target/loongarch: Implement LSX fpu arith instructions
This patch includes:
- VF{ADD/SUB/MUL/DIV}.{S/D};
- VF{MADD/MSUB/NMADD/NMSUB}.{S/D};
- VF{MAX/MIN}.{S/D};
- VF{MAXA/MINA}.{S/D};
- VFLOGB.{S/D};
- VFCLASS.{S/D};
- VF{SQRT/RECIP/RSQRT}.{S/D}.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230504122810.4094787-34-gaosong@loongson.cn>
diff --git a/target/loongarch/lsx_helper.c b/target/loongarch/lsx_helper.c
index 66e9d40..d954f9e 100644
--- a/target/loongarch/lsx_helper.c
+++ b/target/loongarch/lsx_helper.c
@@ -9,6 +9,8 @@
#include "cpu.h"
#include "exec/exec-all.h"
#include "exec/helper-proto.h"
+#include "fpu/softfloat.h"
+#include "internals.h"
#define DO_ADD(a, b) (a + b)
#define DO_SUB(a, b) (a - b)
@@ -2060,3 +2062,187 @@
VFRSTPI(vfrstpi_b, 8, B)
VFRSTPI(vfrstpi_h, 16, H)
+
+static void vec_update_fcsr0_mask(CPULoongArchState *env,
+ uintptr_t pc, int mask)
+{
+ int flags = get_float_exception_flags(&env->fp_status);
+
+ set_float_exception_flags(0, &env->fp_status);
+
+ flags &= ~mask;
+
+ if (flags) {
+ flags = ieee_ex_to_loongarch(flags);
+ UPDATE_FP_CAUSE(env->fcsr0, flags);
+ }
+
+ if (GET_FP_ENABLES(env->fcsr0) & flags) {
+ do_raise_exception(env, EXCCODE_FPE, pc);
+ } else {
+ UPDATE_FP_FLAGS(env->fcsr0, flags);
+ }
+}
+
+static void vec_update_fcsr0(CPULoongArchState *env, uintptr_t pc)
+{
+ vec_update_fcsr0_mask(env, pc, 0);
+}
+
+static inline void vec_clear_cause(CPULoongArchState *env)
+{
+ SET_FP_CAUSE(env->fcsr0, 0);
+}
+
+#define DO_3OP_F(NAME, BIT, E, FN) \
+void HELPER(NAME)(CPULoongArchState *env, \
+ uint32_t vd, uint32_t vj, uint32_t vk) \
+{ \
+ int i; \
+ VReg *Vd = &(env->fpr[vd].vreg); \
+ VReg *Vj = &(env->fpr[vj].vreg); \
+ VReg *Vk = &(env->fpr[vk].vreg); \
+ \
+ vec_clear_cause(env); \
+ for (i = 0; i < LSX_LEN/BIT; i++) { \
+ Vd->E(i) = FN(Vj->E(i), Vk->E(i), &env->fp_status); \
+ vec_update_fcsr0(env, GETPC()); \
+ } \
+}
+
+DO_3OP_F(vfadd_s, 32, UW, float32_add)
+DO_3OP_F(vfadd_d, 64, UD, float64_add)
+DO_3OP_F(vfsub_s, 32, UW, float32_sub)
+DO_3OP_F(vfsub_d, 64, UD, float64_sub)
+DO_3OP_F(vfmul_s, 32, UW, float32_mul)
+DO_3OP_F(vfmul_d, 64, UD, float64_mul)
+DO_3OP_F(vfdiv_s, 32, UW, float32_div)
+DO_3OP_F(vfdiv_d, 64, UD, float64_div)
+DO_3OP_F(vfmax_s, 32, UW, float32_maxnum)
+DO_3OP_F(vfmax_d, 64, UD, float64_maxnum)
+DO_3OP_F(vfmin_s, 32, UW, float32_minnum)
+DO_3OP_F(vfmin_d, 64, UD, float64_minnum)
+DO_3OP_F(vfmaxa_s, 32, UW, float32_maxnummag)
+DO_3OP_F(vfmaxa_d, 64, UD, float64_maxnummag)
+DO_3OP_F(vfmina_s, 32, UW, float32_minnummag)
+DO_3OP_F(vfmina_d, 64, UD, float64_minnummag)
+
+#define DO_4OP_F(NAME, BIT, E, FN, flags) \
+void HELPER(NAME)(CPULoongArchState *env, \
+ uint32_t vd, uint32_t vj, uint32_t vk, uint32_t va) \
+{ \
+ int i; \
+ VReg *Vd = &(env->fpr[vd].vreg); \
+ VReg *Vj = &(env->fpr[vj].vreg); \
+ VReg *Vk = &(env->fpr[vk].vreg); \
+ VReg *Va = &(env->fpr[va].vreg); \
+ \
+ vec_clear_cause(env); \
+ for (i = 0; i < LSX_LEN/BIT; i++) { \
+ Vd->E(i) = FN(Vj->E(i), Vk->E(i), Va->E(i), flags, &env->fp_status); \
+ vec_update_fcsr0(env, GETPC()); \
+ } \
+}
+
+DO_4OP_F(vfmadd_s, 32, UW, float32_muladd, 0)
+DO_4OP_F(vfmadd_d, 64, UD, float64_muladd, 0)
+DO_4OP_F(vfmsub_s, 32, UW, float32_muladd, float_muladd_negate_c)
+DO_4OP_F(vfmsub_d, 64, UD, float64_muladd, float_muladd_negate_c)
+DO_4OP_F(vfnmadd_s, 32, UW, float32_muladd, float_muladd_negate_result)
+DO_4OP_F(vfnmadd_d, 64, UD, float64_muladd, float_muladd_negate_result)
+DO_4OP_F(vfnmsub_s, 32, UW, float32_muladd,
+ float_muladd_negate_c | float_muladd_negate_result)
+DO_4OP_F(vfnmsub_d, 64, UD, float64_muladd,
+ float_muladd_negate_c | float_muladd_negate_result)
+
+#define DO_2OP_F(NAME, BIT, E, FN) \
+void HELPER(NAME)(CPULoongArchState *env, uint32_t vd, uint32_t vj) \
+{ \
+ int i; \
+ VReg *Vd = &(env->fpr[vd].vreg); \
+ VReg *Vj = &(env->fpr[vj].vreg); \
+ \
+ vec_clear_cause(env); \
+ for (i = 0; i < LSX_LEN/BIT; i++) { \
+ Vd->E(i) = FN(env, Vj->E(i)); \
+ } \
+}
+
+#define FLOGB(BIT, T) \
+static T do_flogb_## BIT(CPULoongArchState *env, T fj) \
+{ \
+ T fp, fd; \
+ float_status *status = &env->fp_status; \
+ FloatRoundMode old_mode = get_float_rounding_mode(status); \
+ \
+ set_float_rounding_mode(float_round_down, status); \
+ fp = float ## BIT ##_log2(fj, status); \
+ fd = float ## BIT ##_round_to_int(fp, status); \
+ set_float_rounding_mode(old_mode, status); \
+ vec_update_fcsr0_mask(env, GETPC(), float_flag_inexact); \
+ return fd; \
+}
+
+FLOGB(32, uint32_t)
+FLOGB(64, uint64_t)
+
+#define FCLASS(NAME, BIT, E, FN) \
+void HELPER(NAME)(CPULoongArchState *env, uint32_t vd, uint32_t vj) \
+{ \
+ int i; \
+ VReg *Vd = &(env->fpr[vd].vreg); \
+ VReg *Vj = &(env->fpr[vj].vreg); \
+ \
+ for (i = 0; i < LSX_LEN/BIT; i++) { \
+ Vd->E(i) = FN(env, Vj->E(i)); \
+ } \
+}
+
+FCLASS(vfclass_s, 32, UW, helper_fclass_s)
+FCLASS(vfclass_d, 64, UD, helper_fclass_d)
+
+#define FSQRT(BIT, T) \
+static T do_fsqrt_## BIT(CPULoongArchState *env, T fj) \
+{ \
+ T fd; \
+ fd = float ## BIT ##_sqrt(fj, &env->fp_status); \
+ vec_update_fcsr0(env, GETPC()); \
+ return fd; \
+}
+
+FSQRT(32, uint32_t)
+FSQRT(64, uint64_t)
+
+#define FRECIP(BIT, T) \
+static T do_frecip_## BIT(CPULoongArchState *env, T fj) \
+{ \
+ T fd; \
+ fd = float ## BIT ##_div(float ## BIT ##_one, fj, &env->fp_status); \
+ vec_update_fcsr0(env, GETPC()); \
+ return fd; \
+}
+
+FRECIP(32, uint32_t)
+FRECIP(64, uint64_t)
+
+#define FRSQRT(BIT, T) \
+static T do_frsqrt_## BIT(CPULoongArchState *env, T fj) \
+{ \
+ T fd, fp; \
+ fp = float ## BIT ##_sqrt(fj, &env->fp_status); \
+ fd = float ## BIT ##_div(float ## BIT ##_one, fp, &env->fp_status); \
+ vec_update_fcsr0(env, GETPC()); \
+ return fd; \
+}
+
+FRSQRT(32, uint32_t)
+FRSQRT(64, uint64_t)
+
+DO_2OP_F(vflogb_s, 32, UW, do_flogb_32)
+DO_2OP_F(vflogb_d, 64, UD, do_flogb_64)
+DO_2OP_F(vfsqrt_s, 32, UW, do_fsqrt_32)
+DO_2OP_F(vfsqrt_d, 64, UD, do_fsqrt_64)
+DO_2OP_F(vfrecip_s, 32, UW, do_frecip_32)
+DO_2OP_F(vfrecip_d, 64, UD, do_frecip_64)
+DO_2OP_F(vfrsqrt_s, 32, UW, do_frsqrt_32)
+DO_2OP_F(vfrsqrt_d, 64, UD, do_frsqrt_64)
