulate(
         op_bytes = 16 << vex.l;
         goto simd_0f_ymm;
 
+    case X86EMUL_OPC_EVEX_F3(0x0f38, 0xda): /* vsm4key4 
[xyz]mm/mem,[xyz]mm,[xyz]mm */
+    case X86EMUL_OPC_EVEX_F2(0x0f38, 0xda): /* vsm4rnds4 
[xyz]mm/mem,[xyz]mm,[xyz]mm */
+        vcpu_must_have(avx10);
+        vcpu_must_have(sm4);
+        generate_exception_if(evex.w || evex.brs || evex.opmsk, X86_EXC_UD);
+        avx512_vlen_check(false);
+        op_bytes = 16 << evex.lr;
+        goto simd_zmm;
+
     case X86EMUL_OPC_VEX_66(0x0f38, 0xdc):  /* vaesenc 
{x,y}mm/mem,{x,y}mm,{x,y}mm */
     case X86EMUL_OPC_VEX_66(0x0f38, 0xdd):  /* vaesenclast 
{x,y}mm/mem,{x,y}mm,{x,y}mm */
     case X86EMUL_OPC_VEX_66(0x0f38, 0xde):  /* vaesdec 
{x,y}mm/mem,{x,y}mm,{x,y}mm */