[PATCH v2 11/46] target/loongarch: Implement xavg/xvagr

[Song Gao]

This patch includes:
- XVAVG.{B/H/W/D/}[U];
- XVAVGR.{B/H/W/D}[U].

Signed-off-by: Song Gao <gaosong@loongson.cn>
---
 target/loongarch/disas.c                     | 17 +++++++++++++++++
 target/loongarch/insn_trans/trans_lasx.c.inc | 17 +++++++++++++++++
 target/loongarch/insns.decode                | 17 +++++++++++++++++
 target/loongarch/vec.h                       |  3 +++
 target/loongarch/vec_helper.c                |  9 ++++-----
 5 files changed, 58 insertions(+), 5 deletions(-)

diff --git a/target/loongarch/disas.c b/target/loongarch/disas.c
index 6972e33833..8296aafa98 100644
--- a/target/loongarch/disas.c
+++ b/target/loongarch/disas.c
@@ -1825,6 +1825,23 @@ INSN_LASX(xvaddwod_w_hu_h,   vvv)
 INSN_LASX(xvaddwod_d_wu_w,   vvv)
 INSN_LASX(xvaddwod_q_du_d,   vvv)
 
+INSN_LASX(xvavg_b,           vvv)
+INSN_LASX(xvavg_h,           vvv)
+INSN_LASX(xvavg_w,           vvv)
+INSN_LASX(xvavg_d,           vvv)
+INSN_LASX(xvavg_bu,          vvv)
+INSN_LASX(xvavg_hu,          vvv)
+INSN_LASX(xvavg_wu,          vvv)
+INSN_LASX(xvavg_du,          vvv)
+INSN_LASX(xvavgr_b,          vvv)
+INSN_LASX(xvavgr_h,          vvv)
+INSN_LASX(xvavgr_w,          vvv)
+INSN_LASX(xvavgr_d,          vvv)
+INSN_LASX(xvavgr_bu,         vvv)
+INSN_LASX(xvavgr_hu,         vvv)
+INSN_LASX(xvavgr_wu,         vvv)
+INSN_LASX(xvavgr_du,         vvv)
+
 INSN_LASX(xvreplgr2vr_b,     vr)
 INSN_LASX(xvreplgr2vr_h,     vr)
 INSN_LASX(xvreplgr2vr_w,     vr)
diff --git a/target/loongarch/insn_trans/trans_lasx.c.inc 
b/target/loongarch/insn_trans/trans_lasx.c.inc
index d8230cba9f..ac4cade845 100644
--- a/target/loongarch/insn_trans/trans_lasx.c.inc
+++ b/target/loongarch/insn_trans/trans_lasx.c.inc
@@ -140,6 +140,23 @@ TRANS(xvaddwod_w_hu_h, gvec_vvv, 32, MO_16, do_vaddwod_u_s)
 TRANS(xvaddwod_d_wu_w, gvec_vvv, 32, MO_32, do_vaddwod_u_s)
 TRANS(xvaddwod_q_du_d, gvec_vvv, 32, MO_64, do_vaddwod_u_s)
 
+TRANS(xvavg_b, gvec_vvv, 32, MO_8, do_vavg_s)
+TRANS(xvavg_h, gvec_vvv, 32, MO_16, do_vavg_s)
+TRANS(xvavg_w, gvec_vvv, 32, MO_32, do_vavg_s)
+TRANS(xvavg_d, gvec_vvv, 32, MO_64, do_vavg_s)
+TRANS(xvavg_bu, gvec_vvv, 32, MO_8, do_vavg_u)
+TRANS(xvavg_hu, gvec_vvv, 32, MO_16, do_vavg_u)
+TRANS(xvavg_wu, gvec_vvv, 32, MO_32, do_vavg_u)
+TRANS(xvavg_du, gvec_vvv, 32, MO_64, do_vavg_u)
+TRANS(xvavgr_b, gvec_vvv, 32, MO_8, do_vavgr_s)
+TRANS(xvavgr_h, gvec_vvv, 32, MO_16, do_vavgr_s)
+TRANS(xvavgr_w, gvec_vvv, 32, MO_32, do_vavgr_s)
+TRANS(xvavgr_d, gvec_vvv, 32, MO_64, do_vavgr_s)
+TRANS(xvavgr_bu, gvec_vvv, 32, MO_8, do_vavgr_u)
+TRANS(xvavgr_hu, gvec_vvv, 32, MO_16, do_vavgr_u)
+TRANS(xvavgr_wu, gvec_vvv, 32, MO_32, do_vavgr_u)
+TRANS(xvavgr_du, gvec_vvv, 32, MO_64, do_vavgr_u)
+
 TRANS(xvreplgr2vr_b, gvec_dup, 32, MO_8)
 TRANS(xvreplgr2vr_h, gvec_dup, 32, MO_16)
 TRANS(xvreplgr2vr_w, gvec_dup, 32, MO_32)
diff --git a/target/loongarch/insns.decode b/target/loongarch/insns.decode
index e1d8b30179..a2cb39750d 100644
--- a/target/loongarch/insns.decode
+++ b/target/loongarch/insns.decode
@@ -1406,6 +1406,23 @@ xvaddwod_w_hu_h  0111 01000100 00001 ..... ..... .....   
 @vvv
 xvaddwod_d_wu_w  0111 01000100 00010 ..... ..... .....    @vvv
 xvaddwod_q_du_d  0111 01000100 00011 ..... ..... .....    @vvv
 
+xvavg_b          0111 01000110 01000 ..... ..... .....    @vvv
+xvavg_h          0111 01000110 01001 ..... ..... .....    @vvv
+xvavg_w          0111 01000110 01010 ..... ..... .....    @vvv
+xvavg_d          0111 01000110 01011 ..... ..... .....    @vvv
+xvavg_bu         0111 01000110 01100 ..... ..... .....    @vvv
+xvavg_hu         0111 01000110 01101 ..... ..... .....    @vvv
+xvavg_wu         0111 01000110 01110 ..... ..... .....    @vvv
+xvavg_du         0111 01000110 01111 ..... ..... .....    @vvv
+xvavgr_b         0111 01000110 10000 ..... ..... .....    @vvv
+xvavgr_h         0111 01000110 10001 ..... ..... .....    @vvv
+xvavgr_w         0111 01000110 10010 ..... ..... .....    @vvv
+xvavgr_d         0111 01000110 10011 ..... ..... .....    @vvv
+xvavgr_bu        0111 01000110 10100 ..... ..... .....    @vvv
+xvavgr_hu        0111 01000110 10101 ..... ..... .....    @vvv
+xvavgr_wu        0111 01000110 10110 ..... ..... .....    @vvv
+xvavgr_du        0111 01000110 10111 ..... ..... .....    @vvv
+
 xvreplgr2vr_b    0111 01101001 11110 00000 ..... .....    @vr
 xvreplgr2vr_h    0111 01101001 11110 00001 ..... .....    @vr
 xvreplgr2vr_w    0111 01101001 11110 00010 ..... .....    @vr
diff --git a/target/loongarch/vec.h b/target/loongarch/vec.h
index 6ff89ebda8..361bf87896 100644
--- a/target/loongarch/vec.h
+++ b/target/loongarch/vec.h
@@ -50,4 +50,7 @@
 #define DO_ADD(a, b)  (a + b)
 #define DO_SUB(a, b)  (a - b)
 
+#define DO_VAVG(a, b)  ((a >> 1) + (b >> 1) + (a & b & 1))
+#define DO_VAVGR(a, b) ((a >> 1) + (b >> 1) + ((a | b) & 1))
+
 #endif /* LOONGARCH_VEC_H */
diff --git a/target/loongarch/vec_helper.c b/target/loongarch/vec_helper.c
index 411d94780d..56997455de 100644
--- a/target/loongarch/vec_helper.c
+++ b/target/loongarch/vec_helper.c
@@ -341,17 +341,16 @@ DO_ODD_U_S(vaddwod_h_bu_b, 16, H, UH, B, UB, DO_ADD)
 DO_ODD_U_S(vaddwod_w_hu_h, 32, W, UW, H, UH, DO_ADD)
 DO_ODD_U_S(vaddwod_d_wu_w, 64, D, UD, W, UW, DO_ADD)
 
-#define DO_VAVG(a, b)  ((a >> 1) + (b >> 1) + (a & b & 1))
-#define DO_VAVGR(a, b) ((a >> 1) + (b >> 1) + ((a | b) & 1))
-
 #define DO_3OP(NAME, BIT, E, DO_OP)                         \
 void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t v) \
 {                                                           \
-    int i;                                                  \
+    int i, len;                                             \
     VReg *Vd = (VReg *)vd;                                  \
     VReg *Vj = (VReg *)vj;                                  \
     VReg *Vk = (VReg *)vk;                                  \
-    for (i = 0; i < LSX_LEN/BIT; i++) {                     \
+                                                            \
+    len = (simd_oprsz(v) == 16) ? LSX_LEN : LASX_LEN;       \
+    for (i = 0; i < len / BIT; i++) {                       \
         Vd->E(i) = DO_OP(Vj->E(i), Vk->E(i));               \
     }                                                       \
 }
-- 
2.39.1