changes

Makefile

@@ -14,6 +14,7 @@ ASM_ARM64 := src/kernels/fossbench-arm64.S
 ASM_AMD64 := src/kernels/fossbench-amd64.S
 ASM_I386  := src/kernels/fossbench-i386.S
 ASM_PPC32 := src/kernels/fossbench-ppc32be.S
+SRC_PPC32 := src/kernels/fossbench-ppc32be-chacha.c
 ASM_PPC64 := src/kernels/fossbench-ppc64be.S
 ASM_PPC32LE := src/kernels/fossbench-ppc32le.S
 ASM_PPC64LE := src/kernels/fossbench-ppc64le.S
@@ -37,7 +38,7 @@ else ifneq (,$(filter ppcle powerpcle ppc32le powerpc32le,$(HOST_ARCH)))
 	HOST_KERNEL   := $(ASM_PPC32LE)
 else ifneq (,$(filter ppc powerpc ppc32 powerpc32,$(HOST_ARCH)))
 	HOST_ARCHNAME := ppc32be
-	HOST_KERNEL   := $(ASM_PPC32)
+	HOST_KERNEL   := $(ASM_PPC32) $(SRC_PPC32)
 else ifneq (,$(filter ppc64 powerpc64,$(HOST_ARCH)))
 	HOST_ARCHNAME := ppc64be
 	HOST_KERNEL   := $(ASM_PPC64)
@@ -154,8 +155,8 @@ $(DIST)/fossbench-linux-i386: $(DRIVER) $(DRIVER_DEPS) $(ASM_I386) | $(DIST)
 	$(CC_I386) -m32 -march=pentium4 -fno-pie -no-pie $(CFLAGS) $(PTHREAD) $(LDFLAGS) -o $@ $(DRIVER) $(ASM_I386) $(LDLIBS)
 	@echo "built $@"
 
-$(DIST)/fossbench-linux-ppc32be: $(DRIVER) $(DRIVER_DEPS) $(ASM_PPC32) | $(DIST)
-	$(CC_PPC32BE) $(CFLAGS) $(PTHREAD) $(LDFLAGS) -o $@ $(DRIVER) $(ASM_PPC32) $(LDLIBS)
+$(DIST)/fossbench-linux-ppc32be: $(DRIVER) $(DRIVER_DEPS) $(ASM_PPC32) $(SRC_PPC32) | $(DIST)
+	$(CC_PPC32BE) $(CFLAGS) $(PTHREAD) $(LDFLAGS) -o $@ $(DRIVER) $(ASM_PPC32) $(SRC_PPC32) $(LDLIBS)
 	@echo "built $@"
 
 $(DIST)/fossbench-linux-ppc64be: $(DRIVER) $(DRIVER_DEPS) $(ASM_PPC64) | $(DIST)

src/kernels/fossbench-ppc32be-chacha.c

@@ -0,0 +1,87 @@
+/*
+ * Baseline 32-bit PowerPC ChaCha20 kernel.
+ *
+ * The Wii's Broadway CPU has no AltiVec, so this target must not use the
+ * vectorized implementation emitted for 74xx PowerPC processors.  Keep the
+ * byte loads and stores explicit: ChaCha20 words are little-endian while this
+ * target is big-endian.
+ */
+
+#include <stdint.h>
+
+static uint32_t load32le(const uint8_t *p)
+{
+	return (uint32_t)p[0] | (uint32_t)p[1] << 8 |
+	       (uint32_t)p[2] << 16 | (uint32_t)p[3] << 24;
+}
+
+static void store32le(uint8_t *p, uint32_t x)
+{
+	p[0] = (uint8_t)x;
+	p[1] = (uint8_t)(x >> 8);
+	p[2] = (uint8_t)(x >> 16);
+	p[3] = (uint8_t)(x >> 24);
+}
+
+static uint32_t rotl32(uint32_t x, unsigned int n)
+{
+	return (x << n) | (x >> (32 - n));
+}
+
+#define QR(a, b, c, d) do {                 \
+	x[a] += x[b]; x[d] = rotl32(x[d] ^ x[a], 16); \
+	x[c] += x[d]; x[b] = rotl32(x[b] ^ x[c], 12); \
+	x[a] += x[b]; x[d] = rotl32(x[d] ^ x[a], 8);  \
+	x[c] += x[d]; x[b] = rotl32(x[b] ^ x[c], 7);  \
+} while (0)
+
+uint64_t fb_chacha20(uint8_t *buf, uint64_t len,
+		     const uint8_t key[32], uint64_t rounds)
+{
+	static const uint32_t sigma[4] = {
+		0x61707865, 0x3320646e, 0x79622d32, 0x6b206574
+	};
+	uint32_t base[16] = {0}, x[16];
+	uint32_t checksum = 0, counter = 0;
+	uint64_t pass, off;
+	int i, r;
+
+	len &= ~(uint64_t)63;
+	if (len == 0 || rounds == 0)
+		return 0;
+
+	for (i = 0; i < 4; i++)
+		base[i] = sigma[i];
+	for (i = 0; i < 8; i++)
+		base[4 + i] = load32le(key + 4 * i);
+
+	for (pass = 0; pass < rounds; pass++) {
+		for (off = 0; off < len; off += 64, counter++) {
+			base[12] = counter;
+			for (i = 0; i < 16; i++)
+				x[i] = base[i];
+
+			for (r = 0; r < 10; r++) {
+				QR(0, 4, 8, 12); QR(1, 5, 9, 13);
+				QR(2, 6, 10, 14); QR(3, 7, 11, 15);
+				QR(0, 5, 10, 15); QR(1, 6, 11, 12);
+				QR(2, 7, 8, 13); QR(3, 4, 9, 14);
+			}
+
+			for (i = 0; i < 16; i++) {
+				uint32_t word = x[i] + base[i];
+				uint8_t stream[4];
+				int j;
+
+				store32le(stream, word);
+				for (j = 0; j < 4; j++)
+					buf[off + 4 * i + j] ^= stream[j];
+				checksum ^= word;
+			}
+		}
+	}
+
+	return checksum;
+}
+
+#undef QR

src/kernels/fossbench-ppc32be.S

@@ -1121,11 +1121,11 @@ fb_compress:
         .long .LC46
         .section        ".text"
         .align 2
-        .globl fb_chacha20
+        .globl fb_chacha20_altivec
 .LCL5:
         .long .LCTOC1-.LCF5
-        .type   fb_chacha20, @function
-fb_chacha20:
+        .type   fb_chacha20_altivec, @function
+fb_chacha20_altivec:
         .cfi_startproc
         stwu 1,-608(1)
         .cfi_def_cfa_offset 608
@@ -1966,7 +1966,7 @@ fb_chacha20:
         stb 9,63(23)
         b .L133
         .cfi_endproc
-        .size   fb_chacha20, .-fb_chacha20
+        .size   fb_chacha20_altivec, .-fb_chacha20_altivec
         .section        ".got2","aw"
 .LC50:
         .long .LC49