target/loongarch: Implement xvshuf xvperm{i} xvshuf4i
This patch includes:
- XVSHUF.{B/H/W/D};
- XVPERM.W;
- XVSHUF4i.{B/H/W/D};
- XVPERMI.{W/D/Q};
- XVEXTRINS.{B/H/W/D}.
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914022645.1151356-55-gaosong@loongson.cn>
diff --git a/target/loongarch/vec_helper.c b/target/loongarch/vec_helper.c
index 2bbaee6..6b61a5c 100644
--- a/target/loongarch/vec_helper.c
+++ b/target/loongarch/vec_helper.c
@@ -3381,57 +3381,65 @@
VILVH(vilvh_w, 64, W)
VILVH(vilvh_d, 128, D)
+#define SHF_POS(i, imm) (((i) & 0xfc) + (((imm) >> (2 * ((i) & 0x03))) & 0x03))
+
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)
@@ -3440,38 +3448,92 @@
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)
