From 2beab09f7da6ff61fc7f96b1863ff0e40ee8ef49 Mon Sep 17 00:00:00 2001
From: Madeline Busig <madeline.busig@outlook.com>
Date: Sun, 24 Mar 2024 00:30:56 -0600
Subject: [PATCH] Add armv4t assembly optimized division and modulo by 10

---
 include/armv4t/mtl/armv4t/asm/math.s | 30 ++++++++++++++++++++++++++++
 src/armv4t/math.s                    | 21 +++++++++++++++++++
 2 files changed, 51 insertions(+)
 create mode 100644 include/armv4t/mtl/armv4t/asm/math.s
 create mode 100644 src/armv4t/math.s

diff --git a/include/armv4t/mtl/armv4t/asm/math.s b/include/armv4t/mtl/armv4t/asm/math.s
new file mode 100644
index 0000000..2020215
--- /dev/null
+++ b/include/armv4t/mtl/armv4t/asm/math.s
@@ -0,0 +1,30 @@
+.syntax unified
+
+/*
+ * Calculates rx / 10 and places the result in rd. Clobbers the value of
+ * temporary register rt. The value in rx is unmodified.
+ *
+ * 0xCCCCCCCD >> 35 approximately equals 0.1
+ * Performs a 64 bit multiply of 0xCCCCCCCD and rx, shifts the high 32 bits
+ * by 3, and discards the low bits. This results in a division by 10 that
+ * works for all unsigned values of rx
+ */
+.macro udiv10 rd, rx, rt
+	ldr	\rt, =0xCCCCCCCD
+	umull	\rt, \rd, \rx, \rt
+	lsrs	\rd, $3
+.endm
+
+/*
+ * Calculates rx % 10 and places the result in rd. Clobbers the value of
+ * temporary register rt
+ *
+ * Calculates the modulo by calculating the truncated division by 10,
+ * multiplying by 10, and finding the difference between the original value.
+ */
+.macro umod10 rd, rx, rt
+	udiv10	\rd, \rx, \rt
+	mov	\rt, $10
+	mul	\rd, \rt
+	subs	\rd, \rx, \rd
+.endm
diff --git a/src/armv4t/math.s b/src/armv4t/math.s
new file mode 100644
index 0000000..b7d33d9
--- /dev/null
+++ b/src/armv4t/math.s
@@ -0,0 +1,21 @@
+.syntax unified
+
+.include "mtl/armv4t/asm/math.s"
+
+.section .iwram, "ax", %progbits
+.arm
+.align 2
+
+.global mtl_udiv10
+.type mtl_udiv10 STT_FUNC
+mtl_udiv10:
+	udiv10	r1, r0, r2
+	mov	r0, r1
+bx	lr
+
+.global mtl_umod10
+.type mtl_umod10 STT_FUNC
+mtl_umod10:
+	umod10	r1, r0, r2
+	mov	r0, r1
+bx	lr