[PATCH v5 08/14] RISC-V: Adding T-Head MemPair extension

[Christoph Muellner]

From: Christoph Müllner <christoph.muellner@vrull.eu>

This patch adds support for the T-Head MemPair instructions.
The patch uses the T-Head specific decoder and translation.

Co-developed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Christoph Müllner <christoph.muellner@vrull.eu>
---
Changes in v2:
- Add ISA_EXT_DATA_ENTRY()
- Use single decoder for XThead extensions
- Use get_address() to calculate addresses

Changes in v4:
- Don't translate in case of overlapping registers
- Use temporaries to avoid unexpected behaviour in case of exceptions
- There have also been clarifications in the specification

Changes in v5:
- Use memop_size() instead of hard coded constants

 target/riscv/cpu.c                         |  2 +
 target/riscv/cpu.h                         |  1 +
 target/riscv/insn_trans/trans_xthead.c.inc | 92 ++++++++++++++++++++++
 target/riscv/translate.c                   |  2 +-
 target/riscv/xthead.decode                 | 13 +++
 5 files changed, 109 insertions(+), 1 deletion(-)

diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 88da4de14d..b7047d139d 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -115,6 +115,7 @@ static const struct isa_ext_data isa_edata_arr[] = {
     ISA_EXT_DATA_ENTRY(xtheadcmo, true, PRIV_VERSION_1_11_0, ext_xtheadcmo),
     ISA_EXT_DATA_ENTRY(xtheadcondmov, true, PRIV_VERSION_1_11_0, 
ext_xtheadcondmov),
     ISA_EXT_DATA_ENTRY(xtheadmac, true, PRIV_VERSION_1_11_0, ext_xtheadmac),
+    ISA_EXT_DATA_ENTRY(xtheadmempair, true, PRIV_VERSION_1_11_0, 
ext_xtheadmempair),
     ISA_EXT_DATA_ENTRY(xtheadsync, true, PRIV_VERSION_1_11_0, ext_xtheadsync),
     ISA_EXT_DATA_ENTRY(xventanacondops, true, PRIV_VERSION_1_12_0, 
ext_XVentanaCondOps),
 };
@@ -1101,6 +1102,7 @@ static Property riscv_cpu_extensions[] = {
     DEFINE_PROP_BOOL("xtheadcmo", RISCVCPU, cfg.ext_xtheadcmo, false),
     DEFINE_PROP_BOOL("xtheadcondmov", RISCVCPU, cfg.ext_xtheadcondmov, false),
     DEFINE_PROP_BOOL("xtheadmac", RISCVCPU, cfg.ext_xtheadmac, false),
+    DEFINE_PROP_BOOL("xtheadmempair", RISCVCPU, cfg.ext_xtheadmempair, false),
     DEFINE_PROP_BOOL("xtheadsync", RISCVCPU, cfg.ext_xtheadsync, false),
     DEFINE_PROP_BOOL("xventanacondops", RISCVCPU, cfg.ext_XVentanaCondOps, 
false),
 
diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
index 830b20558c..38e80d44d5 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -479,6 +479,7 @@ struct RISCVCPUConfig {
     bool ext_xtheadcmo;
     bool ext_xtheadcondmov;
     bool ext_xtheadmac;
+    bool ext_xtheadmempair;
     bool ext_xtheadsync;
     bool ext_XVentanaCondOps;
 
diff --git a/target/riscv/insn_trans/trans_xthead.c.inc 
b/target/riscv/insn_trans/trans_xthead.c.inc
index 31a4034927..f1bd0dbad5 100644
--- a/target/riscv/insn_trans/trans_xthead.c.inc
+++ b/target/riscv/insn_trans/trans_xthead.c.inc
@@ -52,6 +52,12 @@
     }                                            \
 } while (0)
 
+#define REQUIRE_XTHEADMEMPAIR(ctx) do {          \
+    if (!ctx->cfg_ptr->ext_xtheadmempair) {      \
+        return false;                            \
+    }                                            \
+} while (0)
+
 #define REQUIRE_XTHEADSYNC(ctx) do {             \
     if (!ctx->cfg_ptr->ext_xtheadsync) {         \
         return false;                            \
@@ -382,6 +388,92 @@ static bool trans_th_mulsw(DisasContext *ctx, arg_th_mulsw 
*a)
     return gen_th_mac(ctx, a, tcg_gen_sub_tl, NULL);
 }
 
+/* XTheadMemPair */
+
+static bool gen_loadpair_tl(DisasContext *ctx, arg_th_pair *a, MemOp memop,
+                            int shamt)
+{
+    if (a->rs == a->rd1 || a->rs == a->rd2 || a->rd1 == a->rd2) {
+        return false;
+    }
+
+    TCGv t1 = tcg_temp_new();
+    TCGv t2 = tcg_temp_new();
+    TCGv addr1 = tcg_temp_new();
+    TCGv addr2 = tcg_temp_new();
+    int imm = a->sh2 << shamt;
+
+    addr1 = get_address(ctx, a->rs, imm);
+    addr2 = get_address(ctx, a->rs, memop_size(memop) + imm);
+
+    tcg_gen_qemu_ld_tl(t1, addr1, ctx->mem_idx, memop);
+    tcg_gen_qemu_ld_tl(t2, addr2, ctx->mem_idx, memop);
+    gen_set_gpr(ctx, a->rd1, t1);
+    gen_set_gpr(ctx, a->rd2, t2);
+
+    tcg_temp_free(t1);
+    tcg_temp_free(t2);
+    tcg_temp_free(addr1);
+    tcg_temp_free(addr2);
+    return true;
+}
+
+static bool trans_th_ldd(DisasContext *ctx, arg_th_pair *a)
+{
+    REQUIRE_XTHEADMEMPAIR(ctx);
+    REQUIRE_64BIT(ctx);
+    return gen_loadpair_tl(ctx, a, MO_TESQ, 4);
+}
+
+static bool trans_th_lwd(DisasContext *ctx, arg_th_pair *a)
+{
+    REQUIRE_XTHEADMEMPAIR(ctx);
+    return gen_loadpair_tl(ctx, a, MO_TESL, 3);
+}
+
+static bool trans_th_lwud(DisasContext *ctx, arg_th_pair *a)
+{
+    REQUIRE_XTHEADMEMPAIR(ctx);
+    return gen_loadpair_tl(ctx, a, MO_TEUL, 3);
+}
+
+static bool gen_storepair_tl(DisasContext *ctx, arg_th_pair *a, MemOp memop,
+                             int shamt)
+{
+    if (a->rs == a->rd1 || a->rs == a->rd2 || a->rd1 == a->rd2) {
+        return false;
+    }
+
+    TCGv data1 = get_gpr(ctx, a->rd1, EXT_NONE);
+    TCGv data2 = get_gpr(ctx, a->rd2, EXT_NONE);
+    TCGv addr1 = tcg_temp_new();
+    TCGv addr2 = tcg_temp_new();
+    int imm = a->sh2 << shamt;
+
+    addr1 = get_address(ctx, a->rs, imm);
+    addr2 = get_address(ctx, a->rs, memop_size(memop) + imm);
+
+    tcg_gen_qemu_st_tl(data1, addr1, ctx->mem_idx, memop);
+    tcg_gen_qemu_st_tl(data2, addr2, ctx->mem_idx, memop);
+
+    tcg_temp_free(addr1);
+    tcg_temp_free(addr2);
+    return true;
+}
+
+static bool trans_th_sdd(DisasContext *ctx, arg_th_pair *a)
+{
+    REQUIRE_XTHEADMEMPAIR(ctx);
+    REQUIRE_64BIT(ctx);
+    return gen_storepair_tl(ctx, a, MO_TESQ, 4);
+}
+
+static bool trans_th_swd(DisasContext *ctx, arg_th_pair *a)
+{
+    REQUIRE_XTHEADMEMPAIR(ctx);
+    return gen_storepair_tl(ctx, a, MO_TESL, 3);
+}
+
 /* XTheadSync */
 
 static bool trans_th_sfence_vmas(DisasContext *ctx, arg_th_sfence_vmas *a)
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index e5a57a8516..f383e69db3 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -135,7 +135,7 @@ static bool has_xthead_p(DisasContext *ctx  
__attribute__((__unused__)))
     return ctx->cfg_ptr->ext_xtheadba || ctx->cfg_ptr->ext_xtheadbb ||
            ctx->cfg_ptr->ext_xtheadbs || ctx->cfg_ptr->ext_xtheadcmo ||
            ctx->cfg_ptr->ext_xtheadcondmov || ctx->cfg_ptr->ext_xtheadmac ||
-           ctx->cfg_ptr->ext_xtheadsync;
+           ctx->cfg_ptr->ext_xtheadmempair || ctx->cfg_ptr->ext_xtheadsync;
 }
 
 #define MATERIALISE_EXT_PREDICATE(ext)  \
diff --git a/target/riscv/xthead.decode b/target/riscv/xthead.decode
index 696de6cecf..ff2a83b56d 100644
--- a/target/riscv/xthead.decode
+++ b/target/riscv/xthead.decode
@@ -11,16 +11,21 @@
 
 # Fields:
 %rd        7:5
+%rd1       7:5
+%rs        15:5
 %rs1       15:5
+%rd2       20:5
 %rs2       20:5
 %sh5       20:5
 %sh6       20:6
+%sh2       25:2
 
 # Argument sets
 &r         rd rs1 rs2                               !extern
 &r2        rd rs1                                   !extern
 &shift     shamt rs1 rd                             !extern
 &th_bfext  msb lsb rs1 rd
+&th_pair   rd1 rs rd2 sh2
 
 # Formats
 @sfence_vm  ....... ..... .....   ... ..... ....... %rs1
@@ -30,6 +35,7 @@
 @th_bfext   msb:6  lsb:6  .....  ... ..... .......  &th_bfext %rs1 %rd
 @sh5        ....... ..... .....  ... ..... .......  &shift  shamt=%sh5      
%rs1 %rd
 @sh6        ...... ...... .....  ... ..... .......  &shift shamt=%sh6 %rs1 %rd
+@th_pair    ..... .. ..... ..... ... ..... .......  &th_pair %rd1 %rs %rd2 %sh2
 
 # XTheadBa
 # Instead of defining a new encoding, we simply use the decoder to
@@ -96,6 +102,13 @@ th_muls          00100 01 ..... ..... 001 ..... 0001011 @r
 th_mulsh         00101 01 ..... ..... 001 ..... 0001011 @r
 th_mulsw         00100 11 ..... ..... 001 ..... 0001011 @r
 
+# XTheadMemPair
+th_ldd           11111 .. ..... ..... 100 ..... 0001011 @th_pair
+th_lwd           11100 .. ..... ..... 100 ..... 0001011 @th_pair
+th_lwud          11110 .. ..... ..... 100 ..... 0001011 @th_pair
+th_sdd           11111 .. ..... ..... 101 ..... 0001011 @th_pair
+th_swd           11100 .. ..... ..... 101 ..... 0001011 @th_pair
+
 # XTheadSync
 th_sfence_vmas   0000010 ..... ..... 000 00000 0001011 @rs2_s
 th_sync          0000000 11000 00000 000 00000 0001011
-- 
2.39.1