simple dev setup for asm unpackers

asm_unpackers/.gitignore

@@ -0,0 +1 @@
+/build/

asm_unpackers/Makefile

@@ -0,0 +1,22 @@
+build/unpack_riscv64: ../c_unpacker/main.c ../c_unpacker/unpack.c
+	mkdir -p build
+	riscv64-linux-gnu-gcc -g -static -o $@ $^
+
+test_riscv64: build/unpack_riscv64
+	qemu-riscv64 $< test_data.upk /tmp/out.bin
+	cmp test_data.bin /tmp/out.bin
+
+build/unpack_armv6m: ../c_unpacker/main.c ../c_unpacker/unpack.c
+	mkdir -p build
+	arm-linux-gnueabihf-gcc -g -static -o $@ $?
+
+test_armv6m: build/unpack_armv6m
+	qemu-arm $< test_data.upk /tmp/out.bin
+	cmp test_data.bin /tmp/out.bin
+
+build/unpack_armv6m.bin: unpack_armv6m.S
+	arm-none-eabi-gcc -march=armv6-m -c -o build/unpack_armv6m.o $?
+	arm-none-eabi-objcopy -O binary --only-section=.text build/unpack_armv6m.o $@
+
+sizes: build/unpack_armv6m.bin
+	ls -l build/*.bin

asm_unpackers/test_data.bin

@@ -0,0 +1,99 @@
+typedef unsigned char u8;
+typedef unsigned short u16;
+typedef unsigned long u32;
+
+u8* upkr_data_ptr;
+u8 upkr_probs[1 + 255 + 1 + 2*32 + 2*32]; 
+#ifdef UPKR_BITSTREAM
+u16 upkr_state;
+u8 upkr_current_byte;
+int upkr_bits_left;
+#else
+u32 upkr_state;
+#endif
+
+int upkr_decode_bit(int context_index) {
+#ifdef UPKR_BITSTREAM
+    while(upkr_state < 32768) {
+        if(upkr_bits_left == 0) {
+            upkr_current_byte = *upkr_data_ptr++;
+            upkr_bits_left = 8;
+        }
+        upkr_state = (upkr_state << 1) + (upkr_current_byte & 1);
+        upkr_current_byte >>= 1;
+        --upkr_bits_left;
+    }
+#else
+    while(upkr_state < 4096) {
+        upkr_state = (upkr_state << 8) | *upkr_data_ptr++;
+    }
+#endif
+   
+    int prob = upkr_probs[context_index];
+    int bit = (upkr_state & 255) < prob ? 1 : 0;
+    
+    int tmp = prob;
+    if(!bit) {
+        tmp = 256 - tmp;
+    }
+    upkr_state = tmp * (upkr_state >> 8) + (upkr_state & 255);
+    tmp += (256 - tmp + 8) >> 4;
+    if(!bit) {
+        upkr_state -= prob;
+        tmp = 256 - tmp;
+    }
+    upkr_probs[context_index] = tmp;
+
+    return bit;
+}
+
+int upkr_decode_length(int context_index) {
+    int length = 0;
+    int bit_pos = 0;
+    while(upkr_decode_bit(context_index)) {
+        length |= upkr_decode_bit(context_index + 1) << bit_pos++;
+        context_index += 2;
+    }
+    return length | (1 << bit_pos);
+}
+
+void* upkr_unpack(void* destination, void* compressed_data) {
+    upkr_data_ptr = (u8*)compressed_data;
+    upkr_state = 0;
+#ifdef UPKR_BITSTREAM
+    upkr_bits_left = 0;
+#endif
+    for(int i = 0; i < sizeof(upkr_probs); ++i)
+        upkr_probs[i] = 128;
+    
+    u8* write_ptr = (u8*)destination;
+    
+    int prev_was_match = 0;
+    int offset = 0;
+    for(;;) {
+        if(upkr_decode_bit(0)) {
+            if(prev_was_match || upkr_decode_bit(256)) {
+                offset = upkr_decode_length(257) - 1;
+                if(offset == 0) {
+                    break;
+                }
+            }
+            int length = upkr_decode_length(257 + 64);
+            while(length--) {
+                *write_ptr = write_ptr[-offset];
+                ++write_ptr;
+            }
+            prev_was_match = 1;
+        } else {
+            int byte = 1;
+            while(byte < 256) {
+                int bit = upkr_decode_bit(byte);
+                byte = (byte << 1) + bit;
+            }
+            *write_ptr++ = byte;
+            prev_was_match = 0;
+        }
+    }
+    
+    return write_ptr;
+}

asm_unpackers/unpack_armv6m.S

@@ -0,0 +1,172 @@
+.syntax unified
+.thumb
+
+.section .text
+
+#define ALIGNUP(n, align) (((n) + (align) - 1) & ~((align) - 1))
+#define PROB_LEN (1 + 255 + 1 + 2*32 + 2*32)
+#define DATA_OFF ALIGNUP(PROB_LEN, 4)
+
+// auto upkr_unpack(uint8_t * out, uint8_t * in) -> tuple<uint8_t *, uint8_t *>
+.global upkr_unpack
+.type upkr_unpack, %function
+// r0 .. out_ptr (returned)
+// r1 .. in_ptr (returned)
+// r2 .. state
+// r3 .. offset
+// r4 .. prev_was_match
+// r5 .. subroutine arg (preserved)
+// r6 .. subroutine ret
+// r7 .. probs ptr
+upkr_unpack:
+    push { r4, r5, r6, r7, lr }
+    sub sp, sp, #DATA_OFF
+
+    mov r7, sp
+    movs r2, #255
+    adds r2, r2, #(PROB_LEN - 255)
+    movs r3, #128
+.Lclear:
+    subs r2, r2, #1
+    strb r3, [r7, r2]
+    bne .Lclear
+
+.Lloop:
+    movs r5, #0
+    bl upkr_decode_bit
+    beq .Ldata
+.Lmatch:
+    // r6 = 1
+    lsls r5, r6, #8
+    cmp r4, #0
+    mov r4, r6 // = 1
+    bne 1f
+
+    bl upkr_decode_bit
+    beq 2f
+
+1:
+    adds r5, r5, #1
+    bl upkr_decode_length
+    subs r5, r5, #1
+    // r4 = 1
+    subs r3, r4, r6
+    beq .Lend
+2:
+
+    adds r5, r5, #65
+    bl upkr_decode_length
+.Lcopy_loop:
+    ldrb r5, [r0, r3]
+    strb r5, [r0]
+    adds r0, r0, #1
+    subs r6, r6, #1
+    bne .Lcopy_loop
+    b .Lloop
+
+.Ldata:
+    movs r4, #0
+    movs r5, #1
+
+.Ldata_loop:
+    bl upkr_decode_bit
+    lsls r5, r5, #1
+    adds r5, r5, r6
+    lsrs r6, r5, #8
+    beq .Ldata_loop
+
+    strb r5, [r0]
+    adds r0, r0, #1
+    b .Lloop
+
+.Lend:
+    add sp, sp, #DATA_OFF
+    pop { r4, r5, r6, r7, pc }
+
+.type upkr_decode_length, %function
+// r0 .. length tmp (saved)
+// r1 ..
+// r2 ..
+// r3 ..
+// r4 .. bit pos (saved)
+// r5 .. content index (saved)
+// r6 .. length (returned)
+// r7 ..
+upkr_decode_length:
+    push { r0, r4, r5, lr }
+
+    movs r4, #0
+    movs r0, #0
+.Lbit_loop:
+    bl upkr_decode_bit
+    beq 1f
+
+    adds r5, r5, #1
+    bl upkr_decode_bit
+    adds r5, r5, #1
+    lsls r6, r6, r4
+    adds r4, r4, #1
+    orrs r0, r0, r6
+    b .Lbit_loop
+1:
+    movs r6, #1
+    lsls r6, r6, r4
+    orrs r6, r6, r0
+
+    pop { r0, r4, r5, pc }
+
+.type upkr_decode_bit, %function
+// r0 .. tmp / prob (saved)
+// r1 .. out_ptr (modified)
+// r2 .. state (modified)
+// r3 .. scratch (saved)
+// r4 ..
+// r5 .. content index (preserved)
+// r6 .. bit (returned)
+// r7 .. probs ptr (preserved)
+upkr_decode_bit:
+    push { r0, r3, lr }
+
+.Lstate_loop:
+    lsrs r3, r2, #12
+    bne 1f
+    lsls r2, r2, #8
+    ldrb r6, [r1]
+    adds r1, r1, #1
+    orrs r2, r2, r6
+    b .Lstate_loop
+1:
+
+    ldrb r0, [r7, r5]
+
+    lsrs r3, r2, #8
+    uxtb r2, r2
+
+    subs r6, r2, r0
+    lsrs r6, r6, #31
+    bne 1f
+
+    subs r2, r2, r0
+    rsbs r0, r0, #0
+    uxtb r0, r0
+1:
+
+    muls r3, r3, r0
+    adds r2, r2, r3
+
+    rsbs r3, r0, #0
+    uxtb r3, r3
+    adds r3, r3, #8
+    lsrs r3, r3, #4
+    adds r0, r0, r3
+
+    cmp r6, #0
+    bne 1f
+
+    rsbs r0, r0, #0
+1:
+
+    strb r0, [r7, r5]
+
+    cmp r6, #0
+    pop { r0, r3, pc }