[Qemu-devel] [PATCH v3 20/29] tcg-aarch64: Improve tcg_out_movi

[Richard Henderson <rth@twiddle.net>]

Handle small positive and negative numbers early.  Check for logical
immediates.  Check if using MOVN for the first set helps.

Signed-off-by: Richard Henderson <rth@twiddle.net>
---
 tcg/aarch64/tcg-target.c | 61 +++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 53 insertions(+), 8 deletions(-)

diff --git a/tcg/aarch64/tcg-target.c b/tcg/aarch64/tcg-target.c
index 59e5026..54f5ce8 100644
--- a/tcg/aarch64/tcg-target.c
+++ b/tcg/aarch64/tcg-target.c
@@ -534,20 +534,65 @@ static inline void tcg_out_movr_sp(TCGContext *s, bool ext,
 }
 
 static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg rd,
-                         tcg_target_long value)
+                         tcg_target_long svalue)
 {
-    AArch64Insn insn = INSN_MOVZ;
+    tcg_target_ulong value = svalue;
+    AArch64Insn insn;
+    bool ext = true;
+    int wantinv, shift;
 
     if (type == TCG_TYPE_I32) {
-        value = (uint32_t)value;
+        value = (uint32_t)svalue;
+        svalue = (int32_t)svalue;
+        ext = false;
+    } else if ((value & ~0xffffffffull) == 0) {
+        svalue = (int32_t)svalue;
+        ext = false;
+    }
+
+    /* Check small positive values.  */
+    if ((value & ~0xffffull) == 0) {
+        tcg_fmt_Rd_uimm_s(s, INSN_MOVZ, 0, rd, value, 0);
+        return;
+    }
+    /* Check small negative values.  */
+    if ((~svalue & 0xffffull) == 0) {
+        tcg_fmt_Rd_uimm_s(s, INSN_MOVN, ext, rd, ~svalue, 0);
+        return;
     }
+    /* Check for bitfield immediates.  */
+    if (is_limm(svalue)) {
+        tcg_fmt_Rdn_limm(s, INSN_ORRI, ext, rd, TCG_REG_XZR, svalue);
+        return;
+    }
+
+    /* Would it take fewer insns to begin with MOVN?  */
+    insn = INSN_MOVZ;
+    wantinv = 0;
+    wantinv += (value & 0x000000000000ffffull) == 0;
+    wantinv += (value & 0x00000000ffff0000ull) == 0;
+    wantinv += (value & 0x0000ffff00000000ull) == 0;
+    wantinv += (value & 0xffff000000000000ull) == 0;
+    wantinv = -wantinv;
+    wantinv += (~value & 0x000000000000ffffull) == 0;
+    wantinv += (~value & 0x00000000ffff0000ull) == 0;
+    wantinv += (~value & 0x0000ffff00000000ull) == 0;
+    wantinv += (~value & 0xffff000000000000ull) == 0;
+    if (wantinv > 0) {
+        insn = INSN_MOVN;
+        value = ~value;
+    }
+
+    shift = ctz64(value) & (63 & -16);
+    tcg_fmt_Rd_uimm_s(s, insn, ext, rd, value >> shift, shift);
+    value &= ~(0xffffUL << shift);
 
-    do {
-        unsigned shift = ctz64(value) & (63 & -16);
-        tcg_fmt_Rd_uimm_s(s, insn, shift >= 32, rd, value >> shift, shift);
+    while (value) {
+        shift = ctz64(value) & (63 & -16);
+        tcg_fmt_Rd_uimm_s(s, INSN_MOVK, ext, rd,
+                          (value ^ -(wantinv > 0)) >> shift, shift);
         value &= ~(0xffffUL << shift);
-        insn = INSN_MOVK;
-    } while (value);
+    }
 }
 
 static inline void tcg_out_ldst_r(TCGContext *s,
-- 
1.8.3.1