opl: enable EMU8950 TLL via pre-computed table in flash (.flashdata section)

The TLL (Total Level Lookup) table is 128KB — too large for the RP2040's
256KB RAM when the firmware also runs code from RAM. Previous attempts to
use a `const` array failed because the firmware uses a copy_to_ram binary
type, which copies ALL sections including `.rodata` into RAM at boot.

The fix: place the pre-computed table in the `.flashdata.tll_table` linker
section. Both pico-sdk's `memmap_copy_to_ram.ld` and the local
`fw_memmap_c2ram.ld` explicitly capture `.flashdata*` into the FLASH
`.rodata` region, which is accessed via XIP cache and never copied to RAM.

- gen_tll_table.py: new script that pre-computes makeTllTable() using the
  EMU8950_NO_FLOAT integer path, outputting tll_table_flash.c with the
  `.flashdata.tll_table` section attribute
- tll_table_flash.c: generated 128KB table (32,768 uint32_t values)
- emu8950.c: support EMU8950_TLL_FLASH — use extern const declaration and
  skip makeTllTable() at runtime when table is pre-computed in flash
- CMakeLists: replace EMU8950_NO_TLL=1 with EMU8950_TLL_FLASH=1

TLL improves OPL2 accuracy by enabling the full Key Scale Level (KSL)
attenuation lookup table rather than the inline approximation.

sw/CMakeLists.txt

@@ -170,8 +170,8 @@ function(config_target TARGET_NAME MULTIFW)
             SOUND_OPL=1
             USE_EMU8950_OPL=1        
             EMU8950_ASM=1
-            EMU8950_NO_TLL=1 # don't use lookup table for total level
-            EMU8950_NO_FLOAT=1 # double check there is no float
+            EMU8950_TLL_FLASH=1 # pre-computed TLL table stored in flash via .flashdata section (not copied to RAM)
+            EMU8950_NO_FLOAT=1 # RP2040 has no FPU; float only used in TLL init anyway
             EMU8950_NO_TIMER=1 # disable timer which isn't used
             EMU8950_NO_TEST_FLAG=1 # disable test flags (which aren't used)
         )

sw/opl/CMakeLists.txt

@@ -2,6 +2,7 @@ add_library(opl INTERFACE)
 target_sources(opl INTERFACE
         # ${CMAKE_CURRENT_LIST_DIR}/opl_api.c
         ${CMAKE_CURRENT_LIST_DIR}/emu8950.c
+        ${CMAKE_CURRENT_LIST_DIR}/tll_table_flash.c
         ${CMAKE_CURRENT_LIST_DIR}/slot_render.cpp
         ${CMAKE_CURRENT_LIST_DIR}/opl_pico.c)
 target_compile_options(opl INTERFACE -fms-extensions) # want OPL_SLOT_RENDER to be unnamed within OPL_SLOT

sw/opl/emu8950.c

@@ -200,8 +200,12 @@ static int16_t kl_tablex16[16] = {dB2x16(0.000), dB2x16(9.000), dB2x16(12.000),
 #endif
 
 #if !EMU8950_NO_TLL
+#ifdef EMU8950_TLL_FLASH
+extern const uint32_t tll_table[8 * 16][1 << TL_BITS][4];
+#else
 static uint32_t tll_table[8 * 16][1 << TL_BITS][4];
 #endif
+#endif
 static int32_t rks_table[2][32][2];
 
 #define min(i, j) (((i) < (j)) ? (i) : (j))
@@ -368,7 +372,7 @@ static void makeSinTable(void) {
 }
 
 static void makeTllTable(void) {
-#if !EMU8950_NO_TLL
+#if !EMU8950_NO_TLL && !defined(EMU8950_TLL_FLASH)
     int32_t tmp;
     int32_t fnum, block, TL, KL, kx;
 

sw/opl/gen_tll_table.py

@@ -0,0 +1,71 @@
+#!/usr/bin/env python3
+"""
+Generates tll_table_flash.c for the EMU8950 OPL2 emulator.
+Pre-computes makeTllTable() so the result can be stored in RP2040 flash (XIP)
+instead of RAM, working around the 128KB RAM constraint.
+
+Replicates the EMU8950_NO_FLOAT integer path from emu8950.c:
+  kl_tablex16[fnum] = (uint16_t)(kl_values[fnum] * 32)
+  dB2x16(3.000)     = 96
+  EG_STEPx16        = 3
+  TL2EG(tl)         = tl << 2
+"""
+
+TL_BITS = 6
+EG_STEPx16 = 3
+
+kl_values = [
+    0.000,  9.000, 12.000, 13.875, 15.000, 16.125,
+   16.875, 17.625, 18.000, 18.750, 19.125, 19.500,
+   19.875, 20.250, 20.625, 21.000,
+]
+kl_tablex16 = [int(v * 32) & 0xFFFF for v in kl_values]  # uint16_t cast
+dB2x16_3 = int(3.000 * 32)  # = 96
+
+def TL2EG(tl):
+    return tl << 2
+
+# Build table[128][64][4]
+table = [[[0] * 4 for _ in range(1 << TL_BITS)] for _ in range(8 * 16)]
+
+for fnum in range(16):
+    for block in range(8):
+        idx = (block << 4) | fnum
+        for TL in range(1 << TL_BITS):
+            for KL in range(4):
+                if KL == 0:
+                    table[idx][TL][KL] = TL2EG(TL)
+                else:
+                    kx = ((KL & 1) << 1) | ((KL >> 1) & 1)
+                    tmp = kl_tablex16[fnum] - dB2x16_3 * (7 - block)
+                    if tmp <= 0:
+                        table[idx][TL][KL] = TL2EG(TL)
+                    else:
+                        table[idx][TL][KL] = (tmp >> (3 - kx)) // EG_STEPx16 + TL2EG(TL)
+
+lines = []
+lines.append("/* Auto-generated by gen_tll_table.py — do not edit manually. */")
+lines.append("/* Pre-computed EMU8950 TLL (Total Level Lookup) table for RP2040 flash (XIP). */")
+lines.append("/* Replicates makeTllTable() integer path; stored as const in .flashdata section so")
+lines.append(" * pico-sdk copy_to_ram linker scripts keep it in flash rather than copying to RAM. */")
+lines.append("#include <stdint.h>")
+lines.append("")
+lines.append("/* __attribute__((section(\".flashdata.tll_table\"))) keeps this in flash even in")
+lines.append(" * copy_to_ram builds (both memmap_copy_to_ram.ld and fw_memmap_c2ram.ld capture")
+lines.append(" * .flashdata* into the FLASH .rodata region, not the RAM .data region). */")
+lines.append("const uint32_t __attribute__((section(\".flashdata.tll_table\")))")
+lines.append("    tll_table[128][64][4] = {")
+for i in range(128):
+    lines.append("    { /* blk_fnum=" + str(i) + " */")
+    for j in range(64):
+        row = ", ".join(str(table[i][j][k]) for k in range(4))
+        lines.append("        {" + row + "},")
+    lines.append("    },")
+lines.append("};")
+
+import os
+out_path = os.path.join(os.path.dirname(__file__), "tll_table_flash.c")
+with open(out_path, "w") as f:
+    f.write("\n".join(lines) + "\n")
+
+print(f"Written {len(table) * 64 * 4} uint32_t values to {out_path}")