- compile with size optimizations, reduces inst cache pressure

- removed most of lodepng's optional features
- lodepng now uses the already existing CRC32 code instead of using its own copy
- fixed GIC interrupt priority

Makefile

@@ -30,7 +30,7 @@ export INCLUDE := -I"$(shell pwd)/common"
 
 export ASFLAGS := -g -x assembler-with-cpp $(INCLUDE)
 export CFLAGS  := -DDBUILTS="\"$(DBUILTS)\"" -DDBUILTL="\"$(DBUILTL)\"" -DVERSION="\"$(VERSION)\"" -DFLAVOR="\"$(FLAVOR)\"" \
-                  -g -O2 -Wall -Wextra -Wpedantic -Wcast-align -Wformat=2 -Wno-main \
+                  -g -Os -Wall -Wextra -Wpedantic -Wcast-align -Wformat=2 -Wno-main \
                   -fomit-frame-pointer -ffast-math -std=gnu11 -MMD -MP \
                   -Wno-unused-function -Wno-format-truncation $(INCLUDE) -ffunction-sections -fdata-sections
 export LDFLAGS := -Tlink.ld -nostartfiles -Wl,--gc-sections,-z,max-page-size=512

arm11/source/main.c

@@ -151,12 +151,13 @@ void PXI_RX_Handler(u32 __attribute__((unused)) irqn)
 void __attribute__((noreturn)) MainLoop(void)
 {
 	// enable PXI RX interrupt
-	GIC_Enable(PXI_RX_INTERRUPT, BIT(0), GIC_HIGHEST_PRIO, PXI_RX_Handler);
+	GIC_Enable(PXI_RX_INTERRUPT, BIT(0), GIC_HIGHEST_PRIO + 2, PXI_RX_Handler);
 
 	// enable MCU interrupts
 	GIC_Enable(MCU_INTERRUPT, BIT(0), GIC_HIGHEST_PRIO + 1, MCU_HandleInterrupts);
 
-	GIC_Enable(VBLANK_INTERRUPT, BIT(0), GIC_HIGHEST_PRIO + 2, VBlank_Handler);
+	// set up VBlank interrupt to always have the highest priority
+	GIC_Enable(VBLANK_INTERRUPT, BIT(0), GIC_HIGHEST_PRIO, VBlank_Handler);
 
 	// ARM9 won't try anything funny until this point
 	PXI_Barrier(ARM11_READY_BARRIER);

arm9/source/crypto/crc32.c

@@ -2,8 +2,9 @@
 // https://github.com/eai04191/beat/blob/master/nall/crc32.hpp
 // Ported by Hyarion for use with VirtualFatFS
 
-#include "crc32.h"
 #include "common.h"
+#include "crc32.h"
+#include "vff.h"
 
 u32 crc32_adjust(u32 crc32, u8 input) {
     static const u32 crc32_table[256] = {

arm9/source/crypto/crc32.h

@@ -3,7 +3,8 @@
 // Ported by Hyarion for use with VirtualFatFS
 
 #pragma once
-#include "vff.h"
+
+#include "common.h"
 
 u32 crc32_adjust(u32 crc32, u8 input);
 u32 crc32_calculate(u32 crc32, const u8* data, u32 length);

arm9/source/game/codelzss.c

@@ -215,7 +215,7 @@ void slideByte(sCompressInfo* a_pInfo, const u8* a_pSrc) {
     }
 }
 
-inline void slide(sCompressInfo* a_pInfo, const u8* a_pSrc, int a_nSize) {
+static inline void slide(sCompressInfo* a_pInfo, const u8* a_pSrc, int a_nSize) {
     for (int i = 0; i < a_nSize; i++) {
         slideByte(a_pInfo, a_pSrc--);
     }

arm9/source/gamecart/protocol.c

@@ -36,13 +36,6 @@ static u32 A0_Response = 0xFFFFFFFFu;
 static u32 rand1 = 0;
 static u32 rand2 = 0;
 
-u32 BSWAP32(u32 val) {
-    return (((val >> 24) & 0xFF)) |
-           (((val >> 16) & 0xFF) << 8) |
-           (((val >> 8) & 0xFF) << 16) |
-           ((val & 0xFF) << 24);
-}
-
 // updated function by profi200
 static void ResetCardSlot(void)
 {

arm9/source/gamecart/protocol.h

@@ -12,8 +12,8 @@
 
 
 #define LATENCY 0x822Cu
+#define BSWAP32(n)  __builtin_bswap32(n)
 
-u32 BSWAP32(u32 val);
 
 void Cart_Init(void);
 int Cart_IsInserted(void);

arm9/source/lodepng/lodepng.h

@@ -39,6 +39,18 @@ compiler command to disable them without modifying this header, e.g.
 In addition to those below, you can also define LODEPNG_NO_COMPILE_CRC to
 allow implementing a custom lodepng_crc32.
 */
+
+#define LODEPNG_NO_COMPILE_CRC
+#define LODEPNG_NO_COMPILE_DISK
+#define LODEPNG_NO_COMPILE_ANCILLARY_CHUNKS
+#define LODEPNG_NO_COMPILE_ERROR_TEXT
+
+#include "crc32.h"
+static inline unsigned lodepng_crc32(const unsigned char *data, size_t length)
+{
+  return crc32_calculate(0, data, length);
+}
+
 /*deflate & zlib. If disabled, you must specify alternative zlib functions in
 the custom_zlib field of the compress and decompress settings*/
 #ifndef LODEPNG_NO_COMPILE_ZLIB

common/arm.h

@@ -18,6 +18,7 @@
 #define SR_NOINT	(SR_NOFIQ | SR_NOIRQ)
 
 #ifdef ARM9
+	#define CPU_FREQ	(134055928)
 	#define CR_MPU		BIT(0)
 	#define CR_DCACHE	BIT(2)
 	#define CR_ICACHE	BIT(12)
@@ -38,6 +39,7 @@
 	#define MAX_IRQ	(32)
 	#define MAX_CPU	(1)
 #else // ARM11
+	#define CPU_FREQ	(268111856)
 	#define CR_MMU		BIT(0)
 	#define CR_ALIGN	BIT(1)
 	#define CR_DCACHE	BIT(2)