diff --git a/src/game/objectives.c b/src/game/objectives.c
index 9c3695d78..1831d442e 100644
--- a/src/game/objectives.c
+++ b/src/game/objectives.c
@@ -107,10 +107,12 @@ void tagsReset(void)
 
 #if PIRACYCHECKS
 	{
-		u32 a = xorBaffbeff(0xb0000a5c ^ 0xbaffbeff);
-		u32 b = xorBabeffff(0x1740fff9 ^ 0xbabeffff);
+		// mtxGetObfuscatedRomBase() returns the value at ROM offset 0xa5c.
+		// This value should be 0x1740fff9.
+		u32 dummy = xorBaffbeff(0xb0000a5c ^ 0xbaffbeff);
+		u32 expected = xorBabeffff(0x1740fff9 ^ 0xbabeffff);
 
-		if (mtxGetObfuscatedRomBase() != b) {
+		if (mtxGetObfuscatedRomBase() != expected) {
 			// Read 4KB from a random ROM location within 128KB from the start of
 			// the ROM, and write it to a random memory location between 0x80010000
 			// and 0x80030ff8. This will corrupt instructions in the lib segment.
diff --git a/src/include/lib/mtx.h b/src/include/lib/mtx.h
index c3873db6e..faf780653 100644
--- a/src/include/lib/mtx.h
+++ b/src/include/lib/mtx.h
@@ -12,7 +12,7 @@ void mtx4RotateVec(Mtxf *mtx, struct coord *vec, struct coord *dst);
 void mtx4TransformVecInPlace(Mtxf *mtx, struct coord *vec);
 void mtx4TransformVec(Mtxf *mtx, struct coord *vec, struct coord *dst);
 void mtx00015be0(Mtxf *matrix1, Mtxf *matrix2);
-void mtx00015be4(Mtxf *arg0, Mtxf *arg1, Mtxf *arg2);
+void mtx00015be4(Mtxf *arg0, Mtxf *arg1, Mtxf *dst);
 void mtx3Copy(f32 src[3][3], f32 dst[3][3]);
 void mtx4Copy(Mtxf *src, Mtxf *dst);
 void mtx3ToMtx4(f32 src[3][3], Mtxf *dst);
diff --git a/src/lib/mtxasm.s b/src/lib/mtxasm.s
index fcddf03e6..4b9eb4b24 100644
--- a/src/lib/mtxasm.s
+++ b/src/lib/mtxasm.s
@@ -461,15 +461,8 @@ glabel mtx00015f88
  	nop
 
 /**
- * Wait for the PI to be idle, then read the osRomBase variable, obfuscate its
- * value and return it.
- *
- * The returned value is (osRomBase | 0xb764b4fd) ^ 0x0764bea1.
- *
- * The source of this is definitely ASM:
- * - Minimal use of temporary registers
- * - The load of a1 for osRecvMesg and osSendMesg should be li, not addiu
- * - The load of a2 for osRecvMesg and osSendMesg use lui instead of li
+ * Wait for the PI to be idle, then read the osRomBase variable, use it to read
+ * the word at ROM offset 0xa5c and return it.
  */
 glabel mtxGetObfuscatedRomBase
 	addiu  $sp, $sp, -0x20
diff --git a/src/lib/mtxc.c b/src/lib/mtxc.c
new file mode 100644
index 000000000..1220b1cab
--- /dev/null
+++ b/src/lib/mtxc.c
@@ -0,0 +1,353 @@
+#include <ultra64.h>
+#include "constants.h"
+#include "bss.h"
+#include "data.h"
+#include "lib/mtx.h"
+#include "types.h"
+
+f32 var8005ef10[] = {65536, 65536};
+
+void mtx4LoadIdentity(Mtxf *mtx)
+{
+	mtx->m[0][0] = 1;
+	mtx->m[0][1] = 0;
+	mtx->m[0][2] = 0;
+	mtx->m[0][3] = 0;
+
+	mtx->m[1][0] = 0;
+	mtx->m[1][1] = 1;
+	mtx->m[1][2] = 0;
+	mtx->m[1][3] = 0;
+
+	mtx->m[2][0] = 0;
+	mtx->m[2][1] = 0;
+	mtx->m[2][2] = 1;
+	mtx->m[2][3] = 0;
+
+	mtx->m[3][0] = 0;
+	mtx->m[3][1] = 0;
+	mtx->m[3][2] = 0;
+	mtx->m[3][3] = 1;
+}
+
+void mtx4MultMtx4InPlace(Mtxf *multmtx, Mtxf *subject)
+{
+	mtx4MultMtx4(multmtx, subject, subject);
+}
+
+void mtx4MultMtx4(Mtxf *mtx1, Mtxf *mtx2, Mtxf *dst)
+{
+	s32 i;
+	f32 m00 = mtx2->m[0][0];
+	f32 m01 = mtx2->m[0][1];
+	f32 m02 = mtx2->m[0][2];
+	f32 m03 = mtx2->m[0][3];
+	f32 m10 = mtx2->m[1][0];
+	f32 m11 = mtx2->m[1][1];
+	f32 m12 = mtx2->m[1][2];
+	f32 m13 = mtx2->m[1][3];
+	f32 m20 = mtx2->m[2][0];
+	f32 m21 = mtx2->m[2][1];
+	f32 m22 = mtx2->m[2][2];
+	f32 m23 = mtx2->m[2][3];
+	f32 m30 = mtx2->m[3][0];
+	f32 m31 = mtx2->m[3][1];
+	f32 m32 = mtx2->m[3][2];
+	f32 m33 = mtx2->m[3][3];
+
+	for (i = 0; i < 4; i++) {
+		dst->m[0][i] = mtx1->m[0][i] * m00 + mtx1->m[1][i] * m01 + mtx1->m[2][i] * m02 + mtx1->m[3][i] * m03;
+		dst->m[1][i] = mtx1->m[0][i] * m10 + mtx1->m[1][i] * m11 + mtx1->m[2][i] * m12 + mtx1->m[3][i] * m13;
+		dst->m[2][i] = mtx1->m[0][i] * m20 + mtx1->m[1][i] * m21 + mtx1->m[2][i] * m22 + mtx1->m[3][i] * m23;
+		dst->m[3][i] = mtx1->m[0][i] * m30 + mtx1->m[1][i] * m31 + mtx1->m[2][i] * m32 + mtx1->m[3][i] * m33;
+	}
+}
+
+void mtx4RotateVecInPlace(Mtxf *mtx, struct coord *vec)
+{
+	mtx4RotateVec(mtx, vec, vec);
+}
+
+void mtx4RotateVec(Mtxf *mtx, struct coord *vec, struct coord *dst)
+{
+	f32 x = vec->x;
+	f32 y = vec->y;
+	f32 z = vec->z;
+
+	dst->x = mtx->m[0][0] * x + mtx->m[1][0] * y + mtx->m[2][0] * z;
+	dst->y = mtx->m[0][1] * x + mtx->m[1][1] * y + mtx->m[2][1] * z;
+	dst->z = mtx->m[0][2] * x + mtx->m[1][2] * y + mtx->m[2][2] * z;
+}
+
+void mtx4TransformVecInPlace(Mtxf *mtx, struct coord *vec)
+{
+	mtx4TransformVec(mtx, vec, vec);
+}
+
+void mtx4TransformVec(Mtxf *mtx, struct coord *vec, struct coord *dst)
+{
+	f32 x = vec->x;
+	f32 y = vec->y;
+	f32 z = vec->z;
+
+	dst->x = mtx->m[0][0] * x + mtx->m[1][0] * y + mtx->m[2][0] * z;
+	dst->y = mtx->m[0][1] * x + mtx->m[1][1] * y + mtx->m[2][1] * z;
+	dst->z = mtx->m[0][2] * x + mtx->m[1][2] * y + mtx->m[2][2] * z;
+
+	dst->x += mtx->m[3][0];
+	dst->y += mtx->m[3][1];
+	dst->z += mtx->m[3][2];
+}
+
+void mtx00015be0(Mtxf *matrix1, Mtxf *matrix2)
+{
+	mtx00015be4(matrix1, matrix2, matrix2);
+}
+
+void mtx00015be4(Mtxf *arg0, Mtxf *arg1, Mtxf *dst)
+{
+	f32 m00 = arg1->m[0][0];
+	f32 m01 = arg1->m[0][1];
+	f32 m02 = arg1->m[0][2];
+	f32 m03 = arg1->m[0][3];
+	f32 m10 = arg1->m[1][0];
+	f32 m11 = arg1->m[1][1];
+	f32 m12 = arg1->m[1][2];
+	f32 m13 = arg1->m[1][3];
+	f32 m20 = arg1->m[2][0];
+	f32 m21 = arg1->m[2][1];
+	f32 m22 = arg1->m[2][2];
+	f32 m23 = arg1->m[2][3];
+	f32 m30 = arg1->m[3][0];
+	f32 m31 = arg1->m[3][1];
+	f32 m32 = arg1->m[3][2];
+	f32 m33 = arg1->m[3][3];
+
+	dst->m[0][0] = arg0->m[0][0] * m00 + arg0->m[1][0] * m01 + arg0->m[2][0] * m02;
+	dst->m[0][1] = arg0->m[0][1] * m00 + arg0->m[1][1] * m01 + arg0->m[2][1] * m02;
+	dst->m[0][2] = arg0->m[0][2] * m00 + arg0->m[1][2] * m01 + arg0->m[2][2] * m02;
+	dst->m[0][3] = 0;
+
+	dst->m[1][0] = arg0->m[0][0] * m10 + arg0->m[1][0] * m11 + arg0->m[2][0] * m12;
+	dst->m[1][1] = arg0->m[0][1] * m10 + arg0->m[1][1] * m11 + arg0->m[2][1] * m12;
+	dst->m[1][2] = arg0->m[0][2] * m10 + arg0->m[1][2] * m11 + arg0->m[2][2] * m12;
+	dst->m[1][3] = 0;
+
+	dst->m[2][0] = arg0->m[0][0] * m20 + arg0->m[1][0] * m21 + arg0->m[2][0] * m22;
+	dst->m[2][1] = arg0->m[0][1] * m20 + arg0->m[1][1] * m21 + arg0->m[2][1] * m22;
+	dst->m[2][2] = arg0->m[0][2] * m20 + arg0->m[1][2] * m21 + arg0->m[2][2] * m22;
+	dst->m[2][3] = 0;
+
+	dst->m[3][0] = arg0->m[0][0] * m30 + arg0->m[1][0] * m31 + arg0->m[2][0] * m32 + arg0->m[3][0];
+	dst->m[3][1] = arg0->m[0][1] * m30 + arg0->m[1][1] * m31 + arg0->m[2][1] * m32 + arg0->m[3][1];
+	dst->m[3][2] = arg0->m[0][2] * m30 + arg0->m[1][2] * m31 + arg0->m[2][2] * m32 + arg0->m[3][2];
+	dst->m[3][3] = 1;
+}
+
+void mtx3Copy(f32 src[3][3], f32 dst[3][3])
+{
+	dst = src;
+}
+
+void mtx4Copy(Mtxf *src, Mtxf *dst)
+{
+	*dst = *src;
+}
+
+void mtx3ToMtx4(f32 src[3][3], Mtxf *dst)
+{
+	dst->m[0][0] = src[0][0];
+	dst->m[0][1] = src[0][1];
+	dst->m[0][2] = src[0][2];
+	dst->m[0][3] = 0;
+
+	dst->m[1][0] = src[1][0];
+	dst->m[1][1] = src[1][1];
+	dst->m[1][2] = src[1][2];
+	dst->m[1][3] = 0;
+
+	dst->m[2][0] = src[2][0];
+	dst->m[2][1] = src[2][1];
+	dst->m[2][2] = src[2][2];
+	dst->m[2][3] = 0;
+
+	dst->m[3][0] = 0;
+	dst->m[3][1] = 0;
+	dst->m[3][2] = 0;
+	dst->m[3][3] = 1;
+}
+
+void mtx4ToMtx3(Mtxf *src, f32 dst[3][3])
+{
+	dst[0][0] = src->m[0][0];
+	dst[0][1] = src->m[0][1];
+	dst[0][2] = src->m[0][2];
+
+	dst[1][0] = src->m[1][0];
+	dst[1][1] = src->m[1][1];
+	dst[1][2] = src->m[1][2];
+
+	dst[2][0] = src->m[2][0];
+	dst[2][1] = src->m[2][1];
+	dst[2][2] = src->m[2][2];
+}
+
+void mtx4SetTranslation(struct coord *pos, Mtxf *mtx)
+{
+	mtx->m[3][0] = pos->x;
+	mtx->m[3][1] = pos->y;
+	mtx->m[3][2] = pos->z;
+}
+
+void mtx00015df0(f32 mult, Mtxf *mtx)
+{
+	mtx->m[0][0] *= mult;
+	mtx->m[0][1] *= mult;
+	mtx->m[0][2] *= mult;
+	mtx->m[0][3] *= mult;
+}
+
+void mtx00015e24(f32 mult, Mtxf *mtx)
+{
+	mtx->m[0][0] *= mult;
+	mtx->m[0][1] *= mult;
+	mtx->m[0][2] *= mult;
+}
+
+void mtx00015e4c(f32 mult, Mtxf *mtx)
+{
+	mtx->m[1][0] *= mult;
+	mtx->m[1][1] *= mult;
+	mtx->m[1][2] *= mult;
+	mtx->m[1][3] *= mult;
+}
+
+void mtx00015e80(f32 mult, Mtxf *mtx)
+{
+	mtx->m[1][0] *= mult;
+	mtx->m[1][1] *= mult;
+	mtx->m[1][2] *= mult;
+}
+
+void mtx00015ea8(f32 mult, Mtxf *mtx)
+{
+	mtx->m[2][0] *= mult;
+	mtx->m[2][1] *= mult;
+	mtx->m[2][2] *= mult;
+	mtx->m[2][3] *= mult;
+}
+
+void mtx00015edc(f32 mult, Mtxf *mtx)
+{
+	mtx->m[2][0] *= mult;
+	mtx->m[2][1] *= mult;
+	mtx->m[2][2] *= mult;
+}
+
+void mtx00015f04(f32 mult, Mtxf *mtx)
+{
+	mtx->m[0][0] *= mult;
+	mtx->m[0][1] *= mult;
+	mtx->m[0][2] *= mult;
+	mtx->m[0][3] *= mult;
+
+	mtx->m[1][0] *= mult;
+	mtx->m[1][1] *= mult;
+	mtx->m[1][2] *= mult;
+	mtx->m[1][3] *= mult;
+
+	mtx->m[2][0] *= mult;
+	mtx->m[2][1] *= mult;
+	mtx->m[2][2] *= mult;
+	mtx->m[2][3] *= mult;
+}
+
+void mtx00015f4c(f32 mult, Mtxf *mtx)
+{
+	mtx->m[0][0] *= mult;
+	mtx->m[0][1] *= mult;
+	mtx->m[0][2] *= mult;
+
+	mtx->m[1][0] *= mult;
+	mtx->m[1][1] *= mult;
+	mtx->m[1][2] *= mult;
+
+	mtx->m[2][0] *= mult;
+	mtx->m[2][1] *= mult;
+	mtx->m[2][2] *= mult;
+}
+
+void mtx00015f88(f32 mult, Mtxf *mtx)
+{
+	mtx->m[0][0] *= mult;
+	mtx->m[0][1] *= mult;
+	mtx->m[0][2] *= mult;
+
+	mtx->m[1][0] *= mult;
+	mtx->m[1][1] *= mult;
+	mtx->m[1][2] *= mult;
+
+	mtx->m[2][0] *= mult;
+	mtx->m[2][1] *= mult;
+	mtx->m[2][2] *= mult;
+
+	mtx->m[3][0] *= mult;
+	mtx->m[3][1] *= mult;
+	mtx->m[3][2] *= mult;
+}
+
+u32 mtxGetObfuscatedRomBase(void)
+{
+	u32 value;
+
+	osRecvMesg(&__osPiAccessQueue, NULL, OS_MESG_BLOCK * 0x10000);
+
+	while (IO_READ(PI_STATUS_REG) & (PI_STATUS_DMA_BUSY | PI_STATUS_IO_BUSY));
+
+	// osRomBase is 0xb0000000
+	// load address is 0xb0000a5c
+	// value is 0x1740fff9
+	value = *(u32 *) (((u32) osRomBase | 0xb764b4fd) ^ 0x0764bea1);
+
+	osSendMesg(&__osPiAccessQueue, 0, 0);
+
+	return value;
+}
+
+void mtx00016054(Mtxf *src, Mtxf *dst)
+{
+	u32 src00 = (s32) (src->m[0][0] * var8005ef10[0]);
+	u32 src01 = (s32) (src->m[0][1] * var8005ef10[0]);
+	u32 src02 = (s32) (src->m[0][2] * var8005ef10[0]);
+	u32 src03 = (s32) (src->m[0][3] * var8005ef10[1]);
+	u32 src10 = (s32) (src->m[1][0] * var8005ef10[0]);
+	u32 src11 = (s32) (src->m[1][1] * var8005ef10[0]);
+	u32 src12 = (s32) (src->m[1][2] * var8005ef10[0]);
+	u32 src13 = (s32) (src->m[1][3] * var8005ef10[1]);
+	u32 src20 = (s32) (src->m[2][0] * var8005ef10[0]);
+	u32 src21 = (s32) (src->m[2][1] * var8005ef10[0]);
+	u32 src22 = (s32) (src->m[2][2] * var8005ef10[0]);
+	u32 src23 = (s32) (src->m[2][3] * var8005ef10[1]);
+	u32 src30 = (s32) (src->m[3][0] * var8005ef10[0]);
+	u32 src31 = (s32) (src->m[3][1] * var8005ef10[0]);
+	u32 src32 = (s32) (src->m[3][2] * var8005ef10[0]);
+	u32 src33 = (s32) (src->m[3][3] * var8005ef10[1]);
+
+	dst->l[0][0] = src00 & 0xffff0000 | src01 >> 16;
+	dst->l[0][1] = src02 & 0xffff0000 | src03 >> 16;
+	dst->l[0][2] = src10 & 0xffff0000 | src11 >> 16;
+	dst->l[0][3] = src12 & 0xffff0000 | src13 >> 16;
+	dst->l[1][0] = src20 & 0xffff0000 | src21 >> 16;
+	dst->l[1][1] = src22 & 0xffff0000 | src23 >> 16;
+	dst->l[1][2] = src30 & 0xffff0000 | src31 >> 16;
+	dst->l[1][3] = src32 & 0xffff0000 | src33 >> 16;
+
+	dst->l[2][0] = src00 << 16 | src01 & 0xffff;
+	dst->l[2][1] = src02 << 16 | src03 & 0xffff;
+	dst->l[2][2] = src10 << 16 | src11 & 0xffff;
+	dst->l[2][3] = src12 << 16 | src13 & 0xffff;
+	dst->l[3][0] = src20 << 16 | src21 & 0xffff;
+	dst->l[3][1] = src22 << 16 | src23 & 0xffff;
+	dst->l[3][2] = src30 << 16 | src31 & 0xffff;
+	dst->l[3][3] = src32 << 16 | src33 & 0xffff;
+}