add debug output to interrupt handler, better stack usage

64drive.inc

@@ -32,3 +32,21 @@ constant CI_WIFI_PARAM_RESULT_1($0428)  // word
 
 constant CI_EEPROM($1000)               // 2048 bytes
 constant CI_WRITEBACK_LBA($1800)        // 1024 bytes (256 words)
+
+macro CI_WAIT() {
+    PI_WAIT()
+-
+    lw      t0, CI_STATUS(t9)
+    srl     t0, t0, 12 // first 12 bits are reserved, so ignore them
+    bnez    t0,-
+    nop // delay slot
+}
+
+macro CI_USB_WRITE_WAIT() {
+    PI_WAIT()
+-
+    lw      t0, CI_USB_COMMAND_STATUS(t9)
+    srl     t0, t0, 4 // shift out ARM status, leaving WRITE status
+    bnez    t0,-
+    nop // delay slot
+}

debug.asm

@@ -50,7 +50,7 @@ Drive64Write:
     sw      t3, PI_RD_LEN(t5)  // "read" from DRAM to cart
 //  PI_WAIT() // if we always wait before doing operations, this shouldn't be necessary
 
-Drive64TestPoint:
+Drive64WriteDirect: // TODO: rewrite so this takes a0, a1 instead of a2, a3
     lui     at, 0x0100      // set printf channel
     or      a3, a3, at
     lli     t1, 0x08        // WRITE mode
@@ -59,20 +59,56 @@ Drive64TestPoint:
     and     a2, a2, t9
     srl     a2, a2, 1
 
-    CI_USB_WRITE_WAIT()
+    lui     t9, K_BASE
-    sw      a2, CI_USB_PARAM_RESULT_0(gp)
+    lw      t9, K_CI_BASE(t9)
+
+    CI_USB_WRITE_WAIT() // clobbers t0, requires t9
+    sw      a2, CI_USB_PARAM_RESULT_0(t9)
     PI_WAIT() // yes, these waits seem to be necessary
-    sw      a3, CI_USB_PARAM_RESULT_1(gp)
+    sw      a3, CI_USB_PARAM_RESULT_1(t9)
     PI_WAIT()
-    sw      t1, CI_USB_COMMAND_STATUS(gp)
+    sw      t1, CI_USB_COMMAND_STATUS(t9)
-//  CI_USB_WRITE_WAIT() // if we always wait before doing operations, this shouldn't be necessary
+// if we always wait before doing operations, this shouldn't be necessary:
+//  CI_USB_WRITE_WAIT()
 
 Drive64WriteExit:
     jr      ra
-    nop // delay slot
+    nop
 
 Drive64TestWrite:
     li      a2, 0xA0000020
     lli     a3, 0x20
-    j       Drive64TestPoint
+    j       Drive64WriteDirect
-    nop // delay slot
+    nop
+
+include "xxd.asm"
+
+DumpAndWrite:
+    // a0: source address
+    // a1: source length
+    // a2: temp string address
+    // a3: temp string maximum length
+    // v0: error code (0 is OK)
+    subiu   sp, sp, 0x20
+    sw      ra, 0x10(sp)
+    // TODO: i think i can just use the a0,a1,a2,a3 slots here?
+    sw      s0, 0x14(sp)
+    sw      s1, 0x18(sp)
+
+    or      s0, a2, r0
+    jal     xxd
+    or      s1, a3, r0
+
+    bnez    v0, DumpAndWriteExit
+
+    ori     a0, s0, r0 // delay slot
+    jal     Drive64Write
+    ori     a1, s1, r0
+    // v0 passthru
+
+DumpAndWriteExit:
+    lw      ra, 0x10(sp)
+    lw      s0, 0x14(sp)
+    lw      s1, 0x18(sp)
+    jr      ra
+    addiu   sp, sp, 0x20

kernel.asm

@@ -28,37 +28,37 @@ Start:
     lui     sp, K_STACK_INIT_BASE
     // SP should always be 8-byte aligned
     // so that SD and LD instructions don't fail on it.
-    subiu   sp, sp, 8
+    // we also need 4 empty words for storing the 32-bit values of a0,a1,a2,a3
+    subiu   sp, sp, 0x10
 
     // TODO: just wipe a portion of RAM?
     sw      r0, K_64DRIVE_MAGIC(k0)
     sw      r0, K_REASON(k0)
 
 Drive64Init:
-    lui     gp, CI_BASE
+    lui     t9, CI_BASE
     lui     t2, 0x5544      // "UD" of "UDEV"
-    lw      t1, CI_HW_MAGIC(gp)
+    lw      t1, CI_HW_MAGIC(t9)
     ori     t2, t2, 0x4556  // "EV" of "UDEV"
 
     beq     t1, t2, Drive64Confirmed
     nop
 
 Drive64TryExtended:
-    lui     gp, CI_BASE_EXTENDED
+    lui     t9, CI_BASE_EXTENDED
-    lw      t1, CI_HW_MAGIC(gp)
+    lw      t1, CI_HW_MAGIC(t9)
     bne     t1, t2, Drive64Done
     nop
 
 Drive64Confirmed:
     sw      t2, K_64DRIVE_MAGIC(k0)
-    sw      gp, K_CI_BASE(k0)
+    sw      t9, K_CI_BASE(k0)
 
     // enable writing to cartROM (SDRAM) for USB writing later
     lli     t1, 0xF0
-
+    CI_WAIT() // clobbers t0, requires t9
-    CI_WAIT()
+    sw      t1, CI_COMMAND(t9)
-    sw      t1, CI_COMMAND(gp) // send our command
+    CI_WAIT() // clobbers t0, requires t9
-    CI_WAIT()
 
 Drive64Done:
 
@@ -199,7 +199,30 @@ InterruptHandler:
     sd      t0, K_DUMP+0x100(k0)
     sd      t1, K_DUMP+0x108(k0)
 
-    // free to modify any GPR here
+    // be wary, this is a tiny temporary stack!
+    ori     sp, k0, K_STACK
+
+IHMain: // free to modify any GPR from here to IHExit
+    la      a2, IHString
+    jal     Drive64WriteDirect
+    lli     a3, 0x20 //IHString.size
+
+    ori     a0, k0, K_DUMP
+    lli     a1, 0x100
+    ori     a2, k0, K_XXD
+    jal     DumpAndWrite
+    lli     a3, 0x400
+
+    ori     a0, k0, K_DUMP
+    addiu   a0, a0, 0x100
+    lli     a1, 0x100
+    ori     a2, k0, K_XXD
+    jal     DumpAndWrite
+    lli     a3, 0x400
+
+IHExit:
+
+    jal     Drive64Write
 
     lui     k0, K_BASE
     ld      t0, K_DUMP+0x100(k0)
@@ -257,4 +280,11 @@ ReturnFromInterrupt:
     jr      k0
     or      k1, r0, r0
 
+include "debug.asm"
+
+align(4)
+IHString:
+    db " ~~~ Interrupt Handled ~~~ ", 0
+
+align(4)
     nops((K_BASE << 16) + 0x10000)

main.asm

@@ -24,7 +24,6 @@ include "kernel.asm"
 
 Main:
     lui     t0, K_BASE
-    lw      gp, K_CI_BASE(t0)
 
     lui     s0, BLAH_BASE
 
@@ -94,8 +93,6 @@ MainLoop:
     j MainLoop
     nop // delay slot
 
-include "debug.asm" // assumes gp is set to CI base
-
 SetupScreen:
     // NTSC: 640x480, 32BPP, Interlace, Resample Only, DRAM Origin VIDEO_BUFFER
     ScreenNTSC(640, 480, BPP32|INTERLACE|AA_MODE_2, VIDEO_BUFFER | UNCACHED)
@@ -116,8 +113,8 @@ LoadRSPBoot:
 
 PushVideoTask:
     // a0: Task RDRAM Pointer (size: 0x40)
-    subiu   sp, sp, 0x8
+    subiu   sp, sp, 0x18
-    sw      ra, 0(sp)
+    sw      ra, 0x10(sp)
 
     lli     t0, 1 // mode: video
     lli     t1, 4 // flags: ???
@@ -154,9 +151,9 @@ PushVideoTask:
     jal     PushRSPTask // a0 passthru
     nop
 
-    lw      ra, 0(sp)
+    lw      ra, 0x10(sp)
     jr      ra
-    addiu   sp, sp, 0x8
+    addiu   sp, sp, 0x18
 
 PushRSPTask:
     lli     t3, 0x40 - 1 // DMA quirk
@@ -168,8 +165,6 @@ PushRSPTask:
     jr      ra
     nop
 
-include "xxd.asm"
-
 align(16); insert F3DZEX_BOOT, "F3DZEX2.boot.bin"
 align(16); insert F3DZEX_DMEM, "F3DZEX2.data.bin"
 align(16); insert F3DZEX_IMEM, "F3DZEX2.bin"

main.inc

@@ -9,8 +9,10 @@ constant K_REASON(0x0600)
 constant K_CAUSE(0x0604)
 constant K_STATUS(0x0608)
 constant K_EPC(0x060C)
-constant K_64DRIVE_MAGIC(0x0800)
+constant K_64DRIVE_MAGIC(0x0700)
-constant K_CI_BASE(0x0804)
+constant K_CI_BASE(0x0704)
+constant K_STACK(0xC00 - 8)
+constant K_XXD(0x0C00) // size: 0x400 (any larger and you overwrite kernel code)
 
 constant K_STACK_INIT_BASE(0x803F) // note that this gets subtracted by 8
                                    // and that the stack grows *backwards.*
@@ -60,22 +62,3 @@ macro SP_DMA_WAIT() { // external
     bnez    t0,-
     nop
 }
-
-macro CI_WAIT() {
-    PI_WAIT()
--
-    lw      t0, CI_STATUS(gp)
-    srl     t0, t0, 12 // first 12 bits are reserved, so ignore them
-    bnez    t0,-
-    nop // delay slot
-}
-
-macro CI_USB_WRITE_WAIT() {
-    PI_WAIT()
--
-    lw      t0, CI_USB_COMMAND_STATUS(gp)
-    srl     t0, t0, 4 // shift out ARM status, leaving WRITE status
-    bnez    t0,-
-    nop // delay slot
-}
-