gpu : added vsh support to makefile + some code. need to setup AEMSTROPATH variable for it to work.

gpu/Makefile

@@ -1,29 +1,50 @@
 CC = arm-none-eabi-gcc
-LINK = arm-none-eabi-ld
+LINK = arm-none-eabi-gcc
+AS = arm-none-eabi-as
 OBJCOPY = arm-none-eabi-objcopy
 CTRULIB = "../libctru"
+AEMSTROPATH = "../../aemstro"
 CFLAGS += -Wall -std=c99 -march=armv6 -O3 -I"$(CTRULIB)/include" -I$(DEVKITPRO)/libnds/include
-LDFLAGS += --script=ccd00.ld -L"$(DEVKITARM)/arm-none-eabi/lib" -L"$(CTRULIB)/lib"
+# LDFLAGS += --script=ccd00.ld -L"$(DEVKITARM)/arm-none-eabi/lib" -L"$(DEVKITARM)/lib/gcc/arm-none-eabi/4.7.1" -L"$(CTRULIB)/lib"
+# LDFLAGS += --script=ccd00.ld -L"$(DEVKITARM)/arm-none-eabi/lib" -L"$(CTRULIB)/lib"
+LDFLAGS += -nostartfiles --specs=ccd00.specs -L"$(DEVKITARM)/arm-none-eabi/lib" -L"$(CTRULIB)/lib"
 
 CFILES = $(wildcard source/*.c)
-OFILES = $(CFILES:source/%.c=build/%.o)
+VSHFILES = $(wildcard source/*.vsh)
+VSHBINFILES = $(VSHFILES:source/%.vsh=data/%.vsh.bin)
+BINFILES = $(wildcard data/*.bin)
+BINFILES += $(VSHBINFILES)
+OFILES = $(BINFILES:data/%.bin=build/%.bin.o)
+OFILES += $(CFILES:source/%.c=build/%.o)
 DFILES = $(CFILES:source/%.c=build/%.d)
 SFILES = $(wildcard source/*.s)
 OFILES += $(SFILES:source/%.s=build/%.o)
 PROJECTNAME = ${shell basename "$(CURDIR)"}
 
+#---------------------------------------------------------------------------------
+# canned command sequence for binary data, taken from devkitARM
+#---------------------------------------------------------------------------------
+define bin2o
+	bin2s $< | $(AS) -o $(@)
+	echo "extern const u8" `(echo $(<F) | sed -e 's/^\([0-9]\)/_\1/' | tr . _)`"_end[];" > source/`(echo $(<F) | tr . _)`.h
+	echo "extern const u8" `(echo $(<F) | sed -e 's/^\([0-9]\)/_\1/' | tr . _)`"[];" >> source/`(echo $(<F) | tr . _)`.h
+	echo "extern const u32" `(echo $(<F) | sed -e 's/^\([0-9]\)/_\1/' | tr . _)`_size";" >> source/`(echo $(<F) | tr . _)`.h
+endef
+
 .PHONY:=all dir
 
 all: dir $(PROJECTNAME).bin
 
 dir:
 	@mkdir -p build
+	@mkdir -p data
 
 $(PROJECTNAME).bin: $(PROJECTNAME).elf
 	$(OBJCOPY) -O binary $< $@
 
-$(PROJECTNAME).elf: $(OFILES)
+$(PROJECTNAME).elf: $(VSHBINFILES) $(OFILES)
-	$(LINK) $(LDFLAGS) -o $(PROJECTNAME).elf $(filter-out build/crt0.o, $(OFILES)) -lctru -lc
+	# $(LINK) $(LDFLAGS) -o $(PROJECTNAME).elf $(filter-out build/crt0.o, $(OFILES)) -lctru -lc -lgcc
+	$(LINK) $(LDFLAGS) -o $(PROJECTNAME).elf $(filter-out build/crt0.o, $(OFILES)) -g -lctru -lm -lc -lg
 
 clean:
 	@rm -f build/*.o build/*.d
@@ -32,6 +53,9 @@ clean:
 
 -include $(DFILES)
 
+data/%.vsh.bin: source/%.vsh
+	@python $(AEMSTROPATH)/aemstro_as.py $< $@
+
 build/%.o: source/%.c
 	$(CC) $(CFLAGS) -c $< -o $@
 	@$(CC) -MM $< > build/$*.d
@@ -39,3 +63,7 @@ build/%.o: source/%.c
 build/%.o: source/%.s
 	$(CC) $(CFLAGS) -c $< -o $@
 	@$(CC) -MM $< > build/$*.d
+
+build/%.bin.o: data/%.bin
+	@echo $(notdir $<)
+	@$(bin2o)

gpu/source/main.c

@@ -8,13 +8,17 @@
 #include <ctr/GX.h>
 #include <ctr/GPU.h>
 #include <ctr/HID.h>
+#include <ctr/SHDR.h>
 #include <ctr/svc.h>
 #include "costable.h"
+#include "test_vsh_bin.h"
 
 u8* gspHeap;
 u32* gxCmdBuf;
 
 u8 currentBuffer;
+u8* topLeftFramebuffers[2];
+u8* topLeftFramebuffersPA[2];
 
 Handle gspEvent, gspSharedMemHandle;
 
@@ -29,6 +33,13 @@ void gspGpuInit()
 	u32 regData=0x01FF0000;
 	GSPGPU_WriteHWRegs(NULL, 0x202A04, &regData, 4);
 
+	//grab main left screen framebuffer addresses
+	GSPGPU_ReadHWRegs(NULL, 0x400468, (u32*)&topLeftFramebuffersPA, 8);
+
+	//convert PA to VA (assuming FB in VRAM)
+	topLeftFramebuffers[0]=topLeftFramebuffersPA[0]+0x7000000;
+	topLeftFramebuffers[1]=topLeftFramebuffersPA[1]+0x7000000;
+
 	//setup our gsp shared mem section
 	u8 threadID;
 	svc_createEvent(&gspEvent, 0x0);
@@ -88,6 +99,8 @@ int main()
 	u32* gpuCmd=(u32*)gspHeap;
 	u32 gpuCmdSize=0x10000;
 
+	DVLB_s* shader=SHDR_ParseSHBIN((u32*)test_vsh_bin,test_vsh_bin_size);
+
 	APP_STATUS status;
 	while((status=aptGetStatus())!=APP_EXITING)
 	{

gpu/source/test.vsh

@@ -0,0 +1,43 @@
+; setup constants
+	.const 5, 0.0, 1.0, 2.0, 3.0
+
+; setup outmap
+	.out o0, result.position
+	.out o1, result.color
+	.out o2, result.texcoord0
+	.out o3, result.texcoord1
+	.out o4, result.texcoord2
+
+; setup uniform map (not required)
+	.uniform 0x10, 0x13, mdlvMtx
+	.uniform 0x14, 0x17, projMtx
+
+;code
+	main:
+		; result.pos = mdlvMtx * in.pos
+		dp4 d40, d40, d00 (0x0)
+		dp4 d40, d41, d00 (0x1)
+		dp4 d40, d42, d00 (0x2)
+		mov d40, d25 (0x4)
+		; result.pos = projMtx * in.pos
+		dp4 d00, d44, d40 (0x0)
+		dp4 d00, d45, d40 (0x1)
+		dp4 d00, d46, d40 (0x2)
+		dp4 d00, d47, d40 (0x3)
+		; result.color = in.pos
+		mov d04, d25 (0x5)
+		; result.texcoord = const
+		mov d08, d25 (0x5)
+		mov d0C, d25 (0x5)
+		mov d10, d25 (0x5)
+		flush
+		end
+	endmain:
+
+;operand descriptors
+	.opdesc x___, xyzw, xyzw ; 0x0
+	.opdesc _y__, xyzw, xyzw ; 0x1
+	.opdesc __z_, xyzw, xyzw ; 0x2
+	.opdesc ___w, xyzw, xyzw ; 0x3
+	.opdesc ___w, yyyy, xyzw ; 0x4
+	.opdesc xyzw, xyzw, xyzw ; 0x5