add ac_yield_load and ac_yield_tail for delay slot optimization opportunities.

This commit is contained in:
ed
2026-08-04 17:25:02 -04:00
parent 888ffce859
commit b5953a723b
7 changed files with 185 additions and 19 deletions
+21 -9
View File
@@ -400,7 +400,11 @@ atom_label(disconnected) /* === Disconnected body. */
load_upper_i(R_T4, 0x8080), or_i_self(R_T4, 0x8080),
store_word( R_T4, R_PadState, O_(PadState,left_x)),
store_byte( R_RawId, R_PadState, O_(PadState,id)),
jump_rel(atom_offset(disconnected, snap_end)), nop,
jump_rel(atom_offset(disconnected, snap_end)),
/* BD-slot: load next atom's entry point (replaces the nop).
* The unconditional branch always jumps to snap_end, where mac_yield_tail()
* transfers control to R_AtomJmp without re-loading it. */
mac_yield_load(),
atom_label(skip_disconnected)
/* === Case 2: Pending (status == 0 && id == 0)
@@ -418,7 +422,8 @@ atom_label(pending) /* === Pending body */
load_upper_i(R_T4, 0x8080), or_i_self(R_T4, 0x8080),
store_word( R_T4, R_PadState, O_(PadState,left_x)),
store_byte( R_RawId, R_PadState, O_(PadState,id)),
jump_rel(atom_offset(pending, snap_end)), nop,
jump_rel(atom_offset(pending, snap_end)),
mac_yield_load(),
atom_label(id_dispatch) /* === Case 3-6: ID dispatch */
add_ui(R_T4, R_0, 0x41), branch_ne(R_RawId, R_T4, atom_offset(id_dispatch, try_analog_stick)),
@@ -440,7 +445,8 @@ atom_label(id_dispatch) /* === Case 3-6: ID dispatch */
add_ui( R_T4, R_0, 0x41),
store_byte( R_T4, R_PadState, O_(PadState,id)),
jump_rel(atom_offset(id_dispatch, snap_end)), nop,
jump_rel(atom_offset(id_dispatch, snap_end)),
mac_yield_load(),
atom_label(try_analog_stick) /* === Case 4: AnalogStick (id == 0x53)*/
add_ui(R_T4, R_0, 0x53), branch_ne(R_RawId, R_T4, atom_offset(try_analog_stick, try_analog_pad)),
@@ -461,7 +467,8 @@ atom_label(analog_stick) /* === AnalogStick body
store_half( R_T4, R_PadState, O_(PadState,right_x)),
add_ui( R_T5, R_0, 0x53), /* R_T5 = id value (clobbers left_xy, already stored) */
store_byte( R_T5, R_PadState, O_(PadState,id)),
jump_rel(atom_offset(analog_stick, snap_end)), nop,
jump_rel(atom_offset(analog_stick, snap_end)),
mac_yield_load(),
atom_label(try_analog_pad) /* === Case 5-6: AnalogPad (id & 0xF0 == 0x70) */
and_i( R_T4, R_RawId, 0xF0),
@@ -483,7 +490,8 @@ atom_label(analog_pad) /* === AnalogPad body
store_half( R_T4, R_PadState, O_(PadState,right_x)),
store_byte( R_RawId, R_PadState, O_(PadState,id)),
jump_rel(atom_offset(analog_pad, snap_end)), nop,
jump_rel(atom_offset(analog_pad, snap_end)),
mac_yield_load(),
atom_label(try_unsupported) /* === Case 7: Unsupported — fall through from the AnalogPad range-check miss. */
add_ui( R_T4, R_0, PadStatus_Unsupported),
@@ -497,7 +505,8 @@ atom_label(try_unsupported) /* === Case 7: Unsupported — fall through from the
/* Fall through to snap_end. */
atom_label(snap_end)
mac_yield(),
/* NOT mac_yield() — R_AtomJmp was already loaded in the BD-slot of the case-exit branch. */
mac_yield_tail(),
};
/* ----- pad_apply_input -----
@@ -581,7 +590,8 @@ atom_label(dead_check_upper)
/* R_T4 = (0x90 < left_x) ? 1 : 0 → (left_x > 0x90) ? 1 : 0 */
set_lt_u(R_T4, R_T4, R_T3), branch_ne(R_T4, R_0, atom_offset(dead_zone_high_check, dead_high_active)),
add_ui( R_T4, R_0, 0x80), /* BD-slot: pre-load 0x80 for dead_high_active */
jump_rel(atom_offset(dead_zone_skip, exit_stick)), nop,
jump_rel(atom_offset(dead_zone_skip, exit_stick)),
mac_yield_load(),
atom_label(dead_low_active)
/* R_T3 = left_x (from line 632 lbu; not clobbered between dead_zone_low_check branch + its BD-slot `add_ui R_T4, 0x80`).
@@ -603,7 +613,8 @@ atom_label(dead_low_active)
add_u( R_T0, R_T0, R_T4),
store_half( R_T0, R_FloorRot, O_(V3_S2,y)),
jump_rel(atom_offset(end_low, exit_stick)), nop,
jump_rel(atom_offset(end_low, exit_stick)),
mac_yield_load(),
atom_label(dead_high_active)
/* R_T3 = left_x (from line 641 lbu in dead_check_upper; not clobbered between dead_zone_high_check branch + its BD-slot `add_ui R_T4, 0x80`).
@@ -625,7 +636,8 @@ atom_label(dead_high_active)
store_half( R_T0, R_FloorRot, O_(V3_S2,y)),
atom_label(exit_stick)
mac_yield(),
/* NOT mac_yield() — R_AtomJmp was already loaded in the BD-slot of the dead-zone/exit branch. */
mac_yield_tail(),
};
#pragma endregion Baked Atoms