#include "vec.h"
#include "tcg/tcg-gvec-desc.h"
-#define DO_ADD(a, b) (a + b)
-#define DO_SUB(a, b) (a - b)
-
#define DO_ODD_EVEN(NAME, BIT, E1, E2, DO_OP) \
void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
{ \
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 desc) \
{ \
DO_3OP(vavgr_wu, 32, UW, DO_VAVGR)
DO_3OP(vavgr_du, 64, UD, DO_VAVGR)
-#define DO_VABSD(a, b) ((a > b) ? (a -b) : (b-a))
-
DO_3OP(vabsd_b, 8, B, DO_VABSD)
DO_3OP(vabsd_h, 16, H, DO_VABSD)
DO_3OP(vabsd_w, 32, W, DO_VABSD)
DO_3OP(vabsd_wu, 32, UW, DO_VABSD)
DO_3OP(vabsd_du, 64, UD, DO_VABSD)
-#define DO_VABS(a) ((a < 0) ? (-a) : (a))
-
#define DO_VADDA(NAME, BIT, E) \
void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
{ \
DO_VADDA(vadda_w, 32, W)
DO_VADDA(vadda_d, 64, D)
-#define DO_MIN(a, b) (a < b ? a : b)
-#define DO_MAX(a, b) (a > b ? a : b)
-
#define VMINMAXI(NAME, BIT, E, DO_OP) \
void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
{ \
DO_VMUH(vmuh_hu, 16, UW, UH, DO_MUH)
DO_VMUH(vmuh_wu, 32, UD, UW, DO_MUH)
-#define DO_MUL(a, b) (a * b)
-
DO_EVEN(vmulwev_h_b, 16, H, B, DO_MUL)
DO_EVEN(vmulwev_w_h, 32, W, H, DO_MUL)
DO_EVEN(vmulwev_d_w, 64, D, W, DO_MUL)
DO_ODD_U_S(vmulwod_w_hu_h, 32, W, UW, H, UH, DO_MUL)
DO_ODD_U_S(vmulwod_d_wu_w, 64, D, UD, W, UW, DO_MUL)
-#define DO_MADD(a, b, c) (a + b * c)
-#define DO_MSUB(a, b, c) (a - b * c)
-
#define VMADDSUB(NAME, BIT, E, DO_OP) \
void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
{ \
VMADDWOD_U_S(vmaddwod_w_hu_h, 32, W, UW, H, UH, DO_MUL)
VMADDWOD_U_S(vmaddwod_d_wu_w, 64, D, UD, W, UW, DO_MUL)
-#define DO_DIVU(N, M) (unlikely(M == 0) ? 0 : N / M)
-#define DO_REMU(N, M) (unlikely(M == 0) ? 0 : N % M)
-#define DO_DIV(N, M) (unlikely(M == 0) ? 0 :\
- unlikely((N == -N) && (M == (__typeof(N))(-1))) ? N : N / M)
-#define DO_REM(N, M) (unlikely(M == 0) ? 0 :\
- unlikely((N == -N) && (M == (__typeof(N))(-1))) ? 0 : N % M)
-
#define VDIV(NAME, BIT, E, DO_OP) \
void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
{ \
VDIV(vmod_wu, 32, UW, DO_REMU)
VDIV(vmod_du, 64, UD, DO_REMU)
-#define VSAT_S(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t max, uint32_t v) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- typedef __typeof(Vd->E(0)) TD; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E(i) = Vj->E(i) > (TD)max ? (TD)max : \
- Vj->E(i) < (TD)~max ? (TD)~max: Vj->E(i); \
- } \
+#define VSAT_S(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t max, uint32_t desc) \
+{ \
+ int i; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ typedef __typeof(Vd->E(0)) TD; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
+ Vd->E(i) = Vj->E(i) > (TD)max ? (TD)max : \
+ Vj->E(i) < (TD)~max ? (TD)~max: Vj->E(i); \
+ } \
}
VSAT_S(vsat_b, 8, B)
VSAT_S(vsat_w, 32, W)
VSAT_S(vsat_d, 64, D)
-#define VSAT_U(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t max, uint32_t v) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- typedef __typeof(Vd->E(0)) TD; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E(i) = Vj->E(i) > (TD)max ? (TD)max : Vj->E(i); \
- } \
+#define VSAT_U(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t max, uint32_t desc) \
+{ \
+ int i; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ typedef __typeof(Vd->E(0)) TD; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
+ Vd->E(i) = Vj->E(i) > (TD)max ? (TD)max : Vj->E(i); \
+ } \
}
VSAT_U(vsat_bu, 8, UB)
VSAT_U(vsat_wu, 32, UW)
VSAT_U(vsat_du, 64, UD)
-#define VEXTH(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = Vj->E2(i + LSX_LEN/BIT); \
- } \
+#define VEXTH(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + i * ofs) = Vj->E2(j + ofs + ofs * 2 * i); \
+ } \
+ } \
}
void HELPER(vexth_q_d)(void *vd, void *vj, uint32_t desc)
{
+ int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- Vd->Q(0) = int128_makes64(Vj->D(1));
+ for (i = 0; i < oprsz / 16; i++) {
+ Vd->Q(i) = int128_makes64(Vj->D(2 * i + 1));
+ }
}
void HELPER(vexth_qu_du)(void *vd, void *vj, uint32_t desc)
{
+ int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- Vd->Q(0) = int128_make64((uint64_t)Vj->D(1));
+ for (i = 0; i < oprsz / 16; i++) {
+ Vd->Q(i) = int128_make64(Vj->UD(2 * i + 1));
+ }
}
VEXTH(vexth_h_b, 16, H, B)
VEXTH(vexth_wu_hu, 32, UW, UH)
VEXTH(vexth_du_wu, 64, UD, UW)
-#define DO_SIGNCOV(a, b) (a == 0 ? 0 : a < 0 ? -b : b)
+#define VEXT2XV(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint32_t desc) \
+{ \
+ int i; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
+ temp.E1(i) = Vj->E2(i); \
+ } \
+ *Vd = temp; \
+}
+
+VEXT2XV(vext2xv_h_b, 16, H, B)
+VEXT2XV(vext2xv_w_b, 32, W, B)
+VEXT2XV(vext2xv_d_b, 64, D, B)
+VEXT2XV(vext2xv_w_h, 32, W, H)
+VEXT2XV(vext2xv_d_h, 64, D, H)
+VEXT2XV(vext2xv_d_w, 64, D, W)
+VEXT2XV(vext2xv_hu_bu, 16, UH, UB)
+VEXT2XV(vext2xv_wu_bu, 32, UW, UB)
+VEXT2XV(vext2xv_du_bu, 64, UD, UB)
+VEXT2XV(vext2xv_wu_hu, 32, UW, UH)
+VEXT2XV(vext2xv_du_hu, 64, UD, UH)
+VEXT2XV(vext2xv_du_wu, 64, UD, UW)
DO_3OP(vsigncov_b, 8, B, DO_SIGNCOV)
DO_3OP(vsigncov_h, 16, H, DO_SIGNCOV)
void HELPER(vmskltz_b)(void *vd, void *vj, uint32_t desc)
{
+ int i;
uint16_t temp = 0;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- temp = do_vmskltz_b(Vj->D(0));
- temp |= (do_vmskltz_b(Vj->D(1)) << 8);
- Vd->D(0) = temp;
- Vd->D(1) = 0;
+ for (i = 0; i < oprsz / 16; i++) {
+ temp = 0;
+ temp = do_vmskltz_b(Vj->D(2 * i));
+ temp |= (do_vmskltz_b(Vj->D(2 * i + 1)) << 8);
+ Vd->D(2 * i) = temp;
+ Vd->D(2 * i + 1) = 0;
+ }
}
static uint64_t do_vmskltz_h(int64_t val)
void HELPER(vmskltz_h)(void *vd, void *vj, uint32_t desc)
{
+ int i;
uint16_t temp = 0;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- temp = do_vmskltz_h(Vj->D(0));
- temp |= (do_vmskltz_h(Vj->D(1)) << 4);
- Vd->D(0) = temp;
- Vd->D(1) = 0;
+ for (i = 0; i < oprsz / 16; i++) {
+ temp = 0;
+ temp = do_vmskltz_h(Vj->D(2 * i));
+ temp |= (do_vmskltz_h(Vj->D(2 * i + 1)) << 4);
+ Vd->D(2 * i) = temp;
+ Vd->D(2 * i + 1) = 0;
+ }
}
static uint64_t do_vmskltz_w(int64_t val)
void HELPER(vmskltz_w)(void *vd, void *vj, uint32_t desc)
{
+ int i;
uint16_t temp = 0;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- temp = do_vmskltz_w(Vj->D(0));
- temp |= (do_vmskltz_w(Vj->D(1)) << 2);
- Vd->D(0) = temp;
- Vd->D(1) = 0;
+ for (i = 0; i < oprsz / 16; i++) {
+ temp = 0;
+ temp = do_vmskltz_w(Vj->D(2 * i));
+ temp |= (do_vmskltz_w(Vj->D(2 * i + 1)) << 2);
+ Vd->D(2 * i) = temp;
+ Vd->D(2 * i + 1) = 0;
+ }
}
static uint64_t do_vmskltz_d(int64_t val)
}
void HELPER(vmskltz_d)(void *vd, void *vj, uint32_t desc)
{
+ int i;
uint16_t temp = 0;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- temp = do_vmskltz_d(Vj->D(0));
- temp |= (do_vmskltz_d(Vj->D(1)) << 1);
- Vd->D(0) = temp;
- Vd->D(1) = 0;
+ for (i = 0; i < oprsz / 16; i++) {
+ temp = 0;
+ temp = do_vmskltz_d(Vj->D(2 * i));
+ temp |= (do_vmskltz_d(Vj->D(2 * i + 1)) << 1);
+ Vd->D(2 * i) = temp;
+ Vd->D(2 * i + 1) = 0;
+ }
}
void HELPER(vmskgez_b)(void *vd, void *vj, uint32_t desc)
{
+ int i;
uint16_t temp = 0;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- temp = do_vmskltz_b(Vj->D(0));
- temp |= (do_vmskltz_b(Vj->D(1)) << 8);
- Vd->D(0) = (uint16_t)(~temp);
- Vd->D(1) = 0;
+ for (i = 0; i < oprsz / 16; i++) {
+ temp = 0;
+ temp = do_vmskltz_b(Vj->D(2 * i));
+ temp |= (do_vmskltz_b(Vj->D(2 * i + 1)) << 8);
+ Vd->D(2 * i) = (uint16_t)(~temp);
+ Vd->D(2 * i + 1) = 0;
+ }
}
static uint64_t do_vmskez_b(uint64_t a)
void HELPER(vmsknz_b)(void *vd, void *vj, uint32_t desc)
{
+ int i;
uint16_t temp = 0;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- temp = do_vmskez_b(Vj->D(0));
- temp |= (do_vmskez_b(Vj->D(1)) << 8);
- Vd->D(0) = (uint16_t)(~temp);
- Vd->D(1) = 0;
+ for (i = 0; i < oprsz / 16; i++) {
+ temp = 0;
+ temp = do_vmskez_b(Vj->D(2 * i));
+ temp |= (do_vmskez_b(Vj->D(2 * i + 1)) << 8);
+ Vd->D(2 * i) = (uint16_t)(~temp);
+ Vd->D(2 * i + 1) = 0;
+ }
}
-void HELPER(vnori_b)(void *vd, void *vj, uint64_t imm, uint32_t v)
+void HELPER(vnori_b)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
- for (i = 0; i < LSX_LEN/8; i++) {
+ for (i = 0; i < simd_oprsz(desc); i++) {
Vd->B(i) = ~(Vj->B(i) | (uint8_t)imm);
}
}
-#define VSLLWIL(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- typedef __typeof(temp.E1(0)) TD; \
- \
- temp.D(0) = 0; \
- temp.D(1) = 0; \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = (TD)Vj->E2(i) << (imm % BIT); \
- } \
- *Vd = temp; \
+#define VSLLWIL(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ typedef __typeof(temp.E1(0)) TD; \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * i) = (TD)Vj->E2(j + ofs * 2 * i) << (imm % BIT); \
+ } \
+ } \
+ *Vd = temp; \
}
+
void HELPER(vextl_q_d)(void *vd, void *vj, uint32_t desc)
{
+ int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- Vd->Q(0) = int128_makes64(Vj->D(0));
+ for (i = 0; i < oprsz / 16; i++) {
+ Vd->Q(i) = int128_makes64(Vj->D(2 * i));
+ }
}
void HELPER(vextl_qu_du)(void *vd, void *vj, uint32_t desc)
{
+ int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- Vd->Q(0) = int128_make64(Vj->D(0));
+ for (i = 0; i < oprsz / 16; i++) {
+ Vd->Q(i) = int128_make64(Vj->UD(2 * i));
+ }
}
VSLLWIL(vsllwil_h_b, 16, H, B)
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = do_vsrlr_ ## E(Vj->E(i), ((T)Vk->E(i))%BIT); \
} \
}
int i; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = do_vsrlr_ ## E(Vj->E(i), imm); \
} \
}
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = do_vsrar_ ## E(Vj->E(i), ((T)Vk->E(i))%BIT); \
} \
}
int i; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = do_vsrar_ ## E(Vj->E(i), imm); \
} \
}
VSRARI(vsrari_w, 32, W)
VSRARI(vsrari_d, 64, D)
-#define R_SHIFT(a, b) (a >> b)
-
-#define VSRLN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = R_SHIFT((T)Vj->E2(i),((T)Vk->E2(i)) % BIT); \
- } \
- Vd->D(1) = 0; \
+#define VSRLN(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = R_SHIFT(Vj->E2(j + ofs * i), \
+ Vk->E2(j + ofs * i) % BIT); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSRLN(vsrln_b_h, 16, uint16_t, B, H)
-VSRLN(vsrln_h_w, 32, uint32_t, H, W)
-VSRLN(vsrln_w_d, 64, uint64_t, W, D)
+VSRLN(vsrln_b_h, 16, B, UH)
+VSRLN(vsrln_h_w, 32, H, UW)
+VSRLN(vsrln_w_d, 64, W, UD)
-#define VSRAN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = R_SHIFT(Vj->E2(i), ((T)Vk->E2(i)) % BIT); \
- } \
- Vd->D(1) = 0; \
+#define VSRAN(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = R_SHIFT(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSRAN(vsran_b_h, 16, uint16_t, B, H)
-VSRAN(vsran_h_w, 32, uint32_t, H, W)
-VSRAN(vsran_w_d, 64, uint64_t, W, D)
+VSRAN(vsran_b_h, 16, B, H, UH)
+VSRAN(vsran_h_w, 32, H, W, UW)
+VSRAN(vsran_w_d, 64, W, D, UD)
-#define VSRLNI(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i, max; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- temp.D(0) = 0; \
- temp.D(1) = 0; \
- max = LSX_LEN/BIT; \
- for (i = 0; i < max; i++) { \
- temp.E1(i) = R_SHIFT((T)Vj->E2(i), imm); \
- temp.E1(i + max) = R_SHIFT((T)Vd->E2(i), imm); \
- } \
- *Vd = temp; \
+#define VSRLNI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = R_SHIFT(Vj->E2(j + ofs * i), imm); \
+ temp.E1(j + ofs * (2 * i + 1)) = R_SHIFT(Vd->E2(j + ofs * i), \
+ imm); \
+ } \
+ } \
+ *Vd = temp; \
}
void HELPER(vsrlni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
- VReg temp;
+ int i;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
- temp.D(0) = 0;
- temp.D(1) = 0;
- temp.D(0) = int128_getlo(int128_urshift(Vj->Q(0), imm % 128));
- temp.D(1) = int128_getlo(int128_urshift(Vd->Q(0), imm % 128));
+ for (i = 0; i < 2; i++) {
+ temp.D(2 * i) = int128_getlo(int128_urshift(Vj->Q(i), imm % 128));
+ temp.D(2 * i +1) = int128_getlo(int128_urshift(Vd->Q(i), imm % 128));
+ }
*Vd = temp;
}
-VSRLNI(vsrlni_b_h, 16, uint16_t, B, H)
-VSRLNI(vsrlni_h_w, 32, uint32_t, H, W)
-VSRLNI(vsrlni_w_d, 64, uint64_t, W, D)
+VSRLNI(vsrlni_b_h, 16, B, UH)
+VSRLNI(vsrlni_h_w, 32, H, UW)
+VSRLNI(vsrlni_w_d, 64, W, UD)
-#define VSRANI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i, max; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- temp.D(0) = 0; \
- temp.D(1) = 0; \
- max = LSX_LEN/BIT; \
- for (i = 0; i < max; i++) { \
- temp.E1(i) = R_SHIFT(Vj->E2(i), imm); \
- temp.E1(i + max) = R_SHIFT(Vd->E2(i), imm); \
- } \
- *Vd = temp; \
+#define VSRANI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = R_SHIFT(Vj->E2(j + ofs * i), imm); \
+ temp.E1(j + ofs * (2 * i + 1)) = R_SHIFT(Vd->E2(j + ofs * i), \
+ imm); \
+ } \
+ } \
+ *Vd = temp; \
}
void HELPER(vsrani_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
- VReg temp;
+ int i;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
- temp.D(0) = 0;
- temp.D(1) = 0;
- temp.D(0) = int128_getlo(int128_rshift(Vj->Q(0), imm % 128));
- temp.D(1) = int128_getlo(int128_rshift(Vd->Q(0), imm % 128));
+ for (i = 0; i < 2; i++) {
+ temp.D(2 * i) = int128_getlo(int128_rshift(Vj->Q(i), imm % 128));
+ temp.D(2 * i + 1) = int128_getlo(int128_rshift(Vd->Q(i), imm % 128));
+ }
*Vd = temp;
}
VSRANI(vsrani_h_w, 32, H, W)
VSRANI(vsrani_w_d, 64, W, D)
-#define VSRLRN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_vsrlr_ ## E2(Vj->E2(i), ((T)Vk->E2(i))%BIT); \
- } \
- Vd->D(1) = 0; \
+#define VSRLRN(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_vsrlr_ ##E2(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSRLRN(vsrlrn_b_h, 16, uint16_t, B, H)
-VSRLRN(vsrlrn_h_w, 32, uint32_t, H, W)
-VSRLRN(vsrlrn_w_d, 64, uint64_t, W, D)
+VSRLRN(vsrlrn_b_h, 16, B, H, UH)
+VSRLRN(vsrlrn_h_w, 32, H, W, UW)
+VSRLRN(vsrlrn_w_d, 64, W, D, UD)
-#define VSRARN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_vsrar_ ## E2(Vj->E2(i), ((T)Vk->E2(i))%BIT); \
- } \
- Vd->D(1) = 0; \
+#define VSRARN(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_vsrar_ ## E2(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSRARN(vsrarn_b_h, 16, uint8_t, B, H)
-VSRARN(vsrarn_h_w, 32, uint16_t, H, W)
-VSRARN(vsrarn_w_d, 64, uint32_t, W, D)
-
-#define VSRLRNI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i, max; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- temp.D(0) = 0; \
- temp.D(1) = 0; \
- max = LSX_LEN/BIT; \
- for (i = 0; i < max; i++) { \
- temp.E1(i) = do_vsrlr_ ## E2(Vj->E2(i), imm); \
- temp.E1(i + max) = do_vsrlr_ ## E2(Vd->E2(i), imm); \
- } \
- *Vd = temp; \
+VSRARN(vsrarn_b_h, 16, B, H, UH)
+VSRARN(vsrarn_h_w, 32, H, W, UW)
+VSRARN(vsrarn_w_d, 64, W, D, UD)
+
+#define VSRLRNI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_vsrlr_ ## E2(Vj->E2(j + ofs * i), imm); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_vsrlr_ ## E2(Vd->E2(j + ofs * i), \
+ imm); \
+ } \
+ } \
+ *Vd = temp; \
}
void HELPER(vsrlrni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
- VReg temp;
+ int i;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
- Int128 r1, r2;
-
- if (imm == 0) {
- temp.D(0) = int128_getlo(Vj->Q(0));
- temp.D(1) = int128_getlo(Vd->Q(0));
- } else {
- r1 = int128_and(int128_urshift(Vj->Q(0), (imm -1)), int128_one());
- r2 = int128_and(int128_urshift(Vd->Q(0), (imm -1)), int128_one());
+ Int128 r[4];
+ int oprsz = simd_oprsz(desc);
- temp.D(0) = int128_getlo(int128_add(int128_urshift(Vj->Q(0), imm), r1));
- temp.D(1) = int128_getlo(int128_add(int128_urshift(Vd->Q(0), imm), r2));
+ for (i = 0; i < oprsz / 16; i++) {
+ if (imm == 0) {
+ temp.D(2 * i) = int128_getlo(Vj->Q(i));
+ temp.D(2 * i + 1) = int128_getlo(Vd->Q(i));
+ } else {
+ r[2 * i] = int128_and(int128_urshift(Vj->Q(i), (imm - 1)),
+ int128_one());
+ r[2 * i + 1] = int128_and(int128_urshift(Vd->Q(i), (imm - 1)),
+ int128_one());
+ temp.D(2 * i) = int128_getlo(int128_add(int128_urshift(Vj->Q(i),
+ imm), r[2 * i]));
+ temp.D(2 * i + 1) = int128_getlo(int128_add(int128_urshift(Vd->Q(i),
+ imm), r[ 2 * i + 1]));
+ }
}
*Vd = temp;
}
VSRLRNI(vsrlrni_h_w, 32, H, W)
VSRLRNI(vsrlrni_w_d, 64, W, D)
-#define VSRARNI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i, max; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- temp.D(0) = 0; \
- temp.D(1) = 0; \
- max = LSX_LEN/BIT; \
- for (i = 0; i < max; i++) { \
- temp.E1(i) = do_vsrar_ ## E2(Vj->E2(i), imm); \
- temp.E1(i + max) = do_vsrar_ ## E2(Vd->E2(i), imm); \
- } \
- *Vd = temp; \
+#define VSRARNI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_vsrar_ ## E2(Vj->E2(j + ofs * i), imm); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_vsrar_ ## E2(Vd->E2(j + ofs * i), \
+ imm); \
+ } \
+ } \
+ *Vd = temp; \
}
void HELPER(vsrarni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
- VReg temp;
+ int i;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
- Int128 r1, r2;
-
- if (imm == 0) {
- temp.D(0) = int128_getlo(Vj->Q(0));
- temp.D(1) = int128_getlo(Vd->Q(0));
- } else {
- r1 = int128_and(int128_rshift(Vj->Q(0), (imm -1)), int128_one());
- r2 = int128_and(int128_rshift(Vd->Q(0), (imm -1)), int128_one());
+ Int128 r[4];
+ int oprsz = simd_oprsz(desc);
- temp.D(0) = int128_getlo(int128_add(int128_rshift(Vj->Q(0), imm), r1));
- temp.D(1) = int128_getlo(int128_add(int128_rshift(Vd->Q(0), imm), r2));
+ for (i = 0; i < oprsz / 16; i++) {
+ if (imm == 0) {
+ temp.D(2 * i) = int128_getlo(Vj->Q(i));
+ temp.D(2 * i + 1) = int128_getlo(Vd->Q(i));
+ } else {
+ r[2 * i] = int128_and(int128_rshift(Vj->Q(i), (imm - 1)),
+ int128_one());
+ r[2 * i + 1] = int128_and(int128_rshift(Vd->Q(i), (imm - 1)),
+ int128_one());
+ temp.D(2 * i) = int128_getlo(int128_add(int128_rshift(Vj->Q(i),
+ imm), r[2 * i]));
+ temp.D(2 * i + 1) = int128_getlo(int128_add(int128_rshift(Vd->Q(i),
+ imm), r[2 * i + 1]));
+ }
}
*Vd = temp;
}
SSRLNS(H, uint32_t, int32_t, uint16_t)
SSRLNS(W, uint64_t, int64_t, uint32_t)
-#define VSSRLN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssrlns_ ## E1(Vj->E2(i), (T)Vk->E2(i)% BIT, BIT/2 -1); \
- } \
- Vd->D(1) = 0; \
+#define VSSRLN(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssrlns_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT / 2 - 1); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSSRLN(vssrln_b_h, 16, uint16_t, B, H)
-VSSRLN(vssrln_h_w, 32, uint32_t, H, W)
-VSSRLN(vssrln_w_d, 64, uint64_t, W, D)
+VSSRLN(vssrln_b_h, 16, B, H, UH)
+VSSRLN(vssrln_h_w, 32, H, W, UW)
+VSSRLN(vssrln_w_d, 64, W, D, UD)
#define SSRANS(E, T1, T2) \
static T1 do_ssrans_ ## E(T1 e2, int sa, int sh) \
shft_res = e2 >> sa; \
} \
T2 mask; \
- mask = (1ll << sh) -1; \
+ mask = (1ll << sh) - 1; \
if (shft_res > mask) { \
return mask; \
- } else if (shft_res < -(mask +1)) { \
+ } else if (shft_res < -(mask + 1)) { \
return ~mask; \
} else { \
return shft_res; \
SSRANS(H, int32_t, int16_t)
SSRANS(W, int64_t, int32_t)
-#define VSSRAN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssrans_ ## E1(Vj->E2(i), (T)Vk->E2(i)%BIT, BIT/2 -1); \
- } \
- Vd->D(1) = 0; \
+#define VSSRAN(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssrans_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT / 2 - 1); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSSRAN(vssran_b_h, 16, uint16_t, B, H)
-VSSRAN(vssran_h_w, 32, uint32_t, H, W)
-VSSRAN(vssran_w_d, 64, uint64_t, W, D)
+VSSRAN(vssran_b_h, 16, B, H, UH)
+VSSRAN(vssran_h_w, 32, H, W, UW)
+VSSRAN(vssran_w_d, 64, W, D, UD)
#define SSRLNU(E, T1, T2, T3) \
static T1 do_ssrlnu_ ## E(T3 e2, int sa, int sh) \
shft_res = (((T1)e2) >> sa); \
} \
T2 mask; \
- mask = (1ull << sh) -1; \
+ mask = (1ull << sh) - 1; \
if (shft_res > mask) { \
return mask; \
} else { \
SSRLNU(H, uint32_t, uint16_t, int32_t)
SSRLNU(W, uint64_t, uint32_t, int64_t)
-#define VSSRLNU(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssrlnu_ ## E1(Vj->E2(i), (T)Vk->E2(i)%BIT, BIT/2); \
- } \
- Vd->D(1) = 0; \
+#define VSSRLNU(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssrlnu_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT / 2); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSSRLNU(vssrln_bu_h, 16, uint16_t, B, H)
-VSSRLNU(vssrln_hu_w, 32, uint32_t, H, W)
-VSSRLNU(vssrln_wu_d, 64, uint64_t, W, D)
+VSSRLNU(vssrln_bu_h, 16, B, H, UH)
+VSSRLNU(vssrln_hu_w, 32, H, W, UW)
+VSSRLNU(vssrln_wu_d, 64, W, D, UD)
#define SSRANU(E, T1, T2, T3) \
static T1 do_ssranu_ ## E(T3 e2, int sa, int sh) \
shft_res = 0; \
} \
T2 mask; \
- mask = (1ull << sh) -1; \
+ mask = (1ull << sh) - 1; \
if (shft_res > mask) { \
return mask; \
} else { \
SSRANU(H, uint32_t, uint16_t, int32_t)
SSRANU(W, uint64_t, uint32_t, int64_t)
-#define VSSRANU(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssranu_ ## E1(Vj->E2(i), (T)Vk->E2(i)%BIT, BIT/2); \
- } \
- Vd->D(1) = 0; \
-}
-
-VSSRANU(vssran_bu_h, 16, uint16_t, B, H)
-VSSRANU(vssran_hu_w, 32, uint32_t, H, W)
-VSSRANU(vssran_wu_d, 64, uint64_t, W, D)
-
-#define VSSRLNI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssrlns_ ## E1(Vj->E2(i), imm, BIT/2 -1); \
- temp.E1(i + LSX_LEN/BIT) = do_ssrlns_ ## E1(Vd->E2(i), imm, BIT/2 -1);\
- } \
- *Vd = temp; \
+#define VSSRANU(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssranu_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT / 2); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-void HELPER(vssrlni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
-{
- Int128 shft_res1, shft_res2, mask;
- VReg *Vd = (VReg *)vd;
- VReg *Vj = (VReg *)vj;
+VSSRANU(vssran_bu_h, 16, B, H, UH)
+VSSRANU(vssran_hu_w, 32, H, W, UW)
+VSSRANU(vssran_wu_d, 64, W, D, UD)
+
+#define VSSRLNI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssrlns_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssrlns_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ } \
+ } \
+ *Vd = temp; \
+}
+
+static void do_vssrlni_q(VReg *Vd, VReg *Vj,
+ uint64_t imm, int idx, Int128 mask)
+{
+ Int128 shft_res1, shft_res2;
if (imm == 0) {
- shft_res1 = Vj->Q(0);
- shft_res2 = Vd->Q(0);
+ shft_res1 = Vj->Q(idx);
+ shft_res2 = Vd->Q(idx);
} else {
- shft_res1 = int128_urshift(Vj->Q(0), imm);
- shft_res2 = int128_urshift(Vd->Q(0), imm);
+ shft_res1 = int128_urshift(Vj->Q(idx), imm);
+ shft_res2 = int128_urshift(Vd->Q(idx), imm);
}
- mask = int128_sub(int128_lshift(int128_one(), 63), int128_one());
if (int128_ult(mask, shft_res1)) {
- Vd->D(0) = int128_getlo(mask);
+ Vd->D(idx * 2) = int128_getlo(mask);
}else {
- Vd->D(0) = int128_getlo(shft_res1);
+ Vd->D(idx * 2) = int128_getlo(shft_res1);
}
if (int128_ult(mask, shft_res2)) {
- Vd->D(1) = int128_getlo(mask);
+ Vd->D(idx * 2 + 1) = int128_getlo(mask);
}else {
- Vd->D(1) = int128_getlo(shft_res2);
+ Vd->D(idx * 2 + 1) = int128_getlo(shft_res2);
+ }
+}
+
+void HELPER(vssrlni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
+ Int128 mask;
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ mask = int128_sub(int128_lshift(int128_one(), 63), int128_one());
+
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrlni_q(Vd, Vj, imm, i, mask);
}
}
VSSRLNI(vssrlni_h_w, 32, H, W)
VSSRLNI(vssrlni_w_d, 64, W, D)
-#define VSSRANI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssrans_ ## E1(Vj->E2(i), imm, BIT/2 -1); \
- temp.E1(i + LSX_LEN/BIT) = do_ssrans_ ## E1(Vd->E2(i), imm, BIT/2 -1); \
- } \
- *Vd = temp; \
-}
-
-void HELPER(vssrani_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
-{
- Int128 shft_res1, shft_res2, mask, min;
- VReg *Vd = (VReg *)vd;
- VReg *Vj = (VReg *)vj;
+#define VSSRANI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssrans_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssrans_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ } \
+ } \
+ *Vd = temp; \
+}
+
+static void do_vssrani_d_q(VReg *Vd, VReg *Vj,
+ uint64_t imm, int idx, Int128 mask, Int128 min)
+{
+ Int128 shft_res1, shft_res2;
if (imm == 0) {
- shft_res1 = Vj->Q(0);
- shft_res2 = Vd->Q(0);
+ shft_res1 = Vj->Q(idx);
+ shft_res2 = Vd->Q(idx);
} else {
- shft_res1 = int128_rshift(Vj->Q(0), imm);
- shft_res2 = int128_rshift(Vd->Q(0), imm);
+ shft_res1 = int128_rshift(Vj->Q(idx), imm);
+ shft_res2 = int128_rshift(Vd->Q(idx), imm);
}
- mask = int128_sub(int128_lshift(int128_one(), 63), int128_one());
- min = int128_lshift(int128_one(), 63);
- if (int128_gt(shft_res1, mask)) {
- Vd->D(0) = int128_getlo(mask);
+ if (int128_gt(shft_res1, mask)) {
+ Vd->D(idx * 2) = int128_getlo(mask);
} else if (int128_lt(shft_res1, int128_neg(min))) {
- Vd->D(0) = int128_getlo(min);
+ Vd->D(idx * 2) = int128_getlo(min);
} else {
- Vd->D(0) = int128_getlo(shft_res1);
+ Vd->D(idx * 2) = int128_getlo(shft_res1);
}
if (int128_gt(shft_res2, mask)) {
- Vd->D(1) = int128_getlo(mask);
+ Vd->D(idx * 2 + 1) = int128_getlo(mask);
} else if (int128_lt(shft_res2, int128_neg(min))) {
- Vd->D(1) = int128_getlo(min);
+ Vd->D(idx * 2 + 1) = int128_getlo(min);
} else {
- Vd->D(1) = int128_getlo(shft_res2);
+ Vd->D(idx * 2 + 1) = int128_getlo(shft_res2);
}
}
+void HELPER(vssrani_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
+ Int128 mask, min;
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ mask = int128_sub(int128_lshift(int128_one(), 63), int128_one());
+ min = int128_lshift(int128_one(), 63);
+
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrani_d_q(Vd, Vj, imm, i, mask, min);
+ }
+}
+
+
VSSRANI(vssrani_b_h, 16, B, H)
VSSRANI(vssrani_h_w, 32, H, W)
VSSRANI(vssrani_w_d, 64, W, D)
-#define VSSRLNUI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssrlnu_ ## E1(Vj->E2(i), imm, BIT/2); \
- temp.E1(i + LSX_LEN/BIT) = do_ssrlnu_ ## E1(Vd->E2(i), imm, BIT/2); \
- } \
- *Vd = temp; \
+#define VSSRLNUI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssrlnu_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssrlnu_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ } \
+ } \
+ *Vd = temp; \
}
void HELPER(vssrlni_du_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
- Int128 shft_res1, shft_res2, mask;
+ int i;
+ Int128 mask;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- if (imm == 0) {
- shft_res1 = Vj->Q(0);
- shft_res2 = Vd->Q(0);
- } else {
- shft_res1 = int128_urshift(Vj->Q(0), imm);
- shft_res2 = int128_urshift(Vd->Q(0), imm);
- }
mask = int128_sub(int128_lshift(int128_one(), 64), int128_one());
- if (int128_ult(mask, shft_res1)) {
- Vd->D(0) = int128_getlo(mask);
- }else {
- Vd->D(0) = int128_getlo(shft_res1);
- }
-
- if (int128_ult(mask, shft_res2)) {
- Vd->D(1) = int128_getlo(mask);
- }else {
- Vd->D(1) = int128_getlo(shft_res2);
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrlni_q(Vd, Vj, imm, i, mask);
}
}
VSSRLNUI(vssrlni_hu_w, 32, H, W)
VSSRLNUI(vssrlni_wu_d, 64, W, D)
-#define VSSRANUI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssranu_ ## E1(Vj->E2(i), imm, BIT/2); \
- temp.E1(i + LSX_LEN/BIT) = do_ssranu_ ## E1(Vd->E2(i), imm, BIT/2); \
- } \
- *Vd = temp; \
-}
-
-void HELPER(vssrani_du_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
-{
- Int128 shft_res1, shft_res2, mask;
- VReg *Vd = (VReg *)vd;
- VReg *Vj = (VReg *)vj;
+#define VSSRANUI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssranu_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssranu_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ } \
+ } \
+ *Vd = temp; \
+}
+
+static void do_vssrani_du_q(VReg *Vd, VReg *Vj,
+ uint64_t imm, int idx, Int128 mask)
+{
+ Int128 shft_res1, shft_res2;
if (imm == 0) {
- shft_res1 = Vj->Q(0);
- shft_res2 = Vd->Q(0);
+ shft_res1 = Vj->Q(idx);
+ shft_res2 = Vd->Q(idx);
} else {
- shft_res1 = int128_rshift(Vj->Q(0), imm);
- shft_res2 = int128_rshift(Vd->Q(0), imm);
+ shft_res1 = int128_rshift(Vj->Q(idx), imm);
+ shft_res2 = int128_rshift(Vd->Q(idx), imm);
}
- if (int128_lt(Vj->Q(0), int128_zero())) {
+ if (int128_lt(Vj->Q(idx), int128_zero())) {
shft_res1 = int128_zero();
}
- if (int128_lt(Vd->Q(0), int128_zero())) {
+ if (int128_lt(Vd->Q(idx), int128_zero())) {
shft_res2 = int128_zero();
}
-
- mask = int128_sub(int128_lshift(int128_one(), 64), int128_one());
-
if (int128_ult(mask, shft_res1)) {
- Vd->D(0) = int128_getlo(mask);
+ Vd->D(idx * 2) = int128_getlo(mask);
}else {
- Vd->D(0) = int128_getlo(shft_res1);
+ Vd->D(idx * 2) = int128_getlo(shft_res1);
}
if (int128_ult(mask, shft_res2)) {
- Vd->D(1) = int128_getlo(mask);
+ Vd->D(idx * 2 + 1) = int128_getlo(mask);
}else {
- Vd->D(1) = int128_getlo(shft_res2);
+ Vd->D(idx * 2 + 1) = int128_getlo(shft_res2);
+ }
+
+}
+
+void HELPER(vssrani_du_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
+ Int128 mask;
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ mask = int128_sub(int128_lshift(int128_one(), 64), int128_one());
+
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrani_du_q(Vd, Vj, imm, i, mask);
}
}
\
shft_res = do_vsrlr_ ## E2(e2, sa); \
T1 mask; \
- mask = (1ull << sh) -1; \
+ mask = (1ull << sh) - 1; \
if (shft_res > mask) { \
return mask; \
} else { \
SSRLRNS(H, W, uint32_t, int32_t, uint16_t)
SSRLRNS(W, D, uint64_t, int64_t, uint32_t)
-#define VSSRLRN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssrlrns_ ## E1(Vj->E2(i), (T)Vk->E2(i)%BIT, BIT/2 -1); \
- } \
- Vd->D(1) = 0; \
+#define VSSRLRN(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssrlrns_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT / 2 - 1); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSSRLRN(vssrlrn_b_h, 16, uint16_t, B, H)
-VSSRLRN(vssrlrn_h_w, 32, uint32_t, H, W)
-VSSRLRN(vssrlrn_w_d, 64, uint64_t, W, D)
+VSSRLRN(vssrlrn_b_h, 16, B, H, UH)
+VSSRLRN(vssrlrn_h_w, 32, H, W, UW)
+VSSRLRN(vssrlrn_w_d, 64, W, D, UD)
#define SSRARNS(E1, E2, T1, T2) \
static T1 do_ssrarns_ ## E1(T1 e2, int sa, int sh) \
\
shft_res = do_vsrar_ ## E2(e2, sa); \
T2 mask; \
- mask = (1ll << sh) -1; \
+ mask = (1ll << sh) - 1; \
if (shft_res > mask) { \
return mask; \
} else if (shft_res < -(mask +1)) { \
SSRARNS(H, W, int32_t, int16_t)
SSRARNS(W, D, int64_t, int32_t)
-#define VSSRARN(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssrarns_ ## E1(Vj->E2(i), (T)Vk->E2(i)%BIT, BIT/2 -1); \
- } \
- Vd->D(1) = 0; \
+#define VSSRARN(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssrarns_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT/ 2 - 1); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSSRARN(vssrarn_b_h, 16, uint16_t, B, H)
-VSSRARN(vssrarn_h_w, 32, uint32_t, H, W)
-VSSRARN(vssrarn_w_d, 64, uint64_t, W, D)
+VSSRARN(vssrarn_b_h, 16, B, H, UH)
+VSSRARN(vssrarn_h_w, 32, H, W, UW)
+VSSRARN(vssrarn_w_d, 64, W, D, UD)
#define SSRLRNU(E1, E2, T1, T2, T3) \
static T1 do_ssrlrnu_ ## E1(T3 e2, int sa, int sh) \
shft_res = do_vsrlr_ ## E2(e2, sa); \
\
T2 mask; \
- mask = (1ull << sh) -1; \
+ mask = (1ull << sh) - 1; \
if (shft_res > mask) { \
return mask; \
} else { \
SSRLRNU(H, W, uint32_t, uint16_t, int32_t)
SSRLRNU(W, D, uint64_t, uint32_t, int64_t)
-#define VSSRLRNU(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssrlrnu_ ## E1(Vj->E2(i), (T)Vk->E2(i)%BIT, BIT/2); \
- } \
- Vd->D(1) = 0; \
+#define VSSRLRNU(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssrlrnu_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT / 2); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSSRLRNU(vssrlrn_bu_h, 16, uint16_t, B, H)
-VSSRLRNU(vssrlrn_hu_w, 32, uint32_t, H, W)
-VSSRLRNU(vssrlrn_wu_d, 64, uint64_t, W, D)
+VSSRLRNU(vssrlrn_bu_h, 16, B, H, UH)
+VSSRLRNU(vssrlrn_hu_w, 32, H, W, UW)
+VSSRLRNU(vssrlrn_wu_d, 64, W, D, UD)
#define SSRARNU(E1, E2, T1, T2, T3) \
static T1 do_ssrarnu_ ## E1(T3 e2, int sa, int sh) \
shft_res = do_vsrar_ ## E2(e2, sa); \
} \
T2 mask; \
- mask = (1ull << sh) -1; \
+ mask = (1ull << sh) - 1; \
if (shft_res > mask) { \
return mask; \
} else { \
SSRARNU(H, W, uint32_t, uint16_t, int32_t)
SSRARNU(W, D, uint64_t, uint32_t, int64_t)
-#define VSSRARNU(NAME, BIT, T, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E1(i) = do_ssrarnu_ ## E1(Vj->E2(i), (T)Vk->E2(i)%BIT, BIT/2); \
- } \
- Vd->D(1) = 0; \
+#define VSSRARNU(NAME, BIT, E1, E2, E3) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ Vd->E1(j + ofs * 2 * i) = do_ssrarnu_ ## E1(Vj->E2(j + ofs * i), \
+ Vk->E3(j + ofs * i) % BIT, \
+ BIT / 2); \
+ } \
+ Vd->D(2 * i + 1) = 0; \
+ } \
}
-VSSRARNU(vssrarn_bu_h, 16, uint16_t, B, H)
-VSSRARNU(vssrarn_hu_w, 32, uint32_t, H, W)
-VSSRARNU(vssrarn_wu_d, 64, uint64_t, W, D)
+VSSRARNU(vssrarn_bu_h, 16, B, H, UH)
+VSSRARNU(vssrarn_hu_w, 32, H, W, UW)
+VSSRARNU(vssrarn_wu_d, 64, W, D, UD)
+
+#define VSSRLRNI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssrlrns_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssrlrns_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ } \
+ } \
+ *Vd = temp; \
+}
+
+static void do_vssrlrni_q(VReg *Vd, VReg * Vj,
+ uint64_t imm, int idx, Int128 mask)
+{
+ Int128 shft_res1, shft_res2, r1, r2;
+ if (imm == 0) {
+ shft_res1 = Vj->Q(idx);
+ shft_res2 = Vd->Q(idx);
+ } else {
+ r1 = int128_and(int128_urshift(Vj->Q(idx), (imm - 1)), int128_one());
+ r2 = int128_and(int128_urshift(Vd->Q(idx), (imm - 1)), int128_one());
+ shft_res1 = (int128_add(int128_urshift(Vj->Q(idx), imm), r1));
+ shft_res2 = (int128_add(int128_urshift(Vd->Q(idx), imm), r2));
+ }
-#define VSSRLRNI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssrlrns_ ## E1(Vj->E2(i), imm, BIT/2 -1); \
- temp.E1(i + LSX_LEN/BIT) = do_ssrlrns_ ## E1(Vd->E2(i), imm, BIT/2 -1);\
- } \
- *Vd = temp; \
+ if (int128_ult(mask, shft_res1)) {
+ Vd->D(idx * 2) = int128_getlo(mask);
+ }else {
+ Vd->D(idx * 2) = int128_getlo(shft_res1);
+ }
+
+ if (int128_ult(mask, shft_res2)) {
+ Vd->D(idx * 2 + 1) = int128_getlo(mask);
+ }else {
+ Vd->D(idx * 2 + 1) = int128_getlo(shft_res2);
+ }
}
-#define VSSRLRNI_Q(NAME, sh) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- Int128 shft_res1, shft_res2, mask, r1, r2; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- if (imm == 0) { \
- shft_res1 = Vj->Q(0); \
- shft_res2 = Vd->Q(0); \
- } else { \
- r1 = int128_and(int128_urshift(Vj->Q(0), (imm -1)), int128_one()); \
- r2 = int128_and(int128_urshift(Vd->Q(0), (imm -1)), int128_one()); \
- \
- shft_res1 = (int128_add(int128_urshift(Vj->Q(0), imm), r1)); \
- shft_res2 = (int128_add(int128_urshift(Vd->Q(0), imm), r2)); \
- } \
- \
- mask = int128_sub(int128_lshift(int128_one(), sh), int128_one()); \
- \
- if (int128_ult(mask, shft_res1)) { \
- Vd->D(0) = int128_getlo(mask); \
- }else { \
- Vd->D(0) = int128_getlo(shft_res1); \
- } \
- \
- if (int128_ult(mask, shft_res2)) { \
- Vd->D(1) = int128_getlo(mask); \
- }else { \
- Vd->D(1) = int128_getlo(shft_res2); \
- } \
+void HELPER(vssrlrni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
+ Int128 mask;
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ mask = int128_sub(int128_lshift(int128_one(), 63), int128_one());
+
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrlrni_q(Vd, Vj, imm, i, mask);
+ }
}
VSSRLRNI(vssrlrni_b_h, 16, B, H)
VSSRLRNI(vssrlrni_h_w, 32, H, W)
VSSRLRNI(vssrlrni_w_d, 64, W, D)
-VSSRLRNI_Q(vssrlrni_d_q, 63)
-
-#define VSSRARNI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssrarns_ ## E1(Vj->E2(i), imm, BIT/2 -1); \
- temp.E1(i + LSX_LEN/BIT) = do_ssrarns_ ## E1(Vd->E2(i), imm, BIT/2 -1); \
- } \
- *Vd = temp; \
-}
-void HELPER(vssrarni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
-{
- Int128 shft_res1, shft_res2, mask1, mask2, r1, r2;
- VReg *Vd = (VReg *)vd;
- VReg *Vj = (VReg *)vj;
+#define VSSRARNI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssrarns_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssrarns_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2 - 1); \
+ } \
+ } \
+ *Vd = temp; \
+}
+
+static void do_vssrarni_d_q(VReg *Vd, VReg *Vj,
+ uint64_t imm, int idx, Int128 mask1, Int128 mask2)
+{
+ Int128 shft_res1, shft_res2, r1, r2;
if (imm == 0) {
- shft_res1 = Vj->Q(0);
- shft_res2 = Vd->Q(0);
+ shft_res1 = Vj->Q(idx);
+ shft_res2 = Vd->Q(idx);
} else {
- r1 = int128_and(int128_rshift(Vj->Q(0), (imm -1)), int128_one());
- r2 = int128_and(int128_rshift(Vd->Q(0), (imm -1)), int128_one());
-
- shft_res1 = int128_add(int128_rshift(Vj->Q(0), imm), r1);
- shft_res2 = int128_add(int128_rshift(Vd->Q(0), imm), r2);
+ r1 = int128_and(int128_rshift(Vj->Q(idx), (imm - 1)), int128_one());
+ r2 = int128_and(int128_rshift(Vd->Q(idx), (imm - 1)), int128_one());
+ shft_res1 = int128_add(int128_rshift(Vj->Q(idx), imm), r1);
+ shft_res2 = int128_add(int128_rshift(Vd->Q(idx), imm), r2);
}
-
- mask1 = int128_sub(int128_lshift(int128_one(), 63), int128_one());
- mask2 = int128_lshift(int128_one(), 63);
-
- if (int128_gt(shft_res1, mask1)) {
- Vd->D(0) = int128_getlo(mask1);
+ if (int128_gt(shft_res1, mask1)) {
+ Vd->D(idx * 2) = int128_getlo(mask1);
} else if (int128_lt(shft_res1, int128_neg(mask2))) {
- Vd->D(0) = int128_getlo(mask2);
+ Vd->D(idx * 2) = int128_getlo(mask2);
} else {
- Vd->D(0) = int128_getlo(shft_res1);
+ Vd->D(idx * 2) = int128_getlo(shft_res1);
}
if (int128_gt(shft_res2, mask1)) {
- Vd->D(1) = int128_getlo(mask1);
+ Vd->D(idx * 2 + 1) = int128_getlo(mask1);
} else if (int128_lt(shft_res2, int128_neg(mask2))) {
- Vd->D(1) = int128_getlo(mask2);
+ Vd->D(idx * 2 + 1) = int128_getlo(mask2);
} else {
- Vd->D(1) = int128_getlo(shft_res2);
+ Vd->D(idx * 2 + 1) = int128_getlo(shft_res2);
+ }
+}
+
+void HELPER(vssrarni_d_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
+ Int128 mask1, mask2;
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ mask1 = int128_sub(int128_lshift(int128_one(), 63), int128_one());
+ mask2 = int128_lshift(int128_one(), 63);
+
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrarni_d_q(Vd, Vj, imm, i, mask1, mask2);
}
}
VSSRARNI(vssrarni_h_w, 32, H, W)
VSSRARNI(vssrarni_w_d, 64, W, D)
-#define VSSRLRNUI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssrlrnu_ ## E1(Vj->E2(i), imm, BIT/2); \
- temp.E1(i + LSX_LEN/BIT) = do_ssrlrnu_ ## E1(Vd->E2(i), imm, BIT/2); \
- } \
- *Vd = temp; \
+#define VSSRLRNUI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssrlrnu_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssrlrnu_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ } \
+ } \
+ *Vd = temp; \
+}
+
+void HELPER(vssrlrni_du_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
+ Int128 mask;
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ mask = int128_sub(int128_lshift(int128_one(), 64), int128_one());
+
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrlrni_q(Vd, Vj, imm, i, mask);
+ }
}
VSSRLRNUI(vssrlrni_bu_h, 16, B, H)
VSSRLRNUI(vssrlrni_hu_w, 32, H, W)
VSSRLRNUI(vssrlrni_wu_d, 64, W, D)
-VSSRLRNI_Q(vssrlrni_du_q, 64)
-#define VSSRARNUI(NAME, BIT, E1, E2) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E1(i) = do_ssrarnu_ ## E1(Vj->E2(i), imm, BIT/2); \
- temp.E1(i + LSX_LEN/BIT) = do_ssrarnu_ ## E1(Vd->E2(i), imm, BIT/2); \
- } \
- *Vd = temp; \
-}
-
-void HELPER(vssrarni_du_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
-{
- Int128 shft_res1, shft_res2, mask1, mask2, r1, r2;
- VReg *Vd = (VReg *)vd;
- VReg *Vj = (VReg *)vj;
+#define VSSRARNUI(NAME, BIT, E1, E2) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E1(j + ofs * 2 * i) = do_ssrarnu_ ## E1(Vj->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ temp.E1(j + ofs * (2 * i + 1)) = do_ssrarnu_ ## E1(Vd->E2(j + ofs * i), \
+ imm, BIT / 2); \
+ } \
+ } \
+ *Vd = temp; \
+}
+
+static void do_vssrarni_du_q(VReg *Vd, VReg *Vj,
+ uint64_t imm, int idx, Int128 mask1, Int128 mask2)
+{
+ Int128 shft_res1, shft_res2, r1, r2;
if (imm == 0) {
- shft_res1 = Vj->Q(0);
- shft_res2 = Vd->Q(0);
+ shft_res1 = Vj->Q(idx);
+ shft_res2 = Vd->Q(idx);
} else {
- r1 = int128_and(int128_rshift(Vj->Q(0), (imm -1)), int128_one());
- r2 = int128_and(int128_rshift(Vd->Q(0), (imm -1)), int128_one());
-
- shft_res1 = int128_add(int128_rshift(Vj->Q(0), imm), r1);
- shft_res2 = int128_add(int128_rshift(Vd->Q(0), imm), r2);
+ r1 = int128_and(int128_rshift(Vj->Q(idx), (imm - 1)), int128_one());
+ r2 = int128_and(int128_rshift(Vd->Q(idx), (imm - 1)), int128_one());
+ shft_res1 = int128_add(int128_rshift(Vj->Q(idx), imm), r1);
+ shft_res2 = int128_add(int128_rshift(Vd->Q(idx), imm), r2);
}
- if (int128_lt(Vj->Q(0), int128_zero())) {
+ if (int128_lt(Vj->Q(idx), int128_zero())) {
shft_res1 = int128_zero();
}
- if (int128_lt(Vd->Q(0), int128_zero())) {
+ if (int128_lt(Vd->Q(idx), int128_zero())) {
shft_res2 = int128_zero();
}
- mask1 = int128_sub(int128_lshift(int128_one(), 64), int128_one());
- mask2 = int128_lshift(int128_one(), 64);
-
if (int128_gt(shft_res1, mask1)) {
- Vd->D(0) = int128_getlo(mask1);
+ Vd->D(idx * 2) = int128_getlo(mask1);
} else if (int128_lt(shft_res1, int128_neg(mask2))) {
- Vd->D(0) = int128_getlo(mask2);
+ Vd->D(idx * 2) = int128_getlo(mask2);
} else {
- Vd->D(0) = int128_getlo(shft_res1);
+ Vd->D(idx * 2) = int128_getlo(shft_res1);
}
if (int128_gt(shft_res2, mask1)) {
- Vd->D(1) = int128_getlo(mask1);
+ Vd->D(idx * 2 + 1) = int128_getlo(mask1);
} else if (int128_lt(shft_res2, int128_neg(mask2))) {
- Vd->D(1) = int128_getlo(mask2);
+ Vd->D(idx * 2 + 1) = int128_getlo(mask2);
} else {
- Vd->D(1) = int128_getlo(shft_res2);
+ Vd->D(idx * 2 + 1) = int128_getlo(shft_res2);
+ }
+}
+
+void HELPER(vssrarni_du_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
+ Int128 mask1, mask2;
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ mask1 = int128_sub(int128_lshift(int128_one(), 64), int128_one());
+ mask2 = int128_lshift(int128_one(), 64);
+
+ for (i = 0; i < oprsz / 16; i++) {
+ do_vssrarni_du_q(Vd, Vj, imm, i, mask1, mask2);
}
}
int i; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) \
+ for (i = 0; i < oprsz / (BIT / 8); i++) \
{ \
Vd->E(i) = DO_OP(Vj->E(i)); \
} \
}
-#define DO_CLO_B(N) (clz32(~N & 0xff) - 24)
-#define DO_CLO_H(N) (clz32(~N & 0xffff) - 16)
-#define DO_CLO_W(N) (clz32(~N))
-#define DO_CLO_D(N) (clz64(~N))
-#define DO_CLZ_B(N) (clz32(N) - 24)
-#define DO_CLZ_H(N) (clz32(N) - 16)
-#define DO_CLZ_W(N) (clz32(N))
-#define DO_CLZ_D(N) (clz64(N))
-
DO_2OP(vclo_b, 8, UB, DO_CLO_B)
DO_2OP(vclo_h, 16, UH, DO_CLO_H)
DO_2OP(vclo_w, 32, UW, DO_CLO_W)
int i; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) \
+ for (i = 0; i < oprsz / (BIT / 8); i++) \
{ \
Vd->E(i) = FN(Vj->E(i)); \
} \
VPCNT(vpcnt_w, 32, UW, ctpop32)
VPCNT(vpcnt_d, 64, UD, ctpop64)
-#define DO_BITCLR(a, bit) (a & ~(1ull << bit))
-#define DO_BITSET(a, bit) (a | 1ull << bit)
-#define DO_BITREV(a, bit) (a ^ (1ull << bit))
-
-#define DO_BIT(NAME, BIT, E, DO_OP) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t v) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E(i) = DO_OP(Vj->E(i), Vk->E(i)%BIT); \
- } \
+#define DO_BIT(NAME, BIT, E, DO_OP) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
+ Vd->E(i) = DO_OP(Vj->E(i), Vk->E(i)%BIT); \
+ } \
}
DO_BIT(vbitclr_b, 8, UB, DO_BITCLR)
DO_BIT(vbitrev_w, 32, UW, DO_BITREV)
DO_BIT(vbitrev_d, 64, UD, DO_BITREV)
-#define DO_BITI(NAME, BIT, E, DO_OP) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t v) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E(i) = DO_OP(Vj->E(i), imm); \
- } \
+#define DO_BITI(NAME, BIT, E, DO_OP) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
+ Vd->E(i) = DO_OP(Vj->E(i), imm); \
+ } \
}
DO_BITI(vbitclri_b, 8, UB, DO_BITCLR)
#define VFRSTP(NAME, BIT, MASK, E) \
void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
{ \
- int i, m; \
+ int i, j, m, ofs; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- if (Vj->E(i) < 0) { \
- break; \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ m = Vk->E(i * ofs) & MASK; \
+ for (j = 0; j < ofs; j++) { \
+ if (Vj->E(j + ofs * i) < 0) { \
+ break; \
+ } \
} \
+ Vd->E(m + i * ofs) = j; \
} \
- m = Vk->E(0) & MASK; \
- Vd->E(m) = i; \
}
VFRSTP(vfrstp_b, 8, 0xf, B)
#define VFRSTPI(NAME, BIT, E) \
void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
{ \
- int i, m; \
+ int i, j, m, ofs; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- if (Vj->E(i) < 0) { \
- break; \
+ ofs = LSX_LEN / BIT; \
+ m = imm % ofs; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ if (Vj->E(j + ofs * i) < 0) { \
+ break; \
+ } \
} \
+ Vd->E(m + i * ofs) = j; \
} \
- m = imm % (LSX_LEN/BIT); \
- Vd->E(m) = i; \
}
VFRSTPI(vfrstpi_b, 8, B)
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
\
vec_clear_cause(env); \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = FN(Vj->E(i), Vk->E(i), &env->fp_status); \
vec_update_fcsr0(env, GETPC()); \
} \
VReg *Vj = (VReg *)vj; \
VReg *Vk = (VReg *)vk; \
VReg *Va = (VReg *)va; \
+ int oprsz = simd_oprsz(desc); \
\
vec_clear_cause(env); \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = FN(Vj->E(i), Vk->E(i), Va->E(i), flags, &env->fp_status); \
vec_update_fcsr0(env, GETPC()); \
} \
int i; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
vec_clear_cause(env); \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = FN(env, Vj->E(i)); \
} \
}
int i; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
Vd->E(i) = FN(env, Vj->E(i)); \
} \
}
void HELPER(vfcvtl_s_h)(void *vd, void *vj,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 32;
vec_clear_cause(env);
- for (i = 0; i < LSX_LEN/32; i++) {
- temp.UW(i) = float16_cvt_float32(Vj->UH(i), &env->fp_status);
+ for (i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.UW(j + ofs * i) =float16_cvt_float32(Vj->UH(j + ofs * 2 * i),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
void HELPER(vfcvtl_d_s)(void *vd, void *vj,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 64;
vec_clear_cause(env);
- for (i = 0; i < LSX_LEN/64; i++) {
- temp.UD(i) = float32_cvt_float64(Vj->UW(i), &env->fp_status);
+ for (i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.UD(j + ofs * i) = float32_cvt_float64(Vj->UW(j + ofs * 2 * i),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
void HELPER(vfcvth_s_h)(void *vd, void *vj,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 32;
vec_clear_cause(env);
- for (i = 0; i < LSX_LEN/32; i++) {
- temp.UW(i) = float16_cvt_float32(Vj->UH(i + 4), &env->fp_status);
+ for (i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.UW(j + ofs * i) = float16_cvt_float32(Vj->UH(j + ofs * (2 * i + 1)),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
void HELPER(vfcvth_d_s)(void *vd, void *vj,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 64;
vec_clear_cause(env);
- for (i = 0; i < LSX_LEN/64; i++) {
- temp.UD(i) = float32_cvt_float64(Vj->UW(i + 2), &env->fp_status);
+ for (i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.UD(j + ofs * i) = float32_cvt_float64(Vj->UW(j + ofs * (2 * i + 1)),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
void HELPER(vfcvt_h_s)(void *vd, void *vj, void *vk,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
VReg *Vk = (VReg *)vk;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 32;
vec_clear_cause(env);
- for(i = 0; i < LSX_LEN/32; i++) {
- temp.UH(i + 4) = float32_cvt_float16(Vj->UW(i), &env->fp_status);
- temp.UH(i) = float32_cvt_float16(Vk->UW(i), &env->fp_status);
+ for(i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.UH(j + ofs * (2 * i + 1)) = float32_cvt_float16(Vj->UW(j + ofs * i),
+ &env->fp_status);
+ temp.UH(j + ofs * 2 * i) = float32_cvt_float16(Vk->UW(j + ofs * i),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
void HELPER(vfcvt_s_d)(void *vd, void *vj, void *vk,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
VReg *Vk = (VReg *)vk;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 64;
vec_clear_cause(env);
- for(i = 0; i < LSX_LEN/64; i++) {
- temp.UW(i + 2) = float64_cvt_float32(Vj->UD(i), &env->fp_status);
- temp.UW(i) = float64_cvt_float32(Vk->UD(i), &env->fp_status);
+ for(i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.UW(j + ofs * (2 * i + 1)) = float64_cvt_float32(Vj->UD(j + ofs * i),
+ &env->fp_status);
+ temp.UW(j + ofs * 2 * i) = float64_cvt_float32(Vk->UD(j + ofs * i),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
vec_clear_cause(env);
- for (i = 0; i < 4; i++) {
+ for (i = 0; i < oprsz / 4; i++) {
Vd->W(i) = float32_round_to_int(Vj->UW(i), &env->fp_status);
vec_update_fcsr0(env, GETPC());
}
int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
vec_clear_cause(env);
- for (i = 0; i < 2; i++) {
+ for (i = 0; i < oprsz / 8; i++) {
Vd->D(i) = float64_round_to_int(Vj->UD(i), &env->fp_status);
vec_update_fcsr0(env, GETPC());
}
int i; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
vec_clear_cause(env); \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
FloatRoundMode old_mode = get_float_rounding_mode(&env->fp_status); \
set_float_rounding_mode(MODE, &env->fp_status); \
Vd->E(i) = float## BIT ## _round_to_int(Vj->E(i), &env->fp_status); \
FTINT(rz_w_d, float64, int32, uint64_t, uint32_t, float_round_to_zero)
FTINT(rne_w_d, float64, int32, uint64_t, uint32_t, float_round_nearest_even)
-#define FTINT_W_D(NAME, FN) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, \
- CPULoongArchState *env, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- vec_clear_cause(env); \
- for (i = 0; i < 2; i++) { \
- temp.W(i + 2) = FN(env, Vj->UD(i)); \
- temp.W(i) = FN(env, Vk->UD(i)); \
- } \
- *Vd = temp; \
+#define FTINT_W_D(NAME, FN) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, \
+ CPULoongArchState *env, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / 64; \
+ vec_clear_cause(env); \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.W(j + ofs * (2 * i + 1)) = FN(env, Vj->UD(j + ofs * i)); \
+ temp.W(j + ofs * 2 * i) = FN(env, Vk->UD(j + ofs * i)); \
+ } \
+ } \
+ *Vd = temp; \
}
FTINT_W_D(vftint_w_d, do_float64_to_int32)
FTINT(rzh_l_s, float32, int64, uint32_t, uint64_t, float_round_to_zero)
FTINT(rneh_l_s, float32, int64, uint32_t, uint64_t, float_round_nearest_even)
-#define FTINTL_L_S(NAME, FN) \
-void HELPER(NAME)(void *vd, void *vj, \
- CPULoongArchState *env, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- vec_clear_cause(env); \
- for (i = 0; i < 2; i++) { \
- temp.D(i) = FN(env, Vj->UW(i)); \
- } \
- *Vd = temp; \
+#define FTINTL_L_S(NAME, FN) \
+void HELPER(NAME)(void *vd, void *vj, \
+ CPULoongArchState *env, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / 64; \
+ vec_clear_cause(env); \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.D(j + ofs * i) = FN(env, Vj->UW(j + ofs * 2 * i)); \
+ } \
+ } \
+ *Vd = temp; \
}
FTINTL_L_S(vftintl_l_s, do_float32_to_int64)
FTINTL_L_S(vftintrzl_l_s, do_ftintrzl_l_s)
FTINTL_L_S(vftintrnel_l_s, do_ftintrnel_l_s)
-#define FTINTH_L_S(NAME, FN) \
-void HELPER(NAME)(void *vd, void *vj, \
- CPULoongArchState *env, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- vec_clear_cause(env); \
- for (i = 0; i < 2; i++) { \
- temp.D(i) = FN(env, Vj->UW(i + 2)); \
- } \
- *Vd = temp; \
+#define FTINTH_L_S(NAME, FN) \
+void HELPER(NAME)(void *vd, void *vj, \
+ CPULoongArchState *env, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / 64; \
+ vec_clear_cause(env); \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.D(j + ofs * i) = FN(env, Vj->UW(j + ofs * (2 * i + 1))); \
+ } \
+ } \
+ *Vd = temp; \
}
FTINTH_L_S(vftinth_l_s, do_float32_to_int64)
void HELPER(vffintl_d_w)(void *vd, void *vj,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 64;
vec_clear_cause(env);
- for (i = 0; i < 2; i++) {
- temp.D(i) = int32_to_float64(Vj->W(i), &env->fp_status);
+ for (i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.D(j + ofs * i) = int32_to_float64(Vj->W(j + ofs * 2 * i),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
void HELPER(vffinth_d_w)(void *vd, void *vj,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 64;
vec_clear_cause(env);
- for (i = 0; i < 2; i++) {
- temp.D(i) = int32_to_float64(Vj->W(i + 2), &env->fp_status);
+ for (i = 0; i < oprsz /16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.D(j + ofs * i) = int32_to_float64(Vj->W(j + ofs * (2 * i + 1)),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
void HELPER(vffint_s_l)(void *vd, void *vj, void *vk,
CPULoongArchState *env, uint32_t desc)
{
- int i;
- VReg temp;
+ int i, j, ofs;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
VReg *Vk = (VReg *)vk;
+ int oprsz = simd_oprsz(desc);
+ ofs = LSX_LEN / 64;
vec_clear_cause(env);
- for (i = 0; i < 2; i++) {
- temp.W(i + 2) = int64_to_float32(Vj->D(i), &env->fp_status);
- temp.W(i) = int64_to_float32(Vk->D(i), &env->fp_status);
+ for (i = 0; i < oprsz / 16; i++) {
+ for (j = 0; j < ofs; j++) {
+ temp.W(j + ofs * (2 * i + 1)) = int64_to_float32(Vj->D(j + ofs * i),
+ &env->fp_status);
+ temp.W(j + ofs * 2 * i) = int64_to_float32(Vk->D(j + ofs * i),
+ &env->fp_status);
+ }
vec_update_fcsr0(env, GETPC());
}
*Vd = temp;
}
-#define VSEQ(a, b) (a == b ? -1 : 0)
-#define VSLE(a, b) (a <= b ? -1 : 0)
-#define VSLT(a, b) (a < b ? -1 : 0)
-
-#define VCMPI(NAME, BIT, E, DO_OP) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t v) \
-{ \
- int i; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- typedef __typeof(Vd->E(0)) TD; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- Vd->E(i) = DO_OP(Vj->E(i), (TD)imm); \
- } \
+#define VCMPI(NAME, BIT, E, DO_OP) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ typedef __typeof(Vd->E(0)) TD; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
+ Vd->E(i) = DO_OP(Vj->E(i), (TD)imm); \
+ } \
}
VCMPI(vseqi_b, 8, B, VSEQ)
}
#define VFCMP(NAME, BIT, E, FN) \
-void HELPER(NAME)(CPULoongArchState *env, \
+void HELPER(NAME)(CPULoongArchState *env, uint32_t oprsz, \
uint32_t vd, uint32_t vj, uint32_t vk, uint32_t flags) \
{ \
int i; \
VReg *Vk = &(env->fpr[vk].vreg); \
\
vec_clear_cause(env); \
- for (i = 0; i < LSX_LEN/BIT ; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
FloatRelation cmp; \
cmp = FN(Vj->E(i), Vk->E(i), &env->fp_status); \
t.E(i) = vfcmp_common(env, cmp, flags); \
VFCMP(vfcmp_c_d, 64, UD, float64_compare_quiet)
VFCMP(vfcmp_s_d, 64, UD, float64_compare)
-void HELPER(vbitseli_b)(void *vd, void *vj, uint64_t imm, uint32_t v)
+void HELPER(vbitseli_b)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
int i;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
- for (i = 0; i < 16; i++) {
+ for (i = 0; i < simd_oprsz(desc); i++) {
Vd->B(i) = (~Vd->B(i) & Vj->B(i)) | (Vd->B(i) & imm);
}
}
/* Copy from target/arm/tcg/sve_helper.c */
static inline bool do_match2(uint64_t n, uint64_t m0, uint64_t m1, int esz)
{
- uint64_t bits = 8 << esz;
+ int bits = 8 << esz;
uint64_t ones = dup_const(esz, 1);
uint64_t signs = ones << (bits - 1);
uint64_t cmp0, cmp1;
return (cmp0 | cmp1) & signs;
}
-#define SETANYEQZ(NAME, MO) \
-void HELPER(NAME)(CPULoongArchState *env, uint32_t cd, uint32_t vj) \
-{ \
- VReg *Vj = &(env->fpr[vj].vreg); \
- \
- env->cf[cd & 0x7] = do_match2(0, Vj->D(0), Vj->D(1), MO); \
+#define SETANYEQZ(NAME, MO) \
+void HELPER(NAME)(CPULoongArchState *env, \
+ uint32_t oprsz, uint32_t cd, uint32_t vj) \
+{ \
+ VReg *Vj = &(env->fpr[vj].vreg); \
+ \
+ env->cf[cd & 0x7] = do_match2(0, Vj->D(0), Vj->D(1), MO); \
+ if (oprsz == 32) { \
+ env->cf[cd & 0x7] = env->cf[cd & 0x7] || \
+ do_match2(0, Vj->D(2), Vj->D(3), MO); \
+ } \
}
+
SETANYEQZ(vsetanyeqz_b, MO_8)
SETANYEQZ(vsetanyeqz_h, MO_16)
SETANYEQZ(vsetanyeqz_w, MO_32)
SETANYEQZ(vsetanyeqz_d, MO_64)
-#define SETALLNEZ(NAME, MO) \
-void HELPER(NAME)(CPULoongArchState *env, uint32_t cd, uint32_t vj) \
-{ \
- VReg *Vj = &(env->fpr[vj].vreg); \
- \
- env->cf[cd & 0x7]= !do_match2(0, Vj->D(0), Vj->D(1), MO); \
+#define SETALLNEZ(NAME, MO) \
+void HELPER(NAME)(CPULoongArchState *env, \
+ uint32_t oprsz, uint32_t cd, uint32_t vj) \
+{ \
+ VReg *Vj = &(env->fpr[vj].vreg); \
+ \
+ env->cf[cd & 0x7]= !do_match2(0, Vj->D(0), Vj->D(1), MO); \
+ if (oprsz == 32) { \
+ env->cf[cd & 0x7] = env->cf[cd & 0x7] && \
+ !do_match2(0, Vj->D(2), Vj->D(3), MO); \
+ } \
}
+
SETALLNEZ(vsetallnez_b, MO_8)
SETALLNEZ(vsetallnez_h, MO_16)
SETALLNEZ(vsetallnez_w, MO_32)
SETALLNEZ(vsetallnez_d, MO_64)
+#define XVINSVE0(NAME, E, MASK) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ Vd->E(imm & MASK) = Vj->E(0); \
+}
+
+XVINSVE0(xvinsve0_w, W, 0x7)
+XVINSVE0(xvinsve0_d, D, 0x3)
+
+#define XVPICKVE(NAME, E, BIT, MASK) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ Vd->E(0) = Vj->E(imm & MASK); \
+ for (i = 1; i < oprsz / (BIT / 8); i++) { \
+ Vd->E(i) = 0; \
+ } \
+}
+
+XVPICKVE(xvpickve_w, W, 32, 0x7)
+XVPICKVE(xvpickve_d, D, 64, 0x3)
+
#define VPACKEV(NAME, BIT, E) \
void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
{ \
int i; \
- VReg temp; \
+ VReg temp = {}; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
temp.E(2 * i + 1) = Vj->E(2 * i); \
temp.E(2 *i) = Vk->E(2 * i); \
} \
void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
{ \
int i; \
- VReg temp; \
+ VReg temp = {}; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
\
- for (i = 0; i < LSX_LEN/BIT; i++) { \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
temp.E(2 * i + 1) = Vj->E(2 * i + 1); \
temp.E(2 * i) = Vk->E(2 * i + 1); \
} \
VPACKOD(vpackod_w, 64, W)
VPACKOD(vpackod_d, 128, D)
-#define VPICKEV(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E(i + LSX_LEN/BIT) = Vj->E(2 * i); \
- temp.E(i) = Vk->E(2 * i); \
- } \
- *Vd = temp; \
+#define VPICKEV(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E(j + ofs * (2 * i + 1)) = Vj->E(2 * (j + ofs * i)); \
+ temp.E(j + ofs * 2 * i) = Vk->E(2 * (j + ofs * i)); \
+ } \
+ } \
+ *Vd = temp; \
}
VPICKEV(vpickev_b, 16, B)
VPICKEV(vpickev_w, 64, W)
VPICKEV(vpickev_d, 128, D)
-#define VPICKOD(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E(i + LSX_LEN/BIT) = Vj->E(2 * i + 1); \
- temp.E(i) = Vk->E(2 * i + 1); \
- } \
- *Vd = temp; \
+#define VPICKOD(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E(j + ofs * (2 * i + 1)) = Vj->E(2 * (j + ofs * i) + 1); \
+ temp.E(j + ofs * 2 * i) = Vk->E(2 * (j + ofs * i) + 1); \
+ } \
+ } \
+ *Vd = temp; \
}
VPICKOD(vpickod_b, 16, B)
VPICKOD(vpickod_w, 64, W)
VPICKOD(vpickod_d, 128, D)
-#define VILVL(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E(2 * i + 1) = Vj->E(i); \
- temp.E(2 * i) = Vk->E(i); \
- } \
- *Vd = temp; \
+#define VILVL(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E(2 * (j + ofs * i) + 1) = Vj->E(j + ofs * 2 * i); \
+ temp.E(2 * (j + ofs * i)) = Vk->E(j + ofs * 2 * i); \
+ } \
+ } \
+ *Vd = temp; \
}
VILVL(vilvl_b, 16, B)
VILVL(vilvl_w, 64, W)
VILVL(vilvl_d, 128, D)
-#define VILVH(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E(2 * i + 1) = Vj->E(i + LSX_LEN/BIT); \
- temp.E(2 * i) = Vk->E(i + LSX_LEN/BIT); \
- } \
- *Vd = temp; \
+#define VILVH(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, ofs; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ ofs = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / 16; i++) { \
+ for (j = 0; j < ofs; j++) { \
+ temp.E(2 * (j + ofs * i) + 1) = Vj->E(j + ofs * (2 * i + 1)); \
+ temp.E(2 * (j + ofs * i)) = Vk->E(j + ofs * (2 * i + 1)); \
+ } \
+ } \
+ *Vd = temp; \
}
VILVH(vilvh_b, 16, B)
void HELPER(vshuf_b)(void *vd, void *vj, void *vk, void *va, uint32_t desc)
{
- int i, m;
- VReg temp;
+ int i, j, m;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
VReg *Vk = (VReg *)vk;
VReg *Va = (VReg *)va;
+ int oprsz = simd_oprsz(desc);
- m = LSX_LEN/8;
- for (i = 0; i < m ; i++) {
+ m = LSX_LEN / 8;
+ for (i = 0; i < (oprsz / 16) * m; i++) {
+ j = i < m ? 0 : 1;
uint64_t k = (uint8_t)Va->B(i) % (2 * m);
- temp.B(i) = k < m ? Vk->B(k) : Vj->B(k - m);
+ temp.B(i) = k < m ? Vk->B(k + j * m): Vj->B(k + (j - 1) * m);
}
*Vd = temp;
}
-#define VSHUF(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
-{ \
- int i, m; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- VReg *Vk = (VReg *)vk; \
- \
- m = LSX_LEN/BIT; \
- for (i = 0; i < m; i++) { \
- uint64_t k = ((uint8_t) Vd->E(i)) % (2 * m); \
- temp.E(i) = k < m ? Vk->E(k) : Vj->E(k - m); \
- } \
- *Vd = temp; \
+#define VSHUF(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, void *vk, uint32_t desc) \
+{ \
+ int i, j, m; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ VReg *Vk = (VReg *)vk; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ m = LSX_LEN / BIT; \
+ for (i = 0; i < (oprsz / 16) * m; i++) { \
+ j = i < m ? 0 : 1; \
+ uint64_t k = ((uint8_t)Vd->E(i)) % (2 * m); \
+ temp.E(i) = k < m ? Vk->E(k + j * m) : Vj->E(k + (j - 1) * m); \
+ } \
+ *Vd = temp; \
}
VSHUF(vshuf_h, 16, H)
VSHUF(vshuf_w, 32, W)
VSHUF(vshuf_d, 64, D)
-#define VSHUF4I(NAME, BIT, E) \
-void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
-{ \
- int i; \
- VReg temp; \
- VReg *Vd = (VReg *)vd; \
- VReg *Vj = (VReg *)vj; \
- \
- for (i = 0; i < LSX_LEN/BIT; i++) { \
- temp.E(i) = Vj->E(((i) & 0xfc) + (((imm) >> \
- (2 * ((i) & 0x03))) & 0x03)); \
- } \
- *Vd = temp; \
+#define VSHUF4I(NAME, BIT, E) \
+void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
+{ \
+ int i, j, max; \
+ VReg temp = {}; \
+ VReg *Vd = (VReg *)vd; \
+ VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
+ \
+ max = LSX_LEN / BIT; \
+ for (i = 0; i < oprsz / (BIT / 8); i++) { \
+ j = i < max ? 1 : 2; \
+ temp.E(i) = Vj->E(SHF_POS(i - ((j -1)* max), imm) + (j - 1) * max); \
+ } \
+ *Vd = temp; \
}
VSHUF4I(vshuf4i_b, 8, B)
void HELPER(vshuf4i_d)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
+ int i;
+ VReg temp = {};
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
- VReg temp;
- temp.D(0) = (imm & 2 ? Vj : Vd)->D(imm & 1);
- temp.D(1) = (imm & 8 ? Vj : Vd)->D((imm >> 2) & 1);
+ for (i = 0; i < oprsz / 16; i++) {
+ temp.D(2 * i) = (imm & 2 ? Vj : Vd)->D((imm & 1) + 2 * i);
+ temp.D(2 * i + 1) = (imm & 8 ? Vj : Vd)->D(((imm >> 2) & 1) + 2 * i);
+ }
+ *Vd = temp;
+}
+
+void HELPER(vperm_w)(void *vd, void *vj, void *vk, uint32_t desc)
+{
+ int i, m;
+ VReg temp = {};
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ VReg *Vk = (VReg *)vk;
+
+ m = LASX_LEN / 32;
+ for (i = 0; i < m ; i++) {
+ uint64_t k = (uint8_t)Vk->W(i) % 8;
+ temp.W(i) = Vj->W(k);
+ }
*Vd = temp;
}
void HELPER(vpermi_w)(void *vd, void *vj, uint64_t imm, uint32_t desc)
{
+ int i;
+ VReg temp = {};
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+ int oprsz = simd_oprsz(desc);
+
+ for (i = 0; i < oprsz / 16; i++) {
+ temp.W(4 * i) = Vj->W((imm & 0x3) + 4 * i);
+ temp.W(4 * i + 1) = Vj->W(((imm >> 2) & 0x3) + 4 * i);
+ temp.W(4 * i + 2) = Vd->W(((imm >> 4) & 0x3) + 4 * i);
+ temp.W(4 * i + 3) = Vd->W(((imm >> 6) & 0x3) + 4 * i);
+ }
+ *Vd = temp;
+}
+
+void HELPER(vpermi_d)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ VReg temp = {};
+ VReg *Vd = (VReg *)vd;
+ VReg *Vj = (VReg *)vj;
+
+ temp.D(0) = Vj->D(imm & 0x3);
+ temp.D(1) = Vj->D((imm >> 2) & 0x3);
+ temp.D(2) = Vj->D((imm >> 4) & 0x3);
+ temp.D(3) = Vj->D((imm >> 6) & 0x3);
+ *Vd = temp;
+}
+
+void HELPER(vpermi_q)(void *vd, void *vj, uint64_t imm, uint32_t desc)
+{
+ int i;
VReg temp;
VReg *Vd = (VReg *)vd;
VReg *Vj = (VReg *)vj;
- temp.W(0) = Vj->W(imm & 0x3);
- temp.W(1) = Vj->W((imm >> 2) & 0x3);
- temp.W(2) = Vd->W((imm >> 4) & 0x3);
- temp.W(3) = Vd->W((imm >> 6) & 0x3);
+ for (i = 0; i < 2; i++, imm >>= 4) {
+ temp.Q(i) = (imm & 2 ? Vd: Vj)->Q(imm & 1);
+ }
*Vd = temp;
}
#define VEXTRINS(NAME, BIT, E, MASK) \
void HELPER(NAME)(void *vd, void *vj, uint64_t imm, uint32_t desc) \
{ \
- int ins, extr; \
+ int i, ins, extr, max; \
VReg *Vd = (VReg *)vd; \
VReg *Vj = (VReg *)vj; \
+ int oprsz = simd_oprsz(desc); \
\
+ max = LSX_LEN / BIT; \
ins = (imm >> 4) & MASK; \
extr = imm & MASK; \
- Vd->E(ins) = Vj->E(extr); \
+ for (i = 0; i < oprsz / 16; i++) { \
+ Vd->E(ins + i * max) = Vj->E(extr + i * max); \
+ } \
}
VEXTRINS(vextrins_b, 8, B, 0xf)
projects / mirror_qemu.git / blobdiff