AVX-512: Fix rounding mode value in EVEX prefix with SAE

If SAE is set, VL(vector length) is implied to be 512.
EVEX.L'L (=EVEX.RC) is set to 00b by default.

Signed-off-by: Jin Kyu Song <jin.kyu.song@intel.com>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

diff --git a/assemble.c b/assemble.c
index 6ea8be6..ad34523 100644 (file)
--- a/assemble.c
+++ b/assemble.c
@@ -1167,15 +1167,18 @@ static int64_t calcsize(int32_t segment, int64_t offset, int bits,
                 op_er_sae = (ins->evex_brerop >= 0 ?
                              &ins->oprs[ins->evex_brerop] : NULL);
 
-                if (op_er_sae && (op_er_sae->decoflags & ER)) {
-                    /* set EVEX.RC (rounding control) and b */
-                    ins->evex_p[2] |= (((ins->evex_rm - BRC_RN) << 5) & EVEX_P2LL) |
-                                      EVEX_P2B;
+                if (op_er_sae && (op_er_sae->decoflags & (ER | SAE))) {
+                    /* set EVEX.b */
+                    ins->evex_p[2] |= EVEX_P2B;
+                    if (op_er_sae->decoflags & ER) {
+                        /* set EVEX.RC (rounding control) */
+                        ins->evex_p[2] |= ((ins->evex_rm - BRC_RN) << 5)
+                                          & EVEX_P2RC;
+                    }
                 } else {
                     /* set EVEX.L'L (vector length) */
                     ins->evex_p[2] |= ((ins->vex_wlp << (5 - 2)) & EVEX_P2LL);
-                    if ((op_er_sae && (op_er_sae->decoflags & SAE)) ||
-                        (opy->decoflags & BRDCAST_MASK)) {
+                    if (opy->decoflags & BRDCAST_MASK) {
                         /* set EVEX.b */
                         ins->evex_p[2] |= EVEX_P2B;
                     }

diff --git a/nasm.h b/nasm.h
index e9ef585..50e4b63 100644 (file)
--- a/nasm.h
+++ b/nasm.h
@@ -514,7 +514,8 @@ static inline uint8_t get_cond_opcode(enum ccode c)
 #define EVEX_P2AAA      0x07        /* EVEX P[18:16] : Embedded opmask    */
 #define EVEX_P2VP       0x08        /* EVEX P[19] : High-16 NDS reg       */
 #define EVEX_P2B        0x10        /* EVEX P[20] : Broadcast / RC / SAE  */
-#define EVEX_P2LL       0x60        /* EVEX P[22:21] : Vector length / RC */
+#define EVEX_P2LL       0x60        /* EVEX P[22:21] : Vector length      */
+#define EVEX_P2RC       EVEX_P2LL   /* EVEX P[22:21] : Rounding control   */
 #define EVEX_P2Z        0x80        /* EVEX P[23] : Zeroing/Merging       */
 
 /*