implement upkr unpacker in wasm, use to load compressed base

This commit is contained in:
2021-11-20 23:36:16 +01:00
parent f7e3202c39
commit 93b2bb60bd
6 changed files with 149 additions and 19 deletions

View File

@@ -1,8 +1,8 @@
import "env.memory" memory(8); import "env.memory" memory(9);
export fn load_uw8(module_start: i32, module_end: i32, base_start: i32, base_end: i32) -> i32 { export fn load_uw8(module_start: i32, module_end: i32, base_start: i32, base_end: i32) -> i32 {
if ?module_start == 0 { if ?module_start == 0 {
let defer length = module_end - module_start; let lazy length = module_end - module_start;
copy(base_end, module_start, length); copy(base_end, module_start, length);
return base_end + length; return base_end + length;
} }
@@ -14,7 +14,7 @@ export fn load_uw8(module_start: i32, module_end: i32, base_start: i32, base_end
loop sections { loop sections {
if src < module_end & (base_start >= base_end | ?src <= ?base_start) { if src < module_end & (base_start >= base_end | ?src <= ?base_start) {
let defer length2 = copy_section(dest, src); let lazy length2 = copy_section(dest, src);
dest = dest + length2; dest = dest + length2;
if base_start < base_end & ?src == ?base_start { if base_start < base_end & ?src == ?base_start {
base_start = base_start + section_size(base_start); base_start = base_start + section_size(base_start);
@@ -24,7 +24,7 @@ export fn load_uw8(module_start: i32, module_end: i32, base_start: i32, base_end
} }
if base_start < base_end { if base_start < base_end {
let defer length3 = copy_section(dest, base_start); let lazy length3 = copy_section(dest, base_start);
dest = dest + length3; dest = dest + length3;
base_start = base_start + length3; base_start = base_start + length3;
branch sections; branch sections;
@@ -39,7 +39,7 @@ fn section_size(ptr: i32) -> i32 {
let l = 0; let l = 0;
let shift = 0; let shift = 0;
loop size { loop size {
let defer b = ?p; let lazy b = ?p;
l = l | ((b & 127) << shift); l = l | ((b & 127) << shift);
shift = shift + 7; shift = shift + 7;
p = p + 1; p = p + 1;
@@ -49,7 +49,7 @@ fn section_size(ptr: i32) -> i32 {
} }
fn copy_section(dest: i32, src: i32) -> i32 { fn copy_section(dest: i32, src: i32) -> i32 {
let defer length = section_size(src); let lazy length = section_size(src);
copy(dest, src, length); copy(dest, src, length);
length length
} }
@@ -62,3 +62,113 @@ fn copy(dest: i32, src: i32, len: i32) {
} }
} }
} }
// upkr unpacker
global mut upkr_src_ptr: i32 = 0;
global mut upkr_code: i64 = 0i64;
global mut upkr_low: i64 = 0i64;
global mut upkr_range: i64 = 0i64;
// uncompress upkr compressed data at `src` into the buffer at `dest`
// returns the end of the uncompressed data
export fn uncompress(src_ptr: i32, dest_ptr: i32) -> i32 {
upkr_src_ptr = src_ptr;
upkr_code = 0i64;
upkr_low = 0i64;
upkr_range = 1i64;
let offset: i32;
let byte: i32;
let i: i32;
loop init_contexts {
i!0x80000 = 0x8000;
branch_if (i := i + 4) < (256 + 1 + 128) * 4: init_contexts
}
block finished {
loop unpack_loop {
if upkr_bit(0) {
if upkr_bit(256) {
offset = upkr_length(257) - 1;
branch_if !offset: finished
}
let length = upkr_length(257 + 64);
loop copy {
dest_ptr?0 = (dest_ptr - offset)?0;
dest_ptr = dest_ptr + 1;
branch_if (length := length - 1): copy;
}
} else {
// literal
i = 0;
byte = 1;
loop literal {
byte = (byte << 1) | upkr_bit(byte);
branch_if (i := i + 1) < 8: literal;
}
dest_ptr?0 = byte;
dest_ptr = dest_ptr + 1;
}
branch unpack_loop;
}
}
dest_ptr
}
fn upkr_length(context_index: i32) -> i32 {
let length: i32;
let bit_pos: i32;
loop bits {
if upkr_bit(context_index) {
length = length | (upkr_bit(context_index + 1) << bit_pos);
context_index = context_index + 2;
bit_pos = bit_pos + 1;
branch bits;
}
}
length | (1 << bit_pos)
}
fn upkr_bit(context_index: i32) -> i32 {
let prob = ((context_index * 4)!0x80000) as i64;
loop refill {
if upkr_low >> 32i64 == (upkr_low + upkr_range - 1i64) >> 32i64 {
upkr_append_byte();
branch refill;
}
}
if upkr_range < (1i64 << 24i64) {
upkr_append_byte();
upkr_append_byte();
upkr_range = (1i64 << 40i64) - upkr_low;
}
let range = upkr_range / 65536i64;
let bit = (upkr_code - upkr_low) / range < prob;
if bit {
upkr_range = range * prob;
prob = prob + (((1i64 << 16i64) - prob) >> 4i64);
} else {
upkr_low = upkr_low + range * prob;
upkr_range = range * (65536i64 - prob);
prob = prob - (prob >> 4i64);
}
(context_index * 4)!0x80000 = prob as i32;
bit
}
fn upkr_append_byte() {
upkr_code = ((upkr_code & i64.extend_i32_u(-1)) << 8i64) | (?upkr_src_ptr) as i64;
upkr_src_ptr = upkr_src_ptr + 1;
upkr_low = (upkr_low & i64.extend_i32_u(-1)) << 8i64;
upkr_range = upkr_range << 8i64;
}

View File

@@ -178,12 +178,20 @@ impl Loader {
fn load(&mut self, module_data: &[u8]) -> Result<Vec<u8>> { fn load(&mut self, module_data: &[u8]) -> Result<Vec<u8>> {
let memory = self.memory.data_mut(&mut self.store); let memory = self.memory.data_mut(&mut self.store);
let compressed_base_module = include_bytes!("../uw8-tool/base.upk");
memory[..compressed_base_module.len()].copy_from_slice(compressed_base_module);
let base_end = self.instance.get_typed_func::<(i32, i32), i32, _>(&mut self.store, "uncompress")?.call(&mut self.store, (0, 0x84000))? as u32 as usize;
let memory = self.memory.data_mut(&mut self.store);
let base_module = memory[0x84000..base_end].to_vec();
let base_start = module_data.len(); let base_start = module_data.len();
memory[..base_start].copy_from_slice(module_data); memory[..base_start].copy_from_slice(module_data);
let base_module = include_bytes!("../uw8-tool/base1.wasm");
let base_end = base_start + base_module.len(); let base_end = base_start + base_module.len();
memory[base_start..base_end].copy_from_slice(base_module); memory[base_start..base_end].copy_from_slice(&base_module);
let load_uw8 = self let load_uw8 = self
.instance .instance

View File

@@ -192,6 +192,13 @@ impl BaseModule {
File::create(path)?.write_all(&self.to_wasm())?; File::create(path)?.write_all(&self.to_wasm())?;
Ok(()) Ok(())
} }
pub fn create_binary(path: &Path) -> Result<()> {
let base1 = BaseModule::for_format_version(1)?.to_wasm();
let data = upkr::pack(&base1);
File::create(path)?.write_all(&data)?;
Ok(())
}
} }
fn add_function( fn add_function(

5
uw8-tool/src/lib.rs Normal file
View File

@@ -0,0 +1,5 @@
mod base_module;
pub mod pack;
pub use base_module::BaseModule;
pub use pack::{pack_file, unpack, unpack_file};

View File

@@ -1,10 +1,7 @@
mod base_module;
mod pack;
use std::path::PathBuf; use std::path::PathBuf;
use anyhow::Result; use anyhow::Result;
use base_module::BaseModule; use uw8_tool::BaseModule;
use pico_args::Arguments; use pico_args::Arguments;
fn main() -> Result<()> { fn main() -> Result<()> {
@@ -13,20 +10,19 @@ fn main() -> Result<()> {
if let Some(cmd) = args.subcommand()? { if let Some(cmd) = args.subcommand()? {
match cmd.as_str() { match cmd.as_str() {
"make-base" => { "make-base" => {
let version: u8 = args.free_from_str()?; let path: PathBuf = args.free_from_str()?;
BaseModule::for_format_version(version)? BaseModule::create_binary(&path)?;
.write_to_file(format!("base{}.wasm", version))?;
} }
"pack" => { "pack" => {
let version: u8 = args.opt_value_from_str(["-v", "--version"])?.unwrap_or(1); let version: u8 = args.opt_value_from_str(["-v", "--version"])?.unwrap_or(1);
let source: PathBuf = args.free_from_str()?; let source: PathBuf = args.free_from_str()?;
let dest: PathBuf = args.free_from_str()?; let dest: PathBuf = args.free_from_str()?;
pack::pack_file(&source, &dest, version)?; uw8_tool::pack_file(&source, &dest, version)?;
} }
"unpack" => { "unpack" => {
let source: PathBuf = args.free_from_str()?; let source: PathBuf = args.free_from_str()?;
let dest: PathBuf = args.free_from_str()?; let dest: PathBuf = args.free_from_str()?;
pack::unpack_file(&source, &dest)?; uw8_tool::unpack_file(&source, &dest)?;
} }
_ => { _ => {
eprintln!("Unknown subcommand '{}'", cmd); eprintln!("Unknown subcommand '{}'", cmd);

View File

@@ -1,6 +1,6 @@
import loaderUrl from "data-url:../../platform/loader.wasm"; import loaderUrl from "data-url:../../platform/loader.wasm";
import platformUrl from "data-url:../../platform/platform.wasm"; import platformUrl from "data-url:../../platform/platform.wasm";
import baseUrl from "data-url:../../uw8-tool/base1.wasm"; import baseUrl from "data-url:../../uw8-tool/base.upk";
async function loadWasm(url, imports) { async function loadWasm(url, imports) {
let wasm_module = await (await fetch(url)).arrayBuffer(); let wasm_module = await (await fetch(url)).arrayBuffer();
@@ -66,9 +66,13 @@ async function runModule(data) {
let loadMem = loaderImport.env.memory.buffer; let loadMem = loaderImport.env.memory.buffer;
let loader = await loadWasm(loaderUrl, loaderImport); let loader = await loadWasm(loaderUrl, loaderImport);
let baseModule = await (await fetch(baseUrl)).arrayBuffer(); let packedBaseModule = await (await fetch(baseUrl)).arrayBuffer();
if (dataU8Array[0] != 0) { if (dataU8Array[0] != 0) {
new Uint8Array(loadMem).set(new Uint8Array(packedBaseModule));
let baseEnd = loader.exports.uncompress(0, 0x84000);
let baseModule = loadMem.slice(0x84000, baseEnd);
new Uint8Array(loadMem).set(dataU8Array); new Uint8Array(loadMem).set(dataU8Array);
new Uint8Array(loadMem).set(new Uint8Array(baseModule), data.byteLength); new Uint8Array(loadMem).set(new Uint8Array(baseModule), data.byteLength);