[Qemu-devel] [RFC PATCH v1 16/22] target/i386: reimplement (V)PSUBS(B, W)

[Jan Bobek <jan.bobek@gmail.com>]

Use the gvec infrastructure to achieve the desired functionality.

Signed-off-by: Jan Bobek <jan.bobek@gmail.com>
---
 target/i386/ops_sse.h        |  4 ----
 target/i386/ops_sse_header.h |  2 --
 target/i386/translate.c      | 19 +++++++++++++++++--
 3 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/target/i386/ops_sse.h b/target/i386/ops_sse.h
index 8c9b47fca4..f948adbc68 100644
--- a/target/i386/ops_sse.h
+++ b/target/i386/ops_sse.h
@@ -339,8 +339,6 @@ static inline int satsw(int x)
 
 #define FSUBUB(a, b) satub((a) - (b))
 #define FSUBUW(a, b) satuw((a) - (b))
-#define FSUBSB(a, b) satsb((int8_t)(a) - (int8_t)(b))
-#define FSUBSW(a, b) satsw((int16_t)(a) - (int16_t)(b))
 #define FMINUB(a, b) ((a) < (b)) ? (a) : (b)
 #define FMINSW(a, b) ((int16_t)(a) < (int16_t)(b)) ? (a) : (b)
 #define FMAXUB(a, b) ((a) > (b)) ? (a) : (b)
@@ -360,10 +358,8 @@ static inline int satsw(int x)
 #endif
 
 SSE_HELPER_B(helper_psubusb, FSUBUB)
-SSE_HELPER_B(helper_psubsb, FSUBSB)
 
 SSE_HELPER_W(helper_psubusw, FSUBUW)
-SSE_HELPER_W(helper_psubsw, FSUBSW)
 
 SSE_HELPER_B(helper_pminub, FMINUB)
 SSE_HELPER_B(helper_pmaxub, FMAXUB)
diff --git a/target/i386/ops_sse_header.h b/target/i386/ops_sse_header.h
index 8a31ade70c..20fb8aeccc 100644
--- a/target/i386/ops_sse_header.h
+++ b/target/i386/ops_sse_header.h
@@ -61,10 +61,8 @@ DEF_HELPER_3(glue(pslldq, SUFFIX), void, env, Reg, Reg)
     DEF_HELPER_3(glue(name, SUFFIX), void, env, Reg, Reg)
 
 SSE_HELPER_B(psubusb, FSUBUB)
-SSE_HELPER_B(psubsb, FSUBSB)
 
 SSE_HELPER_W(psubusw, FSUBUW)
-SSE_HELPER_W(psubsw, FSUBSW)
 
 SSE_HELPER_B(pminub, FMINUB)
 SSE_HELPER_B(pmaxub, FMAXUB)
diff --git a/target/i386/translate.c b/target/i386/translate.c
index e2ed8c20b3..894471861d 100644
--- a/target/i386/translate.c
+++ b/target/i386/translate.c
@@ -2838,8 +2838,8 @@ static const SSEFunc_0_epp sse_op_table1[256][4] = {
     [0xe5] = MMX_OP2(pmulhw),
     [0xe6] = { NULL, gen_helper_cvttpd2dq, gen_helper_cvtdq2pd, gen_helper_cvtpd2dq },
     [0xe7] = { SSE_SPECIAL , SSE_SPECIAL },  /* movntq, movntq */
-    [0xe8] = MMX_OP2(psubsb),
-    [0xe9] = MMX_OP2(psubsw),
+    [0xe8] = { SSE_TOMBSTONE, SSE_TOMBSTONE },
+    [0xe9] = { SSE_TOMBSTONE, SSE_TOMBSTONE },
     [0xea] = MMX_OP2(pminsw),
     [0xeb] = { SSE_TOMBSTONE, SSE_TOMBSTONE },
     [0xec] = { SSE_TOMBSTONE, SSE_TOMBSTONE },
@@ -3186,6 +3186,11 @@ static inline void gen_gvec_ld_modrm_3(CPUX86State *env, DisasContext *s,
 #define gen_vpsub_xmm(env, s, modrm, vece) gen_gvec_ld_modrm_vxmm((env), (s), (modrm), (vece), tcg_gen_gvec_sub, 0123)
 #define gen_vpsub_ymm(env, s, modrm, vece) gen_gvec_ld_modrm_vymm((env), (s), (modrm), (vece), tcg_gen_gvec_sub, 0123)
 
+#define gen_psubs_mm(env, s, modrm, vece)   gen_gvec_ld_modrm_mm  ((env), (s), (modrm), (vece), tcg_gen_gvec_sssub, 0112)
+#define gen_psubs_xmm(env, s, modrm, vece)  gen_gvec_ld_modrm_xmm ((env), (s), (modrm), (vece), tcg_gen_gvec_sssub, 0112)
+#define gen_vpsubs_xmm(env, s, modrm, vece) gen_gvec_ld_modrm_vxmm((env), (s), (modrm), (vece), tcg_gen_gvec_sssub, 0123)
+#define gen_vpsubs_ymm(env, s, modrm, vece) gen_gvec_ld_modrm_vymm((env), (s), (modrm), (vece), tcg_gen_gvec_sssub, 0123)
+
 #define gen_pand_mm(env, s, modrm)   gen_gvec_ld_modrm_mm  ((env), (s), (modrm), MO_64, tcg_gen_gvec_and, 0112)
 #define gen_pand_xmm(env, s, modrm)  gen_gvec_ld_modrm_xmm ((env), (s), (modrm), MO_64, tcg_gen_gvec_and, 0112)
 #define gen_vpand_xmm(env, s, modrm) gen_gvec_ld_modrm_vxmm((env), (s), (modrm), MO_64, tcg_gen_gvec_and, 0123)
@@ -3381,6 +3386,16 @@ static void gen_sse(CPUX86State *env, DisasContext *s, int b)
     case 0xfb | M_0F | P_66 | VEX_128: gen_vpsub_xmm(env, s, modrm, MO_64); return;
     case 0xfb | M_0F | P_66 | VEX_256: gen_vpsub_ymm(env, s, modrm, MO_64); return;
 
+    case 0xe8 | M_0F:                  gen_psubs_mm(env, s, modrm, MO_8); return;
+    case 0xe8 | M_0F | P_66:           gen_psubs_xmm(env, s, modrm, MO_8); return;
+    case 0xe8 | M_0F | P_66 | VEX_128: gen_vpsubs_xmm(env, s, modrm, MO_8); return;
+    case 0xe8 | M_0F | P_66 | VEX_256: gen_vpsubs_ymm(env, s, modrm, MO_8); return;
+
+    case 0xe9 | M_0F:                  gen_psubs_mm(env, s, modrm, MO_16); return;
+    case 0xe9 | M_0F | P_66:           gen_psubs_xmm(env, s, modrm, MO_16); return;
+    case 0xe9 | M_0F | P_66 | VEX_128: gen_vpsubs_xmm(env, s, modrm, MO_16); return;
+    case 0xe9 | M_0F | P_66 | VEX_256: gen_vpsubs_ymm(env, s, modrm, MO_16); return;
+
     case 0xdb | M_0F:                  gen_pand_mm(env, s, modrm); return;
     case 0xdb | M_0F | P_66:           gen_pand_xmm(env, s, modrm); return;
     case 0xdb | M_0F | P_66 | VEX_128: gen_vpand_xmm(env, s, modrm); return;
-- 
2.20.1