[PATCH qemu v16 05/15] target/riscv: rvv: Add tail agnostic for vector load / store instructions

[~eopxd]

From: eopXD <eop.chen@sifive.com>

Destination register of unit-stride mask load and store instructions are
always written with a tail-agnostic policy.

A vector segment load / store instruction may contain fractional lmul
with nf * lmul > 1. The rest of the elements in the last register should
be treated as tail elements.

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
---
 target/riscv/insn_trans/trans_rvv.c.inc | 11 +++++
 target/riscv/translate.c                |  2 +
 target/riscv/vector_helper.c            | 60 +++++++++++++++++++++++++
 3 files changed, 73 insertions(+)

diff --git a/target/riscv/insn_trans/trans_rvv.c.inc 
b/target/riscv/insn_trans/trans_rvv.c.inc
index efdf5d6d81..1f3eeff9eb 100644
--- a/target/riscv/insn_trans/trans_rvv.c.inc
+++ b/target/riscv/insn_trans/trans_rvv.c.inc
@@ -711,6 +711,7 @@ static bool ld_us_op(DisasContext *s, arg_r2nfvm *a, 
uint8_t eew)
     data = FIELD_DP32(data, VDATA, VM, a->vm);
     data = FIELD_DP32(data, VDATA, LMUL, emul);
     data = FIELD_DP32(data, VDATA, NF, a->nf);
+    data = FIELD_DP32(data, VDATA, VTA, s->vta);
     return ldst_us_trans(a->rd, a->rs1, data, fn, s, false);
 }
 
@@ -748,6 +749,7 @@ static bool st_us_op(DisasContext *s, arg_r2nfvm *a, 
uint8_t eew)
     data = FIELD_DP32(data, VDATA, VM, a->vm);
     data = FIELD_DP32(data, VDATA, LMUL, emul);
     data = FIELD_DP32(data, VDATA, NF, a->nf);
+    data = FIELD_DP32(data, VDATA, VTA, s->vta);
     return ldst_us_trans(a->rd, a->rs1, data, fn, s, true);
 }
 
@@ -774,6 +776,8 @@ static bool ld_us_mask_op(DisasContext *s, arg_vlm_v *a, 
uint8_t eew)
     /* EMUL = 1, NFIELDS = 1 */
     data = FIELD_DP32(data, VDATA, LMUL, 0);
     data = FIELD_DP32(data, VDATA, NF, 1);
+    /* Mask destination register are always tail-agnostic */
+    data = FIELD_DP32(data, VDATA, VTA, s->cfg_vta_all_1s);
     return ldst_us_trans(a->rd, a->rs1, data, fn, s, false);
 }
 
@@ -791,6 +795,8 @@ static bool st_us_mask_op(DisasContext *s, arg_vsm_v *a, 
uint8_t eew)
     /* EMUL = 1, NFIELDS = 1 */
     data = FIELD_DP32(data, VDATA, LMUL, 0);
     data = FIELD_DP32(data, VDATA, NF, 1);
+    /* Mask destination register are always tail-agnostic */
+    data = FIELD_DP32(data, VDATA, VTA, s->cfg_vta_all_1s);
     return ldst_us_trans(a->rd, a->rs1, data, fn, s, true);
 }
 
@@ -862,6 +868,7 @@ static bool ld_stride_op(DisasContext *s, arg_rnfvm *a, 
uint8_t eew)
     data = FIELD_DP32(data, VDATA, VM, a->vm);
     data = FIELD_DP32(data, VDATA, LMUL, emul);
     data = FIELD_DP32(data, VDATA, NF, a->nf);
+    data = FIELD_DP32(data, VDATA, VTA, s->vta);
     return ldst_stride_trans(a->rd, a->rs1, a->rs2, data, fn, s, false);
 }
 
@@ -891,6 +898,7 @@ static bool st_stride_op(DisasContext *s, arg_rnfvm *a, 
uint8_t eew)
     data = FIELD_DP32(data, VDATA, VM, a->vm);
     data = FIELD_DP32(data, VDATA, LMUL, emul);
     data = FIELD_DP32(data, VDATA, NF, a->nf);
+    data = FIELD_DP32(data, VDATA, VTA, s->vta);
     fn = fns[eew];
     if (fn == NULL) {
         return false;
@@ -991,6 +999,7 @@ static bool ld_index_op(DisasContext *s, arg_rnfvm *a, 
uint8_t eew)
     data = FIELD_DP32(data, VDATA, VM, a->vm);
     data = FIELD_DP32(data, VDATA, LMUL, emul);
     data = FIELD_DP32(data, VDATA, NF, a->nf);
+    data = FIELD_DP32(data, VDATA, VTA, s->vta);
     return ldst_index_trans(a->rd, a->rs1, a->rs2, data, fn, s, false);
 }
 
@@ -1043,6 +1052,7 @@ static bool st_index_op(DisasContext *s, arg_rnfvm *a, 
uint8_t eew)
     data = FIELD_DP32(data, VDATA, VM, a->vm);
     data = FIELD_DP32(data, VDATA, LMUL, emul);
     data = FIELD_DP32(data, VDATA, NF, a->nf);
+    data = FIELD_DP32(data, VDATA, VTA, s->vta);
     return ldst_index_trans(a->rd, a->rs1, a->rs2, data, fn, s, true);
 }
 
@@ -1108,6 +1118,7 @@ static bool ldff_op(DisasContext *s, arg_r2nfvm *a, 
uint8_t eew)
     data = FIELD_DP32(data, VDATA, VM, a->vm);
     data = FIELD_DP32(data, VDATA, LMUL, emul);
     data = FIELD_DP32(data, VDATA, NF, a->nf);
+    data = FIELD_DP32(data, VDATA, VTA, s->vta);
     return ldff_trans(a->rd, a->rs1, data, fn, s);
 }
 
diff --git a/target/riscv/translate.c b/target/riscv/translate.c
index 832353be54..384ffcc0fa 100644
--- a/target/riscv/translate.c
+++ b/target/riscv/translate.c
@@ -95,6 +95,7 @@ typedef struct DisasContext {
     int8_t lmul;
     uint8_t sew;
     uint8_t vta;
+    bool cfg_vta_all_1s;
     target_ulong vstart;
     bool vl_eq_vlmax;
     uint8_t ntemp;
@@ -1093,6 +1094,7 @@ static void riscv_tr_init_disas_context(DisasContextBase 
*dcbase, CPUState *cs)
     ctx->sew = FIELD_EX32(tb_flags, TB_FLAGS, SEW);
     ctx->lmul = sextract32(FIELD_EX32(tb_flags, TB_FLAGS, LMUL), 0, 3);
     ctx->vta = FIELD_EX32(tb_flags, TB_FLAGS, VTA) && cpu->cfg.rvv_ta_all_1s;
+    ctx->cfg_vta_all_1s = cpu->cfg.rvv_ta_all_1s;
     ctx->vstart = env->vstart;
     ctx->vl_eq_vlmax = FIELD_EX32(tb_flags, TB_FLAGS, VL_EQ_VLMAX);
     ctx->misa_mxl_max = env->misa_mxl_max;
diff --git a/target/riscv/vector_helper.c b/target/riscv/vector_helper.c
index 61c7074f6a..2a934748b3 100644
--- a/target/riscv/vector_helper.c
+++ b/target/riscv/vector_helper.c
@@ -269,6 +269,9 @@ vext_ldst_stride(void *vd, void *v0, target_ulong base,
     uint32_t i, k;
     uint32_t nf = vext_nf(desc);
     uint32_t max_elems = vext_max_elems(desc, log2_esz);
+    uint32_t esz = 1 << log2_esz;
+    uint32_t total_elems = vext_get_total_elems(env, desc, esz);
+    uint32_t vta = vext_vta(desc);
 
     for (i = env->vstart; i < env->vl; i++, env->vstart++) {
         if (!vm && !vext_elem_mask(v0, i)) {
@@ -283,6 +286,18 @@ vext_ldst_stride(void *vd, void *v0, target_ulong base,
         }
     }
     env->vstart = 0;
+    /* set tail elements to 1s */
+    for (k = 0; k < nf; ++k) {
+        vext_set_elems_1s(vd, vta, (k * max_elems + env->vl) * esz,
+                          (k * max_elems + max_elems) * esz);
+    }
+    if (nf * max_elems % total_elems != 0) {
+        uint32_t vlenb = env_archcpu(env)->cfg.vlen >> 3;
+        uint32_t registers_used =
+            ((nf * max_elems) * esz + (vlenb - 1)) / vlenb;
+        vext_set_elems_1s(vd, vta, (nf * max_elems) * esz,
+                          registers_used * vlenb);
+    }
 }
 
 #define GEN_VEXT_LD_STRIDE(NAME, ETYPE, LOAD_FN)                        \
@@ -328,6 +343,9 @@ vext_ldst_us(void *vd, target_ulong base, CPURISCVState 
*env, uint32_t desc,
     uint32_t i, k;
     uint32_t nf = vext_nf(desc);
     uint32_t max_elems = vext_max_elems(desc, log2_esz);
+    uint32_t esz = 1 << log2_esz;
+    uint32_t total_elems = vext_get_total_elems(env, desc, esz);
+    uint32_t vta = vext_vta(desc);
 
     /* load bytes from guest memory */
     for (i = env->vstart; i < evl; i++, env->vstart++) {
@@ -339,6 +357,18 @@ vext_ldst_us(void *vd, target_ulong base, CPURISCVState 
*env, uint32_t desc,
         }
     }
     env->vstart = 0;
+    /* set tail elements to 1s */
+    for (k = 0; k < nf; ++k) {
+        vext_set_elems_1s(vd, vta, (k * max_elems + evl) * esz,
+                          (k * max_elems + max_elems) * esz);
+    }
+    if (nf * max_elems % total_elems != 0) {
+        uint32_t vlenb = env_archcpu(env)->cfg.vlen >> 3;
+        uint32_t registers_used =
+            ((nf * max_elems) * esz + (vlenb - 1)) / vlenb;
+        vext_set_elems_1s(vd, vta, (nf * max_elems) * esz,
+                          registers_used * vlenb);
+    }
 }
 
 /*
@@ -438,6 +468,9 @@ vext_ldst_index(void *vd, void *v0, target_ulong base,
     uint32_t nf = vext_nf(desc);
     uint32_t vm = vext_vm(desc);
     uint32_t max_elems = vext_max_elems(desc, log2_esz);
+    uint32_t esz = 1 << log2_esz;
+    uint32_t total_elems = vext_get_total_elems(env, desc, esz);
+    uint32_t vta = vext_vta(desc);
 
     /* load bytes from guest memory */
     for (i = env->vstart; i < env->vl; i++, env->vstart++) {
@@ -453,6 +486,18 @@ vext_ldst_index(void *vd, void *v0, target_ulong base,
         }
     }
     env->vstart = 0;
+    /* set tail elements to 1s */
+    for (k = 0; k < nf; ++k) {
+        vext_set_elems_1s(vd, vta, (k * max_elems + env->vl) * esz,
+                          (k * max_elems + max_elems) * esz);
+    }
+    if (nf * max_elems % total_elems != 0) {
+        uint32_t vlenb = env_archcpu(env)->cfg.vlen >> 3;
+        uint32_t registers_used =
+            ((nf * max_elems) * esz + (vlenb - 1)) / vlenb;
+        vext_set_elems_1s(vd, vta, (nf * max_elems) * esz,
+                          registers_used * vlenb);
+    }
 }
 
 #define GEN_VEXT_LD_INDEX(NAME, ETYPE, INDEX_FN, LOAD_FN)                  \
@@ -520,6 +565,9 @@ vext_ldff(void *vd, void *v0, target_ulong base,
     uint32_t nf = vext_nf(desc);
     uint32_t vm = vext_vm(desc);
     uint32_t max_elems = vext_max_elems(desc, log2_esz);
+    uint32_t esz = 1 << log2_esz;
+    uint32_t total_elems = vext_get_total_elems(env, desc, esz);
+    uint32_t vta = vext_vta(desc);
     target_ulong addr, offset, remain;
 
     /* probe every access*/
@@ -575,6 +623,18 @@ ProbeSuccess:
         }
     }
     env->vstart = 0;
+    /* set tail elements to 1s */
+    for (k = 0; k < nf; ++k) {
+        vext_set_elems_1s(vd, vta, (k * max_elems + env->vl) * esz,
+                          (k * max_elems + max_elems) * esz);
+    }
+    if (nf * max_elems % total_elems != 0) {
+        uint32_t vlenb = env_archcpu(env)->cfg.vlen >> 3;
+        uint32_t registers_used =
+            ((nf * max_elems) * esz + (vlenb - 1)) / vlenb;
+        vext_set_elems_1s(vd, vta, (nf * max_elems) * esz,
+                          registers_used * vlenb);
+    }
 }
 
 #define GEN_VEXT_LDFF(NAME, ETYPE, LOAD_FN)               \
-- 
2.34.2