Improve the state of handwritten assembly files (#865)

* Format all handwritten asm and document some

* Use c preprocessor for constants

* Fix

* Fix PI_STATUS_ERROR, some label improvements

* Avoid hi/lo for constants

* Some more comments

* Properly mark functions as functions and their sizes

* Fix merge

* Improvements

* Review suggestions, rework procedure start/end macros to be more like libreultra

* Move IPL3 symbol definitions into ipl3.s

* Fix undefined_syms, add include and language guards to asm.h and fix the comment in gbi.h

* Consistent hex capitalization, add some MIPS builtin defines to CC_CHECK to behave properly

* Add -no-pad-sections assembler option and clean up alignment in gu files and bzero

* Further suggestions and improvements

* Matrix conversion function clarifications

* Fix passing AVOID_UB to gcc

* Suggestions

* Suggestions, global interrupt mask improvements

* Further suggestions, interrupt mask comments

* Comments fixes, rdb.h

* Switch from # comments to // comments, remove unnecesary .set gp=64 directives

* Further review suggestions

* Missed one

src/boot/boot_main.c

@@ -1,10 +1,11 @@
 #include "global.h"
+#include "boot.h"
 
 StackEntry sBootThreadInfo;
 OSThread sIdleThread;
 STACK(sIdleThreadStack, 0x400);
 StackEntry sIdleThreadInfo;
-STACK(sBootThreadStack, 0x400);
+STACK(sBootThreadStack, BOOT_STACK_SIZE);
 
 void cleararena(void) {
     bzero(_dmadataSegmentStart, osMemSize - OS_K0_TO_PHYSICAL(_dmadataSegmentStart));

src/code/main.c

@@ -46,7 +46,7 @@ void Main(void* arg) {
     osSyncPrintf("mainproc 実行開始\n"); // "Start running"
     gScreenWidth = SCREEN_WIDTH;
     gScreenHeight = SCREEN_HEIGHT;
-    gAppNmiBufferPtr = (PreNmiBuff*)osAppNmiBuffer;
+    gAppNmiBufferPtr = (PreNmiBuff*)osAppNMIBuffer;
     PreNmiBuff_Init(gAppNmiBufferPtr);
     Fault_Init();
     SysCfb_Init(0);

src/libultra/io/cartrominit.c

@@ -27,8 +27,9 @@ OSPiHandle* osCartRomInit(void) {
     __CartRomHandle.speed = 0;
     bzero(&__CartRomHandle.transferInfo, sizeof(__OSTranxInfo));
 
-    while (status = HW_REG(PI_STATUS_REG, u32), status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
-        ;
+    status = HW_REG(PI_STATUS_REG, u32);
+    while (status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
+        status = HW_REG(PI_STATUS_REG, u32);
     }
 
     lastLatency = HW_REG(PI_BSD_DOM1_LAT_REG, u32);

src/libultra/io/devmgr.c

@@ -28,7 +28,7 @@ void __osDevMgrMain(void* arg) {
             phi_s2 = ((transfer->transferMode == 2) && (ioMesg->piHandle->transferInfo.cmdType == 0)) ? 1 : 0;
 
             osRecvMesg(arg0->acccessQueue, &sp6C, OS_MESG_BLOCK);
-            __osResetGlobalIntMask(0x00100401);
+            __osResetGlobalIntMask(OS_IM_PI);
             __osEPiRawWriteIo(ioMesg->piHandle, 0x05000510, transfer->bmCtlShadow | 0x80000000);
 
             while (true) {
@@ -43,8 +43,8 @@ void __osDevMgrMain(void* arg) {
                         __osEPiRawWriteIo(ioMesg->piHandle, 0x05000510, transfer->bmCtlShadow | 0x1000000);
                     }
                     block->errStatus = 4;
-                    HW_REG(PI_STATUS_REG, u32) = PI_STATUS_CLEAR_INTR;
-                    __osSetGlobalIntMask(0x00100C01);
+                    HW_REG(PI_STATUS_REG, u32) = PI_STATUS_CLR_INTR;
+                    __osSetGlobalIntMask(OS_IM_CART | OS_IM_PI);
                 }
                 osSendMesg(ioMesg->hdr.retQueue, (OSMesg)ioMesg, OS_MESG_NOBLOCK);
 

src/libultra/io/driverominit.c

@@ -22,8 +22,9 @@ OSPiHandle* osDriveRomInit(void) {
     __DriveRomHandle.speed = 0;
     bzero(&__DriveRomHandle.transferInfo, sizeof(__OSTranxInfo));
 
-    while (status = HW_REG(PI_STATUS_REG, u32), status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
-        ;
+    status = HW_REG(PI_STATUS_REG, u32);
+    while (status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
+        status = HW_REG(PI_STATUS_REG, u32);
     }
 
     HW_REG(PI_BSD_DOM1_LAT_REG, u32) = 0xFF;

src/libultra/io/epirawdma.c

@@ -4,8 +4,9 @@ s32 __osEPiRawStartDma(OSPiHandle* handle, s32 direction, u32 cartAddr, void* dr
     s32 status;
     OSPiHandle* curHandle;
 
-    while (status = HW_REG(PI_STATUS_REG, u32), status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
-        ;
+    status = HW_REG(PI_STATUS_REG, u32);
+    while (status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
+        status = HW_REG(PI_STATUS_REG, u32);
     }
 
     if (__osCurrentHandle[handle->domain]->type != handle->type) {

src/libultra/io/epirawread.c

@@ -4,8 +4,9 @@ s32 __osEPiRawReadIo(OSPiHandle* handle, u32 devAddr, u32* data) {
     s32 status;
     OSPiHandle* curHandle;
 
-    while (status = HW_REG(PI_STATUS_REG, u32), status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
-        ;
+    status = HW_REG(PI_STATUS_REG, u32);
+    while (status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
+        status = HW_REG(PI_STATUS_REG, u32);
     }
 
     if (__osCurrentHandle[handle->domain]->type != handle->type) {

src/libultra/io/epirawwrite.c

@@ -4,8 +4,9 @@ s32 __osEPiRawWriteIo(OSPiHandle* handle, u32 devAddr, u32 data) {
     s32 status;
     OSPiHandle* curHandle;
 
-    while (status = HW_REG(PI_STATUS_REG, u32), status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
-        ;
+    status = HW_REG(PI_STATUS_REG, u32);
+    while (status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
+        status = HW_REG(PI_STATUS_REG, u32);
     }
 
     if (__osCurrentHandle[handle->domain]->type != handle->type) {

src/libultra/io/pirawdma.c

@@ -1,8 +1,9 @@
 #include "global.h"
 
 s32 __osPiRawStartDma(s32 dir, u32 cartAddr, void* dramAddr, size_t size) {
-    register s32 status = HW_REG(PI_STATUS_REG, u32);
+    s32 status;
 
+    status = HW_REG(PI_STATUS_REG, u32);
     while (status & (PI_STATUS_BUSY | PI_STATUS_IOBUSY)) {
         status = HW_REG(PI_STATUS_REG, u32);
     }

src/libultra/os/createthread.c

@@ -1,4 +1,5 @@
 #include "global.h"
+#include "ultra64/asm.h"
 
 __OSThreadTail __osThreadTail = { NULL, OS_PRIORITY_THREADTAIL };
 OSThread* __osRunQueue = (OSThread*)&__osThreadTail;
@@ -16,7 +17,7 @@ void osCreateThread(OSThread* thread, OSId id, void (*entry)(void*), void* arg,
     thread->queue = NULL;
     thread->context.pc = (u32)entry;
     thread->context.a0 = arg;
-    thread->context.sp = (u64)(s32)sp - 16;
+    thread->context.sp = (u64)(s32)sp - FRAMESZ(SZREG * NARGSAVE);
     thread->context.ra = __osCleanupThread;
 
     mask = OS_IM_ALL;

src/libultra/os/initialize.c

@@ -7,6 +7,8 @@ typedef struct {
     u32 ins_0C; // nop
 } struct_exceptionPreamble;
 
+void __osExceptionPreamble(void);
+
 u64 osClockRate = OS_CLOCK_RATE;
 s32 osViClock = VI_NTSC_CLOCK;
 u32 __osShutdown = 0;
@@ -58,7 +60,7 @@ void __osInitialize_common(void) {
     osClockRate = (u64)((osClockRate * 3ll) / 4ull);
 
     if (!osResetType) {
-        bzero(osAppNmiBuffer, sizeof(osAppNmiBuffer));
+        bzero(osAppNMIBuffer, sizeof(osAppNMIBuffer));
     }
 
     if (osTvType == OS_TV_PAL) {

src/libultra/os/resetglobalintmask.c

@@ -3,6 +3,6 @@
 void __osResetGlobalIntMask(OSHWIntr mask) {
     register u32 prevInt = __osDisableInt();
 
-    __OSGlobalIntMask &= ~(mask & ~0x401);
+    __OSGlobalIntMask &= ~(mask & ~OS_IM_RCP);
     __osRestoreInt(prevInt);
 }