@@ -2152,3 +2152,40 @@ DEF_HELPER_6(th_vfsgnjn_vf_d, void, ptr, ptr, i64, ptr, env, i32)
DEF_HELPER_6(th_vfsgnjx_vf_h, void, ptr, ptr, i64, ptr, env, i32)
DEF_HELPER_6(th_vfsgnjx_vf_w, void, ptr, ptr, i64, ptr, env, i32)
DEF_HELPER_6(th_vfsgnjx_vf_d, void, ptr, ptr, i64, ptr, env, i32)
+
+DEF_HELPER_6(th_vmfeq_vv_h, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfeq_vv_w, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfeq_vv_d, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfne_vv_h, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfne_vv_w, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfne_vv_d, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmflt_vv_h, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmflt_vv_w, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmflt_vv_d, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfle_vv_h, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfle_vv_w, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfle_vv_d, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmfeq_vf_h, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfeq_vf_w, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfeq_vf_d, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfne_vf_h, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfne_vf_w, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfne_vf_d, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmflt_vf_h, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmflt_vf_w, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmflt_vf_d, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfle_vf_h, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfle_vf_w, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfle_vf_d, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfgt_vf_h, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfgt_vf_w, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfgt_vf_d, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfge_vf_h, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfge_vf_w, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmfge_vf_d, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmford_vv_h, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmford_vv_w, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmford_vv_d, void, ptr, ptr, ptr, ptr, env, i32)
+DEF_HELPER_6(th_vmford_vf_h, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmford_vf_w, void, ptr, ptr, i64, ptr, env, i32)
+DEF_HELPER_6(th_vmford_vf_d, void, ptr, ptr, i64, ptr, env, i32)
@@ -2106,24 +2106,49 @@ GEN_OPFVF_TRANS_TH(th_vfsgnj_vf, opfvf_check_th)
GEN_OPFVF_TRANS_TH(th_vfsgnjn_vf, opfvf_check_th)
GEN_OPFVF_TRANS_TH(th_vfsgnjx_vf, opfvf_check_th)
+/* Vector Floating-Point Compare Instructions */
+static bool opfvv_cmp_check_th(DisasContext *s, arg_rmrr *a)
+{
+ return (require_xtheadvector(s) &&
+ vext_check_isa_ill(s) &&
+ th_check_reg(s, a->rs2, false) &&
+ th_check_reg(s, a->rs1, false) &&
+ (s->sew != 0) &&
+ ((th_check_overlap_group(a->rd, 1, a->rs1, 1 << s->lmul) &&
+ th_check_overlap_group(a->rd, 1, a->rs2, 1 << s->lmul)) ||
+ (s->lmul == 0)));
+}
+
+GEN_OPFVV_TRANS_TH(th_vmfeq_vv, opfvv_cmp_check_th)
+GEN_OPFVV_TRANS_TH(th_vmfne_vv, opfvv_cmp_check_th)
+GEN_OPFVV_TRANS_TH(th_vmflt_vv, opfvv_cmp_check_th)
+GEN_OPFVV_TRANS_TH(th_vmfle_vv, opfvv_cmp_check_th)
+GEN_OPFVV_TRANS_TH(th_vmford_vv, opfvv_cmp_check_th)
+
+static bool opfvf_cmp_check_th(DisasContext *s, arg_rmrr *a)
+{
+ return (require_xtheadvector(s) &&
+ vext_check_isa_ill(s) &&
+ th_check_reg(s, a->rs2, false) &&
+ (s->sew != 0) &&
+ (th_check_overlap_group(a->rd, 1, a->rs2, 1 << s->lmul) ||
+ (s->lmul == 0)));
+}
+
+GEN_OPFVF_TRANS_TH(th_vmfeq_vf, opfvf_cmp_check_th)
+GEN_OPFVF_TRANS_TH(th_vmfne_vf, opfvf_cmp_check_th)
+GEN_OPFVF_TRANS_TH(th_vmflt_vf, opfvf_cmp_check_th)
+GEN_OPFVF_TRANS_TH(th_vmfle_vf, opfvf_cmp_check_th)
+GEN_OPFVF_TRANS_TH(th_vmfgt_vf, opfvf_cmp_check_th)
+GEN_OPFVF_TRANS_TH(th_vmfge_vf, opfvf_cmp_check_th)
+GEN_OPFVF_TRANS_TH(th_vmford_vf, opfvf_cmp_check_th)
+
#define TH_TRANS_STUB(NAME) \
static bool trans_##NAME(DisasContext *s, arg_##NAME *a) \
{ \
return require_xtheadvector(s); \
}
-TH_TRANS_STUB(th_vmfeq_vv)
-TH_TRANS_STUB(th_vmfeq_vf)
-TH_TRANS_STUB(th_vmfne_vv)
-TH_TRANS_STUB(th_vmfne_vf)
-TH_TRANS_STUB(th_vmflt_vv)
-TH_TRANS_STUB(th_vmflt_vf)
-TH_TRANS_STUB(th_vmfle_vv)
-TH_TRANS_STUB(th_vmfle_vf)
-TH_TRANS_STUB(th_vmfgt_vf)
-TH_TRANS_STUB(th_vmfge_vf)
-TH_TRANS_STUB(th_vmford_vv)
-TH_TRANS_STUB(th_vmford_vf)
TH_TRANS_STUB(th_vfclass_v)
TH_TRANS_STUB(th_vfmerge_vfm)
TH_TRANS_STUB(th_vfmv_v_f)
@@ -4050,19 +4050,19 @@ GEN_VEXT_CMP_VF(vmfeq_vf_h, uint16_t, H2, float16_eq_quiet)
GEN_VEXT_CMP_VF(vmfeq_vf_w, uint32_t, H4, float32_eq_quiet)
GEN_VEXT_CMP_VF(vmfeq_vf_d, uint64_t, H8, float64_eq_quiet)
-static bool vmfne16(uint16_t a, uint16_t b, float_status *s)
+bool vmfne16(uint16_t a, uint16_t b, float_status *s)
{
FloatRelation compare = float16_compare_quiet(a, b, s);
return compare != float_relation_equal;
}
-static bool vmfne32(uint32_t a, uint32_t b, float_status *s)
+bool vmfne32(uint32_t a, uint32_t b, float_status *s)
{
FloatRelation compare = float32_compare_quiet(a, b, s);
return compare != float_relation_equal;
}
-static bool vmfne64(uint64_t a, uint64_t b, float_status *s)
+bool vmfne64(uint64_t a, uint64_t b, float_status *s)
{
FloatRelation compare = float64_compare_quiet(a, b, s);
return compare != float_relation_equal;
@@ -4089,19 +4089,19 @@ GEN_VEXT_CMP_VF(vmfle_vf_h, uint16_t, H2, float16_le)
GEN_VEXT_CMP_VF(vmfle_vf_w, uint32_t, H4, float32_le)
GEN_VEXT_CMP_VF(vmfle_vf_d, uint64_t, H8, float64_le)
-static bool vmfgt16(uint16_t a, uint16_t b, float_status *s)
+bool vmfgt16(uint16_t a, uint16_t b, float_status *s)
{
FloatRelation compare = float16_compare(a, b, s);
return compare == float_relation_greater;
}
-static bool vmfgt32(uint32_t a, uint32_t b, float_status *s)
+bool vmfgt32(uint32_t a, uint32_t b, float_status *s)
{
FloatRelation compare = float32_compare(a, b, s);
return compare == float_relation_greater;
}
-static bool vmfgt64(uint64_t a, uint64_t b, float_status *s)
+bool vmfgt64(uint64_t a, uint64_t b, float_status *s)
{
FloatRelation compare = float64_compare(a, b, s);
return compare == float_relation_greater;
@@ -4111,21 +4111,21 @@ GEN_VEXT_CMP_VF(vmfgt_vf_h, uint16_t, H2, vmfgt16)
GEN_VEXT_CMP_VF(vmfgt_vf_w, uint32_t, H4, vmfgt32)
GEN_VEXT_CMP_VF(vmfgt_vf_d, uint64_t, H8, vmfgt64)
-static bool vmfge16(uint16_t a, uint16_t b, float_status *s)
+bool vmfge16(uint16_t a, uint16_t b, float_status *s)
{
FloatRelation compare = float16_compare(a, b, s);
return compare == float_relation_greater ||
compare == float_relation_equal;
}
-static bool vmfge32(uint32_t a, uint32_t b, float_status *s)
+bool vmfge32(uint32_t a, uint32_t b, float_status *s)
{
FloatRelation compare = float32_compare(a, b, s);
return compare == float_relation_greater ||
compare == float_relation_equal;
}
-static bool vmfge64(uint64_t a, uint64_t b, float_status *s)
+bool vmfge64(uint64_t a, uint64_t b, float_status *s)
{
FloatRelation compare = float64_compare(a, b, s);
return compare == float_relation_greater ||
@@ -404,4 +404,14 @@ uint16_t fsgnjx16(uint16_t a, uint16_t b, float_status *s);
uint32_t fsgnjx32(uint32_t a, uint32_t b, float_status *s);
uint64_t fsgnjx64(uint64_t a, uint64_t b, float_status *s);
+bool vmfne16(uint16_t a, uint16_t b, float_status *s);
+bool vmfne32(uint32_t a, uint32_t b, float_status *s);
+bool vmfne64(uint64_t a, uint64_t b, float_status *s);
+bool vmfgt16(uint16_t a, uint16_t b, float_status *s);
+bool vmfgt32(uint32_t a, uint32_t b, float_status *s);
+bool vmfgt64(uint64_t a, uint64_t b, float_status *s);
+bool vmfge16(uint16_t a, uint16_t b, float_status *s);
+bool vmfge32(uint32_t a, uint32_t b, float_status *s);
+bool vmfge64(uint64_t a, uint64_t b, float_status *s);
+
#endif /* TARGET_RISCV_VECTOR_INTERNALS_H */
@@ -3051,3 +3051,99 @@ THCALL(TH_OPFVF2, th_vfsgnjx_vf_d, OP_UUU_D, H8, H8, fsgnjx64)
GEN_TH_VF(th_vfsgnjx_vf_h, 2, 2, clearh_th)
GEN_TH_VF(th_vfsgnjx_vf_w, 4, 4, clearl_th)
GEN_TH_VF(th_vfsgnjx_vf_d, 8, 8, clearq_th)
+
+/* Vector Floating-Point Compare Instructions */
+#define GEN_TH_CMP_VV_ENV(NAME, ETYPE, H, DO_OP) \
+void HELPER(NAME)(void *vd, void *v0, void *vs1, void *vs2, \
+ CPURISCVState *env, uint32_t desc) \
+{ \
+ uint32_t mlen = th_mlen(desc); \
+ uint32_t vm = th_vm(desc); \
+ uint32_t vl = env->vl; \
+ uint32_t vlmax = th_maxsz(desc) / sizeof(ETYPE); \
+ uint32_t i; \
+ \
+ VSTART_CHECK_EARLY_EXIT(env); \
+ for (i = env->vstart; i < vl; i++) { \
+ ETYPE s1 = *((ETYPE *)vs1 + H(i)); \
+ ETYPE s2 = *((ETYPE *)vs2 + H(i)); \
+ if (!vm && !th_elem_mask(v0, mlen, i)) { \
+ continue; \
+ } \
+ th_set_elem_mask(vd, mlen, i, \
+ DO_OP(s2, s1, &env->fp_status)); \
+ } \
+ env->vstart = 0; \
+ for (; i < vlmax; i++) { \
+ th_set_elem_mask(vd, mlen, i, 0); \
+ } \
+}
+
+GEN_TH_CMP_VV_ENV(th_vmfeq_vv_h, uint16_t, H2, float16_eq_quiet)
+GEN_TH_CMP_VV_ENV(th_vmfeq_vv_w, uint32_t, H4, float32_eq_quiet)
+GEN_TH_CMP_VV_ENV(th_vmfeq_vv_d, uint64_t, H8, float64_eq_quiet)
+
+#define GEN_TH_CMP_VF(NAME, ETYPE, H, DO_OP) \
+void HELPER(NAME)(void *vd, void *v0, uint64_t s1, void *vs2, \
+ CPURISCVState *env, uint32_t desc) \
+{ \
+ uint32_t mlen = th_mlen(desc); \
+ uint32_t vm = th_vm(desc); \
+ uint32_t vl = env->vl; \
+ uint32_t vlmax = th_maxsz(desc) / sizeof(ETYPE); \
+ uint32_t i; \
+ \
+ VSTART_CHECK_EARLY_EXIT(env); \
+ for (i = env->vstart; i < vl; i++) { \
+ ETYPE s2 = *((ETYPE *)vs2 + H(i)); \
+ if (!vm && !th_elem_mask(v0, mlen, i)) { \
+ continue; \
+ } \
+ th_set_elem_mask(vd, mlen, i, \
+ DO_OP(s2, (ETYPE)s1, &env->fp_status)); \
+ } \
+ env->vstart = 0; \
+ for (; i < vlmax; i++) { \
+ th_set_elem_mask(vd, mlen, i, 0); \
+ } \
+}
+
+GEN_TH_CMP_VF(th_vmfeq_vf_h, uint16_t, H2, float16_eq_quiet)
+GEN_TH_CMP_VF(th_vmfeq_vf_w, uint32_t, H4, float32_eq_quiet)
+GEN_TH_CMP_VF(th_vmfeq_vf_d, uint64_t, H8, float64_eq_quiet)
+
+GEN_TH_CMP_VV_ENV(th_vmfne_vv_h, uint16_t, H2, vmfne16)
+GEN_TH_CMP_VV_ENV(th_vmfne_vv_w, uint32_t, H4, vmfne32)
+GEN_TH_CMP_VV_ENV(th_vmfne_vv_d, uint64_t, H8, vmfne64)
+GEN_TH_CMP_VF(th_vmfne_vf_h, uint16_t, H2, vmfne16)
+GEN_TH_CMP_VF(th_vmfne_vf_w, uint32_t, H4, vmfne32)
+GEN_TH_CMP_VF(th_vmfne_vf_d, uint64_t, H8, vmfne64)
+
+GEN_TH_CMP_VV_ENV(th_vmflt_vv_h, uint16_t, H2, float16_lt)
+GEN_TH_CMP_VV_ENV(th_vmflt_vv_w, uint32_t, H4, float32_lt)
+GEN_TH_CMP_VV_ENV(th_vmflt_vv_d, uint64_t, H8, float64_lt)
+GEN_TH_CMP_VF(th_vmflt_vf_h, uint16_t, H2, float16_lt)
+GEN_TH_CMP_VF(th_vmflt_vf_w, uint32_t, H4, float32_lt)
+GEN_TH_CMP_VF(th_vmflt_vf_d, uint64_t, H8, float64_lt)
+
+GEN_TH_CMP_VV_ENV(th_vmfle_vv_h, uint16_t, H2, float16_le)
+GEN_TH_CMP_VV_ENV(th_vmfle_vv_w, uint32_t, H4, float32_le)
+GEN_TH_CMP_VV_ENV(th_vmfle_vv_d, uint64_t, H8, float64_le)
+GEN_TH_CMP_VF(th_vmfle_vf_h, uint16_t, H2, float16_le)
+GEN_TH_CMP_VF(th_vmfle_vf_w, uint32_t, H4, float32_le)
+GEN_TH_CMP_VF(th_vmfle_vf_d, uint64_t, H8, float64_le)
+
+GEN_TH_CMP_VF(th_vmfgt_vf_h, uint16_t, H2, vmfgt16)
+GEN_TH_CMP_VF(th_vmfgt_vf_w, uint32_t, H4, vmfgt32)
+GEN_TH_CMP_VF(th_vmfgt_vf_d, uint64_t, H8, vmfgt64)
+
+GEN_TH_CMP_VF(th_vmfge_vf_h, uint16_t, H2, vmfge16)
+GEN_TH_CMP_VF(th_vmfge_vf_w, uint32_t, H4, vmfge32)
+GEN_TH_CMP_VF(th_vmfge_vf_d, uint64_t, H8, vmfge64)
+
+GEN_TH_CMP_VV_ENV(th_vmford_vv_h, uint16_t, H2, !float16_unordered_quiet)
+GEN_TH_CMP_VV_ENV(th_vmford_vv_w, uint32_t, H4, !float32_unordered_quiet)
+GEN_TH_CMP_VV_ENV(th_vmford_vv_d, uint64_t, H8, !float64_unordered_quiet)
+GEN_TH_CMP_VF(th_vmford_vf_h, uint16_t, H2, !float16_unordered_quiet)
+GEN_TH_CMP_VF(th_vmford_vf_w, uint32_t, H4, !float32_unordered_quiet)
+GEN_TH_CMP_VF(th_vmford_vf_d, uint64_t, H8, !float64_unordered_quiet)
There is no similar instruction in RVV1.0 as th.vmford in XTheadVector. Signed-off-by: Huang Tao <eric.huang@linux.alibaba.com> --- target/riscv/helper.h | 37 +++++++ .../riscv/insn_trans/trans_xtheadvector.c.inc | 49 +++++++--- target/riscv/vector_helper.c | 18 ++-- target/riscv/vector_internals.h | 10 ++ target/riscv/xtheadvector_helper.c | 96 +++++++++++++++++++ 5 files changed, 189 insertions(+), 21 deletions(-)