boot: use #embed for fusee

Also, fix a few very embarassing mistakes in kernel ldr:
* We have been mapping the page table region RWX for a few years now, accidentally.
* My attempt at making initial page tables not use bit 58 was broken in multiple ways.

docs/features/cheats.md

@@ -49,7 +49,7 @@ Code type 0x0 allows writing a static value to a memory address.
 `0TMR00AA AAAAAAAA VVVVVVVV (VVVVVVVV)`
 
 + T: Width of memory write (1, 2, 4, or 8 bytes).
-+ M: Memory region to write to (0 = Main NSO, 1 = Heap, 2 = Alias, 3 = Aslr).
++ M: Memory region to write to (0 = Main NSO, 1 = Heap, 2 = Alias, 3 = Aslr, 4 = non-relative).
 + R: Register to use as an offset from memory region base.
 + A: Immediate offset to use from memory region base.
 + V: Value to write.
@@ -62,11 +62,13 @@ Code type 0x1 performs a comparison of the contents of memory to a static value.
 If the condition is not met, all instructions until the appropriate End or Else conditional block terminator are skipped.
 
 #### Encoding
-`1TMC00AA AAAAAAAA VVVVVVVV (VVVVVVVV)`
+`1TMCXrAA AAAAAAAA VVVVVVVV (VVVVVVVV)`
 
-+ T: Width of memory write (1, 2, 4, or 8 bytes).
++ T: Width of memory read (1, 2, 4, or 8 bytes).
-+ M: Memory region to write to (0 = Main NSO, 1 = Heap, 2 = Alias, 3 = Aslr).
++ M: Memory region to read from (0 = Main NSO, 1 = Heap, 2 = Alias, 3 = Aslr, 4 = non-relative).
 + C: Condition to use, see below.
++ X: Operand Type, see below.
++ r: Offset Register (operand types 1).
 + A: Immediate offset to use from memory region base.
 + V: Value to compare to.
 
@@ -78,6 +80,9 @@ If the condition is not met, all instructions until the appropriate End or Else
 + 5: ==
 + 6: !=
 
+#### Operand Type
++ 0: Memory Base + Relative Offset
++ 1: Memory Base + Offset Register + Relative Offset
 ---
 
 ### Code Type 0x2: End Conditional Block
@@ -126,7 +131,7 @@ Code type 0x5 allows loading a value from memory into a register, either using a
 `5TMR00AA AAAAAAAA`
 
 + T: Width of memory read (1, 2, 4, or 8 bytes).
-+ M: Memory region to write to (0 = Main NSO, 1 = Heap, 2 = Alias, 3 = Aslr).
++ M: Memory region to write to (0 = Main NSO, 1 = Heap, 2 = Alias, 3 = Aslr, 4 = non-relative).
 + R: Register to load value into.
 + A: Immediate offset to use from memory region base.
 
@@ -137,6 +142,22 @@ Code type 0x5 allows loading a value from memory into a register, either using a
 + R: Register to load value into. (This register is also used as the base memory address).
 + A: Immediate offset to use from register R.
 
+#### Load from Register Address Encoding
+`5T0R2SAA AAAAAAAA`
+
++ T: Width of memory read (1, 2, 4, or 8 bytes).
++ R: Register to load value into. 
++ S: Register to use as the base memory address.
++ A: Immediate offset to use from register R.
+
+#### Load From Fixed Address Encoding with offset register
+`5TMR3SAA AAAAAAAA`
+
++ T: Width of memory read (1, 2, 4, or 8 bytes).
++ M: Memory region to write to (0 = Main NSO, 1 = Heap, 2 = Alias, 3 = Aslr, 4 = non-relative).
++ R: Register to load value into.
++ S: Register to use as offset register.
++ A: Immediate offset to use from memory region base.
 ---
 
 ### Code Type 0x6: Store Static Value to Register Memory Address
@@ -250,7 +271,10 @@ Code type 0x9 allows performing arithmetic on registers.
 + 7: Logical Not (discards right-hand operand)
 + 8: Logical Xor
 + 9: None/Move (discards right-hand operand)
-
++ 10: Float Addition, T==4 single T==8 double
++ 11: Float Subtraction, T==4 single T==8 double
++ 12: Float Multiplication, T==4 single T==8 double
++ 13: Float Division, T==4 single T==8 double
 ---
 
 ### Code Type 0xA: Store Register to Memory Address
@@ -380,6 +404,61 @@ Code type 0xC3 reads or writes a static register with a given register.
 
 ---
 
+### Code Type 0xC4: Begin Extended Keypress Conditional Block
+Code type 0xC4 enters or skips a conditional block based on whether a key combination is pressed.
+
+#### Encoding
+`C4r00000 kkkkkkkk kkkkkkkk`
+
++ r: Auto-repeat, see below.
++ kkkkkkkkkk: Keypad mask to check against output of `hidKeysDown()`.
+
+Note that for multiple button combinations, the bitmasks should be OR'd together.
+
+#### Auto-repeat
+
++ 0: The conditional block executes only once when the keypad mask matches. The mask must stop matching to reset for the next trigger.
++ 1: The conditional block executes as long as the keypad mask matches.
+
+#### Keypad Values
+Note: This is the direct output of `hidKeysDown()`.
+
++ 000000001: A
++ 000000002: B
++ 000000004: X
++ 000000008: Y
++ 000000010: Left Stick Pressed
++ 000000020: Right Stick Pressed
++ 000000040: L
++ 000000080: R
++ 000000100: ZL
++ 000000200: ZR
++ 000000400: Plus
++ 000000800: Minus
++ 000001000: Left
++ 000002000: Up
++ 000004000: Right
++ 000008000: Down
++ 000010000: Left Stick Left
++ 000020000: Left Stick Up
++ 000040000: Left Stick Right
++ 000080000: Left Stick Down
++ 000100000: Right Stick Left
++ 000200000: Right Stick Up
++ 000400000: Right Stick Right
++ 000800000: Right Stick Down
++ 001000000: SL Left Joy-Con
++ 002000000: SR Left Joy-Con
++ 004000000: SL Right Joy-Con
++ 008000000: SR Right Joy-Con
++ 010000000: Top button on Poké Ball Plus (Palma) controller
++ 020000000: Verification
++ 040000000: B button on Left NES/HVC controller in Handheld mode
++ 080000000: Left C button in N64 controller
++ 100000000: Up C button in N64 controller
++ 200000000: Right C button in N64 controller
++ 400000000: Down C button in N64 controller
+
 ### Code Type 0xF0: Double Extended-Width Instruction
 Code Type 0xF0 signals to the VM to treat the upper three nybbles of the first dword as instruction type, instead of just the upper nybble.
 

emummc/source/utils/types.h

@@ -18,6 +18,7 @@
 #define _TYPES_H_
 
 #include <stdint.h>
+#include <stdbool.h>
 
 #define ALIGN(x, a) (((x) + (a) - 1) & ~((a) - 1))
 #define MAX(a, b) ((a) > (b) ? (a) : (b))
@@ -58,12 +59,6 @@ typedef u32 Result; ///< Function error code result type.
 #define INVALID_HANDLE ((Handle) 0)
 #define CUR_PROCESS_HANDLE ((Handle) 0xFFFF8001)
 
-#ifndef __cplusplus
-typedef int bool;
-#define true  1
-#define false 0
-#endif /* __cplusplus */
-
 #define BOOT_CFG_AUTOBOOT_EN (1 << 0)
 #define BOOT_CFG_FROM_LAUNCH (1 << 1)
 #define BOOT_CFG_SEPT_RUN    (1 << 7)

exosphere/loader_stub/loader_stub.mk

@@ -15,13 +15,12 @@ ifneq ($(__RECURSIVE__),1)
 export ATMOSPHERE_TOPDIR := $(CURRENT_DIRECTORY)
 
 export VPATH	:=	$(foreach dir,$(SOURCES),$(CURDIR)/$(dir)) $(CURDIR)/include \
-			$(foreach dir,$(DATA),$(CURDIR)/$(dir)) \
+			$(foreach dir,$(DATA),$(CURDIR)/$(dir))
-            $(CURRENT_DIRECTORY)/../program/$(ATMOSPHERE_OUT_DIR)
 
 CFILES      :=	$(call FIND_SOURCE_FILES,$(SOURCES),c)
 CPPFILES    :=	$(call FIND_SOURCE_FILES,$(SOURCES),cpp)
 SFILES      :=	$(call FIND_SOURCE_FILES,$(SOURCES),s)
-BINFILES    := program.lz4 boot_code.lz4
+BINFILES    :=
 
 #---------------------------------------------------------------------------------
 # use CXX for linking C++ projects, CC for standard C
@@ -102,13 +101,7 @@ $(OUTPUT).elf	:	$(OFILES)
 
 $(OFILES)	:	 $(ATMOSPHERE_LIBRARIES_DIR)/libexosphere/$(ATMOSPHERE_LIBRARY_DIR)/libexosphere.a
 
-program.lz4.o: program.lz4
+secmon_loader_main.o: CXXFLAGS += --embed-dir="$(CURRENT_DIRECTORY)/../program/$(ATMOSPHERE_OUT_DIR)/"
-	@echo $(notdir $<)
-	@$(bin2o)
-
-boot_code.lz4.o: boot_code.lz4
-	@echo $(notdir $<)
-	@$(bin2o)
 
 %.elf:
 	@echo linking $(notdir $@)
@@ -117,14 +110,6 @@ boot_code.lz4.o: boot_code.lz4
 
 $(OFILES_SRC)	: $(OFILES_BIN)
 
-#---------------------------------------------------------------------------------
-# you need a rule like this for each extension you use as binary data
-#---------------------------------------------------------------------------------
-%.bin.o	%_bin.h:	%.bin
-#---------------------------------------------------------------------------------
-	@echo $(notdir $<)
-	@$(bin2o)
-
 -include $(DEPENDS)
 
 #---------------------------------------------------------------------------------------

exosphere/loader_stub/source/secmon_loader_main.cpp

@@ -15,21 +15,31 @@
  */
 #include <exosphere.hpp>
 #include "secmon_loader_uncompress.hpp"
-#include "program_lz4.h"
-#include "boot_code_lz4.h"
 
 namespace ams::secmon::loader {
 
-    NORETURN void UncompressAndExecute(const void *program, const void *boot_code) {
+    namespace {
+
+        constexpr const u8 SecmonProgramLz4[] = {
+            #embed <program.lz4>
+        };
+
+        constexpr const u8 SecmonBootCodeLz4[] = {
+            #embed <boot_code.lz4>
+        };
+
+    }
+
+    NORETURN void UncompressAndExecute() {
         /* Uncompress the program image. */
-        Uncompress(secmon::MemoryRegionPhysicalTzramFullProgramImage.GetPointer(), secmon::MemoryRegionPhysicalTzramFullProgramImage.GetSize(), program, program_lz4_size);
+        Uncompress(secmon::MemoryRegionPhysicalTzramFullProgramImage.GetPointer(), secmon::MemoryRegionPhysicalTzramFullProgramImage.GetSize(), SecmonProgramLz4, sizeof(SecmonProgramLz4));
 
         /* Copy the boot image to the end of IRAM */
-        u8 *relocated_boot_code = secmon::MemoryRegionPhysicalIramBootCodeImage.GetEndPointer<u8>() - boot_code_lz4_size;
+        u8 *relocated_boot_code = secmon::MemoryRegionPhysicalIramBootCodeImage.GetEndPointer<u8>() - sizeof(SecmonBootCodeLz4);
-        std::memcpy(relocated_boot_code, boot_code, boot_code_lz4_size);
+        std::memcpy(relocated_boot_code, SecmonBootCodeLz4, sizeof(SecmonBootCodeLz4));
 
         /* Uncompress the boot image. */
-        Uncompress(secmon::MemoryRegionPhysicalIramBootCodeImage.GetPointer(), secmon::MemoryRegionPhysicalIramBootCodeImage.GetSize(), relocated_boot_code, boot_code_lz4_size);
+        Uncompress(secmon::MemoryRegionPhysicalIramBootCodeImage.GetPointer(), secmon::MemoryRegionPhysicalIramBootCodeImage.GetSize(), relocated_boot_code, sizeof(SecmonBootCodeLz4));
 
         /* Jump to the boot image. */
         reinterpret_cast<void (*)()>(secmon::MemoryRegionPhysicalIramBootCodeImage.GetAddress())();

exosphere/loader_stub/source/start.s

@@ -98,8 +98,5 @@ _start:
     ldr x20, =0x7C020000
     mov sp, x20
 
-    adr x0, program_lz4
-    adr x1, boot_code_lz4
-
     /* Uncompress the program and iram boot code images. */
-    b _ZN3ams6secmon6loader20UncompressAndExecuteEPKvS3_
+    b _ZN3ams6secmon6loader20UncompressAndExecuteEv

exosphere/program/program.ld

@@ -106,6 +106,7 @@ SECTIONS
     .debug_code : {
         KEEP (*(.text._ZN3ams3log6PrintfEPKcz .text._ZN3ams3log7VPrintfEPKcSt9__va_list .text._ZN3ams3log4DumpEPKvm))
         KEEP (*(.text._ZN3ams4util10TVSNPrintfEPcmPKcSt9__va_list .text._ZN3ams4util12_GLOBAL__N_114TVSNPrintfImplEPcmPKcSt9__va_list .text._ZZN3ams4util12_GLOBAL__N_114TVSNPrintfImplEPcmPKcSt9__va_listENKUlbmE3_clEbm))
+        KEEP (*(.text._ZN3ams4util12_GLOBAL__N_1L14TVSNPrintfImplEPcmPKcSt9__va_list .text._ZZN3ams4util12_GLOBAL__N_1L14TVSNPrintfImplEPcmPKcSt9__va_listENKUlbmE_clEbm))
         KEEP(secmon_exception_handler.o(.text*))
         secmon_exception_handler.o(.rodata*)
         secmon_exception_handler.o(.data*)

fusee/program/source/mtc/fusee_mtc_ram_training_pattern.inc

@@ -421,6 +421,7 @@ constexpr const u8 EmcRamTrainingPatternData[] {
     0x0A, 0x05, 0x0C, 0x03, 0x0A, 0x05, 0x0C, 0x03, 0x0A, 0x05, 0x0C, 0x03, 0x0A, 0x05, 0x0C, 0x03,
     0x0A, 0x05, 0x0C, 0x03, 0x0A, 0x05, 0x0C, 0x03, 0x0A, 0x05, 0x0C, 0x03, 0x0A, 0x05, 0x0C, 0x03
 };
+static_assert(sizeof(EmcRamTrainingPatternData) % sizeof(EmcRamTrainingPattern) == 0);
 
 ALWAYS_INLINE const EmcRamTrainingPattern *GetEmcRamTrainingPattern() {
     return reinterpret_cast<const EmcRamTrainingPattern *>(EmcRamTrainingPatternData);