tests/tcg: extend SSE tests to AVX

Extracted from a patch by Paul Brook <paul@nowt.org>.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

tests/tcg/i386/Makefile.target

@@ -107,7 +107,7 @@ run-test-mmx: QEMU_OPTS += -cpu max
 run-plugin-test-mmx: QEMU_OPTS += -cpu max
 test-mmx: test-mmx.h
 
-test-avx: CFLAGS += -masm=intel -O -I.
+test-avx: CFLAGS += -mavx -masm=intel -O -I.
 run-test-avx: QEMU_OPTS += -cpu max
 run-plugin-test-avx: QEMU_OPTS += -cpu max
 test-avx: test-avx.h

tests/tcg/i386/test-avx.c

@@ -6,18 +6,18 @@
 typedef void (*testfn)(void);
 
 typedef struct {
-    uint64_t q0, q1;
+    uint64_t q0, q1, q2, q3;
-} __attribute__((aligned(16))) v2di;
+} __attribute__((aligned(32))) v4di;
 
 typedef struct {
     uint64_t mm[8];
-    v2di xmm[16];
+    v4di ymm[16];
     uint64_t r[16];
     uint64_t flags;
     uint32_t ff;
     uint64_t pad;
-    v2di mem[4];
+    v4di mem[4];
-    v2di mem0[4];
+    v4di mem0[4];
 } reg_state;
 
 typedef struct {
@@ -31,20 +31,20 @@ reg_state initI;
 reg_state initF32;
 reg_state initF64;
 
-static void dump_xmm(const char *name, int n, const v2di *r, int ff)
+static void dump_ymm(const char *name, int n, const v4di *r, int ff)
 {
-    printf("%s%d = %016lx %016lx\n",
+    printf("%s%d = %016lx %016lx %016lx %016lx\n",
-           name, n, r->q1, r->q0);
+           name, n, r->q3, r->q2, r->q1, r->q0);
     if (ff == 64) {
-        double v[2];
+        double v[4];
         memcpy(v, r, sizeof(v));
-        printf("        %16g %16g\n",
+        printf("        %16g %16g %16g %16g\n",
-                v[1], v[0]);
-    } else if (ff == 32) {
-        float v[4];
-        memcpy(v, r, sizeof(v));
-        printf(" %8g %8g %8g %8g\n",
                 v[3], v[2], v[1], v[0]);
+    } else if (ff == 32) {
+        float v[8];
+        memcpy(v, r, sizeof(v));
+        printf(" %8g %8g %8g %8g %8g %8g %8g %8g\n",
+                v[7], v[6], v[5], v[4], v[3], v[2], v[1], v[0]);
     }
 }
 
@@ -53,10 +53,10 @@ static void dump_regs(reg_state *s)
     int i;
 
     for (i = 0; i < 16; i++) {
-        dump_xmm("xmm", i, &s->xmm[i], 0);
+        dump_ymm("ymm", i, &s->ymm[i], 0);
     }
     for (i = 0; i < 4; i++) {
-        dump_xmm("mem", i, &s->mem0[i], 0);
+        dump_ymm("mem", i, &s->mem0[i], 0);
     }
 }
 
@@ -74,13 +74,13 @@ static void compare_state(const reg_state *a, const reg_state *b)
         }
     }
     for (i = 0; i < 16; i++) {
-        if (memcmp(&a->xmm[i], &b->xmm[i], 16)) {
+        if (memcmp(&a->ymm[i], &b->ymm[i], 32)) {
-            dump_xmm("xmm", i, &b->xmm[i], a->ff);
+            dump_ymm("ymm", i, &b->ymm[i], a->ff);
         }
     }
     for (i = 0; i < 4; i++) {
-        if (memcmp(&a->mem0[i], &a->mem[i], 16)) {
+        if (memcmp(&a->mem0[i], &a->mem[i], 32)) {
-            dump_xmm("mem", i, &a->mem[i], a->ff);
+            dump_ymm("mem", i, &a->mem[i], a->ff);
         }
     }
     if (a->flags != b->flags) {
@@ -89,9 +89,9 @@ static void compare_state(const reg_state *a, const reg_state *b)
 }
 
 #define LOADMM(r, o) "movq " #r ", " #o "[%0]\n\t"
-#define LOADXMM(r, o) "movdqa " #r ", " #o "[%0]\n\t"
+#define LOADYMM(r, o) "vmovdqa " #r ", " #o "[%0]\n\t"
 #define STOREMM(r, o) "movq " #o "[%1], " #r "\n\t"
-#define STOREXMM(r, o) "movdqa " #o "[%1], " #r "\n\t"
+#define STOREYMM(r, o) "vmovdqa " #o "[%1], " #r "\n\t"
 #define MMREG(F) \
     F(mm0, 0x00) \
     F(mm1, 0x08) \
@@ -101,39 +101,39 @@ static void compare_state(const reg_state *a, const reg_state *b)
     F(mm5, 0x28) \
     F(mm6, 0x30) \
     F(mm7, 0x38)
-#define XMMREG(F) \
+#define YMMREG(F) \
-    F(xmm0, 0x040) \
+    F(ymm0, 0x040) \
-    F(xmm1, 0x050) \
+    F(ymm1, 0x060) \
-    F(xmm2, 0x060) \
+    F(ymm2, 0x080) \
-    F(xmm3, 0x070) \
+    F(ymm3, 0x0a0) \
-    F(xmm4, 0x080) \
+    F(ymm4, 0x0c0) \
-    F(xmm5, 0x090) \
+    F(ymm5, 0x0e0) \
-    F(xmm6, 0x0a0) \
+    F(ymm6, 0x100) \
-    F(xmm7, 0x0b0) \
+    F(ymm7, 0x120) \
-    F(xmm8, 0x0c0) \
+    F(ymm8, 0x140) \
-    F(xmm9, 0x0d0) \
+    F(ymm9, 0x160) \
-    F(xmm10, 0x0e0) \
+    F(ymm10, 0x180) \
-    F(xmm11, 0x0f0) \
+    F(ymm11, 0x1a0) \
-    F(xmm12, 0x100) \
+    F(ymm12, 0x1c0) \
-    F(xmm13, 0x110) \
+    F(ymm13, 0x1e0) \
-    F(xmm14, 0x120) \
+    F(ymm14, 0x200) \
-    F(xmm15, 0x130)
+    F(ymm15, 0x220)
 #define LOADREG(r, o) "mov " #r ", " #o "[rax]\n\t"
 #define STOREREG(r, o) "mov " #o "[rax], " #r "\n\t"
 #define REG(F) \
-    F(rbx, 0x148) \
+    F(rbx, 0x248) \
-    F(rcx, 0x150) \
+    F(rcx, 0x250) \
-    F(rdx, 0x158) \
+    F(rdx, 0x258) \
-    F(rsi, 0x160) \
+    F(rsi, 0x260) \
-    F(rdi, 0x168) \
+    F(rdi, 0x268) \
-    F(r8, 0x180) \
+    F(r8, 0x280) \
-    F(r9, 0x188) \
+    F(r9, 0x288) \
-    F(r10, 0x190) \
+    F(r10, 0x290) \
-    F(r11, 0x198) \
+    F(r11, 0x298) \
-    F(r12, 0x1a0) \
+    F(r12, 0x2a0) \
-    F(r13, 0x1a8) \
+    F(r13, 0x2a8) \
-    F(r14, 0x1b0) \
+    F(r14, 0x2b0) \
-    F(r15, 0x1b8) \
+    F(r15, 0x2b8) \
 
 static void run_test(const TestDef *t)
 {
@@ -143,7 +143,7 @@ static void run_test(const TestDef *t)
     printf("%5d %s\n", t->n, t->s);
     asm volatile(
             MMREG(LOADMM)
-            XMMREG(LOADXMM)
+            YMMREG(LOADYMM)
             "sub rsp, 128\n\t"
             "push rax\n\t"
             "push rbx\n\t"
@@ -156,26 +156,26 @@ static void run_test(const TestDef *t)
             "pop rbx\n\t"
             "shr rbx, 8\n\t"
             "shl rbx, 8\n\t"
-            "mov rcx, 0x1c0[rax]\n\t"
+            "mov rcx, 0x2c0[rax]\n\t"
             "and rcx, 0xff\n\t"
             "or rbx, rcx\n\t"
             "push rbx\n\t"
             "popf\n\t"
             REG(LOADREG)
-            "mov rax, 0x140[rax]\n\t"
+            "mov rax, 0x240[rax]\n\t"
             "call [rsp]\n\t"
             "mov [rsp], rax\n\t"
             "mov rax, 8[rsp]\n\t"
             REG(STOREREG)
             "mov rbx, [rsp]\n\t"
-            "mov 0x140[rax], rbx\n\t"
+            "mov 0x240[rax], rbx\n\t"
             "mov rbx, 0\n\t"
-            "mov 0x170[rax], rbx\n\t"
+            "mov 0x270[rax], rbx\n\t"
-            "mov 0x178[rax], rbx\n\t"
+            "mov 0x278[rax], rbx\n\t"
             "pushf\n\t"
             "pop rbx\n\t"
             "and rbx, 0xff\n\t"
-            "mov 0x1c0[rax], rbx\n\t"
+            "mov 0x2c0[rax], rbx\n\t"
             "add rsp, 16\n\t"
             "pop rdx\n\t"
             "pop rcx\n\t"
@@ -183,15 +183,15 @@ static void run_test(const TestDef *t)
             "pop rax\n\t"
             "add rsp, 128\n\t"
             MMREG(STOREMM)
-            XMMREG(STOREXMM)
+            YMMREG(STOREYMM)
             : : "r"(init), "r"(&result), "r"(t->fn)
             : "memory", "cc",
             "rsi", "rdi",
             "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
             "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
-            "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5",
+            "ymm0", "ymm1", "ymm2", "ymm3", "ymm4", "ymm5",
-            "xmm6", "xmm7", "xmm8", "xmm9", "xmm10", "xmm11",
+            "ymm6", "ymm7", "ymm8", "ymm9", "ymm10", "ymm11",
-            "xmm12", "xmm13", "xmm14", "xmm15"
+            "ymm12", "ymm13", "ymm14", "ymm15"
             );
     compare_state(init, &result);
 }
@@ -223,22 +223,30 @@ static void run_all(void)
 
 float val_f32[] = {2.0, -1.0, 4.8, 0.8, 3, -42.0, 5e6, 7.5, 8.3};
 double val_f64[] = {2.0, -1.0, 4.8, 0.8, 3, -42.0, 5e6, 7.5};
-v2di val_i64[] = {
+v4di val_i64[] = {
-    {0x3d6b3b6a9e4118f2lu, 0x355ae76d2774d78clu},
+    {0x3d6b3b6a9e4118f2lu, 0x355ae76d2774d78clu,
-    {0xd851c54a56bf1f29lu, 0x4a84d1d50bf4c4fflu},
+     0xac3ff76c4daa4b28lu, 0xe7fabd204cb54083lu},
-    {0x5826475e2c5fd799lu, 0xfd32edc01243f5e9lu},
+    {0xd851c54a56bf1f29lu, 0x4a84d1d50bf4c4fflu,
+     0x56621e553d52b56clu, 0xd0069553da8f584alu},
+    {0x5826475e2c5fd799lu, 0xfd32edc01243f5e9lu,
+     0x738ba2c66d3fe126lu, 0x5707219c6e6c26b4lu},
 };
 
-v2di deadbeef = {0xa5a5a5a5deadbeefull, 0xa5a5a5a5deadbeefull};
+v4di deadbeef = {0xa5a5a5a5deadbeefull, 0xa5a5a5a5deadbeefull,
-v2di indexq = {0x000000000000001full, 0x000000000000008full};
+                 0xa5a5a5a5deadbeefull, 0xa5a5a5a5deadbeefull};
-v2di indexd = {0x00000002000000efull, 0xfffffff500000010ull};
+v4di indexq = {0x000000000000001full, 0x000000000000008full,
+               0xffffffffffffffffull, 0xffffffffffffff5full};
+v4di indexd = {0x00000002000000efull, 0xfffffff500000010ull,
+               0x0000000afffffff0ull, 0x000000000000000eull};
 
-void init_f32reg(v2di *r)
+v4di gather_mem[0x20];
+
+void init_f32reg(v4di *r)
 {
     static int n;
-    float v[4];
+    float v[8];
     int i;
-    for (i = 0; i < 4; i++) {
+    for (i = 0; i < 8; i++) {
         v[i] = val_f32[n++];
         if (n == ARRAY_LEN(val_f32)) {
             n = 0;
@@ -247,12 +255,12 @@ void init_f32reg(v2di *r)
     memcpy(r, v, sizeof(*r));
 }
 
-void init_f64reg(v2di *r)
+void init_f64reg(v4di *r)
 {
     static int n;
-    double v[2];
+    double v[4];
     int i;
-    for (i = 0; i < 2; i++) {
+    for (i = 0; i < 4; i++) {
         v[i] = val_f64[n++];
         if (n == ARRAY_LEN(val_f64)) {
             n = 0;
@@ -261,13 +269,15 @@ void init_f64reg(v2di *r)
     memcpy(r, v, sizeof(*r));
 }
 
-void init_intreg(v2di *r)
+void init_intreg(v4di *r)
 {
     static uint64_t mask;
     static int n;
 
     r->q0 = val_i64[n].q0 ^ mask;
     r->q1 = val_i64[n].q1 ^ mask;
+    r->q2 = val_i64[n].q2 ^ mask;
+    r->q3 = val_i64[n].q3 ^ mask;
     n++;
     if (n == ARRAY_LEN(val_i64)) {
         n = 0;
@@ -280,46 +290,53 @@ static void init_all(reg_state *s)
     int i;
 
     s->r[3] = (uint64_t)&s->mem[0]; /* rdx */
+    s->r[4] = (uint64_t)&gather_mem[ARRAY_LEN(gather_mem) / 2]; /* rsi */
     s->r[5] = (uint64_t)&s->mem[2]; /* rdi */
     s->flags = 2;
-    for (i = 0; i < 8; i++) {
+    for (i = 0; i < 16; i++) {
-        s->xmm[i] = deadbeef;
+        s->ymm[i] = deadbeef;
     }
-    s->xmm[13] = indexd;
+    s->ymm[13] = indexd;
-    s->xmm[14] = indexq;
+    s->ymm[14] = indexq;
-    for (i = 0; i < 2; i++) {
+    for (i = 0; i < 4; i++) {
         s->mem0[i] = deadbeef;
     }
 }
 
 int main(int argc, char *argv[])
 {
+    int i;
+
     init_all(&initI);
-    init_intreg(&initI.xmm[10]);
+    init_intreg(&initI.ymm[10]);
-    init_intreg(&initI.xmm[11]);
+    init_intreg(&initI.ymm[11]);
-    init_intreg(&initI.xmm[12]);
+    init_intreg(&initI.ymm[12]);
     init_intreg(&initI.mem0[1]);
     printf("Int:\n");
     dump_regs(&initI);
 
     init_all(&initF32);
-    init_f32reg(&initF32.xmm[10]);
+    init_f32reg(&initF32.ymm[10]);
-    init_f32reg(&initF32.xmm[11]);
+    init_f32reg(&initF32.ymm[11]);
-    init_f32reg(&initF32.xmm[12]);
+    init_f32reg(&initF32.ymm[12]);
     init_f32reg(&initF32.mem0[1]);
     initF32.ff = 32;
     printf("F32:\n");
     dump_regs(&initF32);
 
     init_all(&initF64);
-    init_f64reg(&initF64.xmm[10]);
+    init_f64reg(&initF64.ymm[10]);
-    init_f64reg(&initF64.xmm[11]);
+    init_f64reg(&initF64.ymm[11]);
-    init_f64reg(&initF64.xmm[12]);
+    init_f64reg(&initF64.ymm[12]);
     init_f64reg(&initF64.mem0[1]);
     initF64.ff = 64;
     printf("F64:\n");
     dump_regs(&initF64);
 
+    for (i = 0; i < ARRAY_LEN(gather_mem); i++) {
+        init_intreg(&gather_mem[i]);
+    }
+
     if (argc > 1) {
         int n = atoi(argv[1]);
         run_test(&test_table[n]);

tests/tcg/i386/test-avx.py

@@ -8,6 +8,7 @@ from fnmatch import fnmatch
 
 archs = [
     "SSE", "SSE2", "SSE3", "SSSE3", "SSE4_1", "SSE4_2",
+    "AES", "AVX", "AVX2", "AES+AVX", "VAES+AVX",
 ]
 
 ignore = set(["FISTTP",
@@ -42,7 +43,7 @@ imask = {
     'vROUND[PS][SD]': 0x7,
     'vSHUFPD': 0x0f,
     'vSHUFPS': 0xff,
-    'vAESKEYGENASSIST': 0,
+    'vAESKEYGENASSIST': 0xff,
     'VEXTRACT[FI]128': 0x01,
     'VINSERT[FI]128': 0x01,
     'VPBLENDD': 0xff,
@@ -85,7 +86,7 @@ def mem_w(w):
     else:
         raise Exception()
 
-    return t + " PTR 16[rdx]"
+    return t + " PTR 32[rdx]"
 
 class XMMArg():
     isxmm = True