aboutsummaryrefslogtreecommitdiff
path: root/tcg
diff options
context:
space:
mode:
authorRichard Henderson <richard.henderson@linaro.org>2019-05-14 12:47:02 -0700
committerRichard Henderson <richard.henderson@linaro.org>2019-05-22 15:09:43 -0400
commit02f3a5b4744885258758d07ebe09cf965de78bcf (patch)
tree4626d771631f230e9fcafff6f8e200d246ff64a3 /tcg
parent7e308e003e5b6ddd3130e09711e1d33693230696 (diff)
tcg/aarch64: Build vector immediates with two insns
Use MOVI+ORR or MVNI+BIC in order to build some vector constants, as opposed to dropping them to the constant pool. This includes all 16-bit constants and a similar set of 32-bit constants. Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Diffstat (limited to 'tcg')
-rw-r--r--tcg/aarch64/tcg-target.inc.c47
1 files changed, 47 insertions, 0 deletions
diff --git a/tcg/aarch64/tcg-target.inc.c b/tcg/aarch64/tcg-target.inc.c
index 0b8b733805..52c18074ae 100644
--- a/tcg/aarch64/tcg-target.inc.c
+++ b/tcg/aarch64/tcg-target.inc.c
@@ -273,6 +273,26 @@ static bool is_fimm64(uint64_t v64, int *cmode, int *imm8)
return false;
}
+/*
+ * Return non-zero if v32 can be formed by MOVI+ORR.
+ * Place the parameters for MOVI in (cmode, imm8).
+ * Return the cmode for ORR; the imm8 can be had via extraction from v32.
+ */
+static int is_shimm32_pair(uint32_t v32, int *cmode, int *imm8)
+{
+ int i;
+
+ for (i = 6; i > 0; i -= 2) {
+ /* Mask out one byte we can add with ORR. */
+ uint32_t tmp = v32 & ~(0xffu << (i * 4));
+ if (is_shimm32(tmp, cmode, imm8) ||
+ is_soimm32(tmp, cmode, imm8)) {
+ break;
+ }
+ }
+ return i;
+}
+
static int tcg_target_const_match(tcg_target_long val, TCGType type,
const TCGArgConstraint *arg_ct)
{
@@ -495,6 +515,8 @@ typedef enum {
/* AdvSIMD modified immediate */
I3606_MOVI = 0x0f000400,
I3606_MVNI = 0x2f000400,
+ I3606_BIC = 0x2f001400,
+ I3606_ORR = 0x0f001400,
/* AdvSIMD shift by immediate */
I3614_SSHR = 0x0f000400,
@@ -843,6 +865,14 @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type,
tcg_out_insn(s, 3606, MVNI, q, rd, 0, cmode, imm8);
return;
}
+
+ /*
+ * Otherwise, all remaining constants can be loaded in two insns:
+ * rd = v16 & 0xff, rd |= v16 & 0xff00.
+ */
+ tcg_out_insn(s, 3606, MOVI, q, rd, 0, 0x8, v16 & 0xff);
+ tcg_out_insn(s, 3606, ORR, q, rd, 0, 0xa, v16 >> 8);
+ return;
} else if (v64 == dup_const(MO_32, v64)) {
uint32_t v32 = v64;
uint32_t n32 = ~v32;
@@ -858,6 +888,23 @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type,
tcg_out_insn(s, 3606, MVNI, q, rd, 0, cmode, imm8);
return;
}
+
+ /*
+ * Restrict the set of constants to those we can load with
+ * two instructions. Others we load from the pool.
+ */
+ i = is_shimm32_pair(v32, &cmode, &imm8);
+ if (i) {
+ tcg_out_insn(s, 3606, MOVI, q, rd, 0, cmode, imm8);
+ tcg_out_insn(s, 3606, ORR, q, rd, 0, i, extract32(v32, i * 4, 8));
+ return;
+ }
+ i = is_shimm32_pair(n32, &cmode, &imm8);
+ if (i) {
+ tcg_out_insn(s, 3606, MVNI, q, rd, 0, cmode, imm8);
+ tcg_out_insn(s, 3606, BIC, q, rd, 0, i, extract32(n32, i * 4, 8));
+ return;
+ }
} else if (is_fimm64(v64, &cmode, &imm8)) {
tcg_out_insn(s, 3606, MOVI, q, rd, 1, cmode, imm8);
return;