chaning css for no reason

Reviewed-on: https://git.ablecorp.us/AbleOS/holey-bytes/pulls/24

.cargo/config.toml

@@ -0,0 +1,4 @@
+[alias]
+xtask = "r -p xtask --"
+wasm-build = "b --target wasm32-unknown-unknown --profile=small -Zbuild-std=core,alloc -Zbuild-std-features=optimize_for_size,panic_immediate_abort -p"
+wasm-build-debug = "b --target wasm32-unknown-unknown --profile=small-dev -Zbuild-std=core,alloc -Zbuild-std-features=optimize_for_size -p"

.gitignore

@@ -1 +1,13 @@
+# garbage
 /target
+rustc-ice-*
+
+# sqlite
+db.sqlite
+db.sqlite-journal
+
+# assets
+/depell/src/*.gz
+/depell/src/*.wasm
+#**/*-sv.rs
+/bytecode/src/instrs.rs

Cargo.toml

@@ -1,2 +1,50 @@
+cargo-features = ["profile-rustflags"]
+
 [workspace]
-members = ["hbasm", "hbbytecode", "hbvm"]
+resolver = "2"
+members = [
+	"bytecode",
+	"vm",
+	"xrt",
+	"xtask",
+	"lang",
+	"depell",
+	"depell/wasm-fmt",
+	"depell/wasm-hbc",
+	"depell/wasm-rt",
+]
+
+[workspace.dependencies]
+hbbytecode = { path = "bytecode", default-features = false }
+hbvm = { path = "vm", default-features = false }
+hbxrt = { path = "xrt" }
+hblang = { path = "lang", default-features = false }
+hbjit = { path = "jit" }
+
+[profile.release]
+lto = true
+#debug = true
+strip = true
+codegen-units = 1
+panic = "abort"
+
+[profile.small]
+rustflags = ["-Zfmt-debug=none", "-Zlocation-detail=none"]
+inherits = "release"
+opt-level = "z"
+strip = "debuginfo"
+lto = true
+codegen-units = 1
+panic = "abort"
+
+[profile.small-dev]
+inherits = "dev"
+opt-level = "z"
+strip = "debuginfo"
+panic = "abort"
+
+[profile.fuzz]
+inherits = "dev"
+debug = true
+opt-level = 3
+panic = "abort"

bytecode/Cargo.toml

@@ -0,0 +1,10 @@
+[package]
+name = "hbbytecode"
+version = "0.1.0"
+edition = "2018"
+
+[features]
+default = ["disasm"]
+std = []
+disasm = ["std"]
+

bytecode/build.rs

@@ -0,0 +1,204 @@
+#![feature(iter_next_chunk)]
+
+use std::{collections::HashSet, fmt::Write};
+
+fn main() -> Result<(), Box<dyn std::error::Error>> {
+    println!("cargo:rerun-if-changed=build.rs");
+    println!("cargo:rerun-if-changed=instructions.in");
+
+    let mut generated = String::new();
+    gen_instrs(&mut generated)?;
+    std::fs::write("src/instrs.rs", generated)?;
+
+    Ok(())
+}
+
+fn gen_instrs(generated: &mut String) -> Result<(), Box<dyn std::error::Error>> {
+    writeln!(generated, "#![expect(dead_code)]")?;
+    writeln!(generated, "use crate::*;")?;
+
+    '_opcode_structs: {
+        let mut seen = HashSet::new();
+        for [.., args, _] in instructions() {
+            if !seen.insert(args) {
+                continue;
+            }
+
+            writeln!(generated, "#[derive(Clone, Copy, Debug)]")?;
+            writeln!(generated, "#[repr(packed)]")?;
+            write!(generated, "pub struct Ops{args}(")?;
+            let mut first = true;
+            for ch in args.chars().filter(|&ch| ch != 'N') {
+                if !std::mem::take(&mut first) {
+                    write!(generated, ",")?;
+                }
+                write!(generated, "pub Op{ch}")?;
+            }
+            writeln!(generated, ");")?;
+            writeln!(generated, "unsafe impl BytecodeItem for Ops{args} {{}}")?;
+        }
+    }
+
+    '_max_size: {
+        let max = instructions()
+            .map(
+                |[_, _, ty, _]| {
+                    if ty == "N" {
+                        1
+                    } else {
+                        iter_args(ty).map(arg_to_width).sum::<usize>() + 1
+                    }
+                },
+            )
+            .max()
+            .unwrap();
+
+        writeln!(generated, "pub const MAX_SIZE: usize = {max};")?;
+    }
+
+    '_encoders: {
+        for [op, name, ty, doc] in instructions() {
+            writeln!(generated, "/// {}", doc.trim_matches('"'))?;
+            let name = name.to_lowercase();
+            let args = comma_sep(
+                iter_args(ty)
+                    .enumerate()
+                    .map(|(i, c)| format!("{}{i}: {}", arg_to_name(c), arg_to_type(c))),
+            );
+            writeln!(generated, "pub fn {name}({args}) -> (usize, [u8; MAX_SIZE]) {{")?;
+            let arg_names =
+                comma_sep(iter_args(ty).enumerate().map(|(i, c)| format!("{}{i}", arg_to_name(c))));
+            writeln!(generated, "    unsafe {{ crate::encode({ty}({op}, {arg_names})) }}")?;
+            writeln!(generated, "}}")?;
+        }
+    }
+
+    '_structs: {
+        let mut seen = std::collections::HashSet::new();
+        for [_, _, ty, _] in instructions() {
+            if !seen.insert(ty) {
+                continue;
+            }
+            let types = comma_sep(iter_args(ty).map(arg_to_type).map(|s| s.to_string()));
+            writeln!(generated, "#[repr(packed)] pub struct {ty}(u8, {types});")?;
+        }
+    }
+
+    '_name_list: {
+        writeln!(generated, "pub const COUNT: u8 = {};", instructions().count())?;
+    }
+
+    let instr = "Instr";
+    let oper = "Oper";
+
+    '_instr_enum: {
+        writeln!(generated, "#[derive(Debug, Clone, Copy, PartialEq, Eq)] #[repr(u8)]")?;
+        writeln!(generated, "pub enum {instr} {{")?;
+        for [id, name, ..] in instructions() {
+            writeln!(generated, "    {name} = {id},")?;
+        }
+        writeln!(generated, "}}")?;
+    }
+
+    '_arg_kind: {
+        writeln!(generated, "#[derive(Debug, Clone, Copy, PartialEq, Eq)]")?;
+        writeln!(generated, "pub enum {oper} {{")?;
+        let mut seen = HashSet::new();
+        for ty in instructions().flat_map(|[.., ty, _]| iter_args(ty)) {
+            if !seen.insert(ty) {
+                continue;
+            }
+            writeln!(generated, "    {ty}({}),", arg_to_type(ty))?;
+        }
+        writeln!(generated, "}}")?;
+    }
+
+    '_parse_opers: {
+        writeln!(
+            generated,
+            "/// This assumes the instruction byte is still at the beginning of the buffer"
+        )?;
+        writeln!(generated, "#[cfg(feature = \"disasm\")]")?;
+        writeln!(generated, "pub fn parse_args(bytes: &mut &[u8], kind: {instr}, buf: &mut alloc::vec::Vec<{oper}>) -> Option<()> {{")?;
+        writeln!(generated, "    match kind {{")?;
+        let mut instrs = instructions().collect::<Vec<_>>();
+        instrs.sort_unstable_by_key(|&[.., ty, _]| ty);
+        for group in instrs.chunk_by(|[.., a, _], [.., b, _]| a == b) {
+            let ty = group[0][2];
+            for &[_, name, ..] in group {
+                writeln!(generated, "        | {instr}::{name}")?;
+            }
+            generated.pop();
+            writeln!(generated, " => {{")?;
+            if iter_args(ty).count() != 0 {
+                writeln!(generated, "            let data = crate::decode::<{ty}>(bytes)?;")?;
+                writeln!(
+                    generated,
+                    "            buf.extend([{}]);",
+                    comma_sep(
+                        iter_args(ty).zip(1u32..).map(|(t, i)| format!("{oper}::{t}(data.{i})"))
+                    )
+                )?;
+            } else {
+                writeln!(generated, "            crate::decode::<{ty}>(bytes)?;")?;
+            }
+
+            writeln!(generated, "        }}")?;
+        }
+        writeln!(generated, "    }}")?;
+        writeln!(generated, "    Some(())")?;
+        writeln!(generated, "}}")?;
+    }
+
+    std::fs::write("src/instrs.rs", generated)?;
+    Ok(())
+}
+
+fn comma_sep(items: impl Iterator<Item = String>) -> String {
+    items.map(|item| item.to_string()).collect::<Vec<_>>().join(", ")
+}
+
+fn instructions() -> impl Iterator<Item = [&'static str; 4]> {
+    include_str!("instructions.in")
+        .lines()
+        .filter_map(|line| line.strip_suffix(';'))
+        .map(|line| line.splitn(4, ',').map(str::trim).next_chunk().unwrap())
+}
+
+fn arg_to_type(arg: char) -> &'static str {
+    match arg {
+        'R' | 'B' => "u8",
+        'H' => "u16",
+        'W' => "u32",
+        'D' | 'A' => "u64",
+        'P' => "i16",
+        'O' => "i32",
+        _ => panic!("unknown type: {}", arg),
+    }
+}
+
+fn arg_to_width(arg: char) -> usize {
+    match arg {
+        'R' | 'B' => 1,
+        'H' => 2,
+        'W' => 4,
+        'D' | 'A' => 8,
+        'P' => 2,
+        'O' => 4,
+        _ => panic!("unknown type: {}", arg),
+    }
+}
+
+fn arg_to_name(arg: char) -> &'static str {
+    match arg {
+        'R' => "reg",
+        'B' | 'H' | 'W' | 'D' => "imm",
+        'P' | 'O' => "offset",
+        'A' => "addr",
+        _ => panic!("unknown type: {}", arg),
+    }
+}
+
+fn iter_args(ty: &'static str) -> impl Iterator<Item = char> {
+    ty.chars().filter(|c| *c != 'N')
+}

bytecode/hbbytecode.h

@@ -0,0 +1,75 @@
+/* HoleyBytes Bytecode representation in C
+ * Requires C23 compiler or better
+ *
+ * Uses MSVC pack pragma extension,
+ * proved to work with Clang and GNU® GCC™.
+ */
+
+#pragma once
+#include <assert.h>
+#include <limits.h>
+#include <stdint.h>
+
+static_assert(CHAR_BIT == 8, "Cursed architectures are not supported");
+
+enum hbbc_Opcode: uint8_t {
+    hbbc_Op_UN    , hbbc_Op_TX   , hbbc_Op_NOP   , hbbc_Op_ADD   , hbbc_Op_SUB   , hbbc_Op_MUL   ,
+    hbbc_Op_AND   , hbbc_Op_OR   , hbbc_Op_XOR   , hbbc_Op_SL    , hbbc_Op_SR    , hbbc_Op_SRS   ,
+    hbbc_Op_CMP   , hbbc_Op_CMPU , hbbc_Op_DIR   , hbbc_Op_NEG   , hbbc_Op_NOT   , hbbc_Op_ADDI  , 
+    hbbc_Op_MULI  , hbbc_Op_ANDI , hbbc_Op_ORI   , hbbc_Op_XORI  , hbbc_Op_SLI   , hbbc_Op_SRI   ,
+    hbbc_Op_SRSI  , hbbc_Op_CMPI , hbbc_Op_CMPUI , hbbc_Op_CP    , hbbc_Op_SWA   , hbbc_Op_LI    ,
+    hhbc_Op_LRA   , hbbc_Op_LD   , hbbc_Op_ST    , hbbc_Op_LDR   , hhbc_Op_STR   , hbbc_Op_BMC   ,
+    hbbc_Op_BRC   , hbbc_Op_JMP  , hbbc_Op_JMPR  , hbbc_Op_JAL   , hbbc_Op_JALR  , hbbc_Op_JEQ   ,
+    hbbc_Op_JNE   , hbbc_Op_JLT  , hbbc_Op_JGT   , hbbc_Op_JLTU  , hbbc_Op_JGTU  , hbbc_Op_ECALL ,
+    hbbc_Op_ADDF  , hbbc_Op_SUBF , hbbc_Op_MULF  , hbbc_Op_DIRF  , hbbc_Op_FMAF  , hbbc_Op_NEGF  ,
+    hbbc_Op_ITF   , hbbc_Op_FTI  , hbbc_Op_ADDFI , hbbc_Op_MULFI ,
+} typedef hbbc_Opcode;
+
+static_assert(sizeof(hbbc_Opcode) == 1);
+
+#pragma pack(push, 1)
+struct hbbc_ParamBBBB
+    { uint8_t _0; uint8_t _1; uint8_t _2; uint8_t _3; }
+    typedef hbbc_ParamBBBB;
+    static_assert(sizeof(hbbc_ParamBBBB) == 32 / 8);
+
+struct hbbc_ParamBBB
+    { uint8_t _0; uint8_t _1; uint8_t _2; }
+    typedef hbbc_ParamBBB;
+    static_assert(sizeof(hbbc_ParamBBB) == 24 / 8);
+
+struct hbbc_ParamBBDH
+    { uint8_t _0; uint8_t _1; uint64_t _2; uint16_t _3; }
+    typedef hbbc_ParamBBDH;
+    static_assert(sizeof(hbbc_ParamBBDH) == 96 / 8);
+
+struct hbbc_ParamBBWH
+    { uint8_t _0; uint8_t _1; uint32_t _2; uint16_t _3; }
+    typedef hbbc_ParamBBWH;
+    static_assert(sizeof(hbbc_ParamBBWH) == 64 / 8);
+
+
+struct hbbc_ParamBBD
+    { uint8_t _0; uint8_t _1; uint64_t _2; }
+    typedef hbbc_ParamBBD;
+    static_assert(sizeof(hbbc_ParamBBD) == 80 / 8);
+
+struct hbbc_ParamBBW
+    { uint8_t _0; uint8_t _1; uint32_t _2; }
+    typedef hbbc_ParamBBW;
+    static_assert(sizeof(hbbc_ParamBBW) == 48 / 8);
+
+struct hbbc_ParamBB
+    { uint8_t _0; uint8_t _1; }
+    typedef hbbc_ParamBB;
+    static_assert(sizeof(hbbc_ParamBB) == 16 / 8);
+
+struct hbbc_ParamBD
+    { uint8_t _0; uint64_t _1; }
+    typedef hbbc_ParamBD;
+    static_assert(sizeof(hbbc_ParamBD) == 72 / 8);
+
+typedef uint64_t hbbc_ParamD;
+    static_assert(sizeof(hbbc_ParamD) == 64 / 8);
+
+#pragma pack(pop)

bytecode/instructions.in

@@ -0,0 +1,120 @@
+0x00, UN,       N,    "Cause an unreachable code trap"              ;
+0x01, TX,       N,    "Termiante execution"                         ;
+0x02, NOP,      N,    "Do nothing"                                  ;
+0x03, ADD8,     RRR,  "Addition (8b)"                               ;
+0x04, ADD16,    RRR,  "Addition (16b)"                              ;
+0x05, ADD32,    RRR,  "Addition (32b)"                              ;
+0x06, ADD64,    RRR,  "Addition (64b)"                              ;
+0x07, SUB8,     RRR,  "Subtraction (8b)"                            ;
+0x08, SUB16,    RRR,  "Subtraction (16b)"                           ;
+0x09, SUB32,    RRR,  "Subtraction (32b)"                           ;
+0x0A, SUB64,    RRR,  "Subtraction (64b)"                           ;
+0x0B, MUL8,     RRR,  "Multiplication (8b)"                         ;
+0x0C, MUL16,    RRR,  "Multiplication (16b)"                        ;
+0x0D, MUL32,    RRR,  "Multiplication (32b)"                        ;
+0x0E, MUL64,    RRR,  "Multiplication (64b)"                        ;
+0x0F, AND,      RRR,  "Bitand"                                      ;
+0x10, OR,       RRR,  "Bitor"                                       ;
+0x11, XOR,      RRR,  "Bitxor"                                      ;
+0x12, SLU8,     RRR,  "Unsigned left bitshift (8b)"                 ;
+0x13, SLU16,    RRR,  "Unsigned left bitshift (16b)"                ;
+0x14, SLU32,    RRR,  "Unsigned left bitshift (32b)"                ;
+0x15, SLU64,    RRR,  "Unsigned left bitshift (64b)"                ;
+0x16, SRU8,     RRR,  "Unsigned right bitshift (8b)"                ;
+0x17, SRU16,    RRR,  "Unsigned right bitshift (16b)"               ;
+0x18, SRU32,    RRR,  "Unsigned right bitshift (32b)"               ;
+0x19, SRU64,    RRR,  "Unsigned right bitshift (64b)"               ;
+0x1A, SRS8,     RRR,  "Signed right bitshift (8b)"                  ;
+0x1B, SRS16,    RRR,  "Signed right bitshift (16b)"                 ;
+0x1C, SRS32,    RRR,  "Signed right bitshift (32b)"                 ;
+0x1D, SRS64,    RRR,  "Signed right bitshift (64b)"                 ;
+0x1E, CMPU,     RRR,  "Unsigned comparsion"                         ;
+0x1F, CMPS,     RRR,  "Signed comparsion"                           ;
+0x20, DIRU8,    RRRR, "Merged divide-remainder (unsigned 8b)"       ;
+0x21, DIRU16,   RRRR, "Merged divide-remainder (unsigned 16b)"      ;
+0x22, DIRU32,   RRRR, "Merged divide-remainder (unsigned 32b)"      ;
+0x23, DIRU64,   RRRR, "Merged divide-remainder (unsigned 64b)"      ;
+0x24, DIRS8,    RRRR, "Merged divide-remainder (signed 8b)"         ;
+0x25, DIRS16,   RRRR, "Merged divide-remainder (signed 16b)"        ;
+0x26, DIRS32,   RRRR, "Merged divide-remainder (signed 32b)"        ;
+0x27, DIRS64,   RRRR, "Merged divide-remainder (signed 64b)"        ;
+0x28, NEG,      RR,   "Bit negation"                                ;
+0x29, NOT,      RR,   "Logical negation"                            ;
+0x2A, SXT8,     RR,   "Sign extend 8b to 64b"                       ;
+0x2B, SXT16,    RR,   "Sign extend 16b to 64b"                      ;
+0x2C, SXT32,    RR,   "Sign extend 32b to 64b"                      ;
+0x2D, ADDI8,    RRB,  "Addition with immediate (8b)"                ;
+0x2E, ADDI16,   RRH,  "Addition with immediate (16b)"               ;
+0x2F, ADDI32,   RRW,  "Addition with immediate (32b)"               ;
+0x30, ADDI64,   RRD,  "Addition with immediate (64b)"               ;
+0x31, MULI8,    RRB,  "Multiplication with immediate (8b)"          ;
+0x32, MULI16,   RRH,  "Multiplication with immediate (16b)"         ;
+0x33, MULI32,   RRW,  "Multiplication with immediate (32b)"         ;
+0x34, MULI64,   RRD,  "Multiplication with immediate (64b)"         ;
+0x35, ANDI,     RRD,  "Bitand with immediate"                       ;
+0x36, ORI,      RRD,  "Bitor with immediate"                        ;
+0x37, XORI,     RRD,  "Bitxor with immediate"                       ;
+0x38, SLUI8,    RRB,  "Unsigned left bitshift with immedidate (8b)" ;
+0x39, SLUI16,   RRB,  "Unsigned left bitshift with immedidate (16b)";
+0x3A, SLUI32,   RRB,  "Unsigned left bitshift with immedidate (32b)";
+0x3B, SLUI64,   RRB,  "Unsigned left bitshift with immedidate (64b)";
+0x3C, SRUI8,    RRB,  "Unsigned right bitshift with immediate (8b)" ;
+0x3D, SRUI16,   RRB,  "Unsigned right bitshift with immediate (16b)";
+0x3E, SRUI32,   RRB,  "Unsigned right bitshift with immediate (32b)";
+0x3F, SRUI64,   RRB,  "Unsigned right bitshift with immediate (64b)";
+0x40, SRSI8,    RRB,  "Signed right bitshift with immediate"        ;
+0x41, SRSI16,   RRB,  "Signed right bitshift with immediate"        ;
+0x42, SRSI32,   RRB,  "Signed right bitshift with immediate"        ;
+0x43, SRSI64,   RRB,  "Signed right bitshift with immediate"        ;
+0x44, CMPUI,    RRD,  "Unsigned compare with immediate"             ;
+0x45, CMPSI,    RRD,  "Signed compare with immediate"               ;
+0x46, CP,       RR,   "Copy register"                               ;
+0x47, SWA,      RR,   "Swap registers"                              ;
+0x48, LI8,      RB,   "Load immediate (8b)"                         ;
+0x49, LI16,     RH,   "Load immediate (16b)"                        ;
+0x4A, LI32,     RW,   "Load immediate (32b)"                        ;
+0x4B, LI64,     RD,   "Load immediate (64b)"                        ;
+0x4C, LRA,      RRO,  "Load relative address"                       ;
+0x4D, LD,       RRAH, "Load from absolute address"                  ;
+0x4E, ST,       RRAH, "Store to absolute address"                   ;
+0x4F, LDR,      RROH, "Load from relative address"                  ;
+0x50, STR,      RROH, "Store to relative address"                   ;
+0x51, BMC,      RRH,  "Copy block of memory"                        ;
+0x52, BRC,      RRB,  "Copy register block"                         ;
+0x53, JMP,      O,    "Relative jump"                               ;
+0x54, JAL,      RRO,  "Linking relative jump"                       ;
+0x55, JALA,     RRA,  "Linking absolute jump"                       ;
+0x56, JEQ,      RRP,  "Branch on equal"                             ;
+0x57, JNE,      RRP,  "Branch on nonequal"                          ;
+0x58, JLTU,     RRP,  "Branch on lesser-than (unsigned)"            ;
+0x59, JGTU,     RRP,  "Branch on greater-than (unsigned)"           ;
+0x5A, JLTS,     RRP,  "Branch on lesser-than (signed)"              ;
+0x5B, JGTS,     RRP,  "Branch on greater-than (signed)"             ;
+0x5C, ECA,      N,    "Environment call trap"                       ;
+0x5D, EBP,      N,    "Environment breakpoint"                      ;
+0x5E, FADD32,   RRR,  "Floating point addition (32b)"               ;
+0x5F, FADD64,   RRR,  "Floating point addition (64b)"               ;
+0x60, FSUB32,   RRR,  "Floating point subtraction (32b)"            ;
+0x61, FSUB64,   RRR,  "Floating point subtraction (64b)"            ;
+0x62, FMUL32,   RRR,  "Floating point multiply (32b)"               ;
+0x63, FMUL64,   RRR,  "Floating point multiply (64b)"               ;
+0x64, FDIV32,   RRR,  "Floating point division (32b)"               ;
+0x65, FDIV64,   RRR,  "Floating point division (64b)"               ;
+0x66, FMA32,    RRRR, "Float fused multiply-add (32b)"              ;
+0x67, FMA64,    RRRR, "Float fused multiply-add (64b)"              ;
+0x68, FINV32,   RR,   "Float reciprocal (32b)"                      ;
+0x69, FINV64,   RR,   "Float reciprocal (64b)"                      ;
+0x6A, FCMPLT32, RRR,  "Flaot compare less than (32b)"               ;
+0x6B, FCMPLT64, RRR,  "Flaot compare less than (64b)"               ;
+0x6C, FCMPGT32, RRR,  "Flaot compare greater than (32b)"            ;
+0x6D, FCMPGT64, RRR,  "Flaot compare greater than (64b)"            ;
+0x6E, ITF32,    RR,   "Int to 32 bit float"                         ;
+0x6F, ITF64,    RR,   "Int to 64 bit float"                         ;
+0x70, FTI32,    RRB,  "Float 32 to int"                             ;
+0x71, FTI64,    RRB,  "Float 64 to int"                             ;
+0x72, FC32T64,  RR,   "Float 64 to Float 32"                        ;
+0x73, FC64T32,  RRB,  "Float 32 to Float 64"                        ;
+0x74, LRA16,    RRP,  "Load relative immediate (16 bit)"            ;
+0x75, LDR16,    RRPH, "Load from relative address (16 bit)"         ;
+0x76, STR16,    RRPH, "Store to relative address (16 bit)"          ;
+0x77, JMP16,    P,    "Relative jump (16 bit)"                      ;

bytecode/src/lib.rs

@@ -0,0 +1,284 @@
+#![no_std]
+
+#[cfg(feature = "disasm")]
+extern crate alloc;
+
+pub use crate::instrs::*;
+use core::convert::TryFrom;
+
+mod instrs;
+
+type OpR = u8;
+
+type OpA = u64;
+type OpO = i32;
+type OpP = i16;
+
+type OpB = u8;
+type OpH = u16;
+type OpW = u32;
+type OpD = u64;
+
+/// # Safety
+/// Has to be valid to be decoded from bytecode.
+pub unsafe trait BytecodeItem {}
+unsafe impl BytecodeItem for u8 {}
+
+impl TryFrom<u8> for Instr {
+    type Error = u8;
+
+    #[inline]
+    fn try_from(value: u8) -> Result<Self, Self::Error> {
+        #[cold]
+        fn failed(value: u8) -> Result<Instr, u8> {
+            Err(value)
+        }
+
+        if value < COUNT {
+            unsafe { Ok(core::mem::transmute::<u8, Instr>(value)) }
+        } else {
+            failed(value)
+        }
+    }
+}
+
+#[inline]
+unsafe fn encode<T>(instr: T) -> (usize, [u8; instrs::MAX_SIZE]) {
+    let mut buf = [0; instrs::MAX_SIZE];
+    core::ptr::write(buf.as_mut_ptr() as *mut T, instr);
+    (core::mem::size_of::<T>(), buf)
+}
+
+#[inline]
+#[cfg(feature = "disasm")]
+fn decode<T>(binary: &mut &[u8]) -> Option<T> {
+    let (front, rest) = core::mem::take(binary).split_at_checked(core::mem::size_of::<T>())?;
+    *binary = rest;
+    unsafe { Some(core::ptr::read(front.as_ptr() as *const T)) }
+}
+
+/// Rounding mode
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+#[repr(u8)]
+pub enum RoundingMode {
+    NearestEven = 0,
+    Truncate = 1,
+    Up = 2,
+    Down = 3,
+}
+
+impl TryFrom<u8> for RoundingMode {
+    type Error = ();
+
+    fn try_from(value: u8) -> Result<Self, Self::Error> {
+        (value <= 3).then(|| unsafe { core::mem::transmute(value) }).ok_or(())
+    }
+}
+
+#[cfg(feature = "disasm")]
+#[derive(Clone, Copy)]
+pub enum DisasmItem {
+    Func,
+    Global,
+}
+
+#[cfg(feature = "disasm")]
+#[derive(Debug)]
+pub enum DisasmError<'a> {
+    InvalidInstruction(u8),
+    InstructionOutOfBounds(&'a str),
+    FmtFailed(core::fmt::Error),
+    HasOutOfBoundsJumps,
+}
+
+#[cfg(feature = "disasm")]
+impl From<core::fmt::Error> for DisasmError<'_> {
+    fn from(value: core::fmt::Error) -> Self {
+        Self::FmtFailed(value)
+    }
+}
+
+#[cfg(feature = "disasm")]
+impl core::fmt::Display for DisasmError<'_> {
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+        match *self {
+            DisasmError::InvalidInstruction(b) => write!(f, "invalid instruction opcode: {b}"),
+            DisasmError::InstructionOutOfBounds(name) => {
+                write!(f, "instruction would go out of bounds of {name} symbol")
+            }
+            DisasmError::FmtFailed(error) => write!(f, "fmt failed: {error}"),
+            DisasmError::HasOutOfBoundsJumps => write!(
+                f,
+                "the code contained jumps that dont got neither to a \
+                valid symbol or local insturction"
+            ),
+        }
+    }
+}
+
+#[cfg(feature = "disasm")]
+impl core::error::Error for DisasmError<'_> {}
+
+#[cfg(feature = "disasm")]
+pub fn disasm<'a>(
+    binary: &mut &[u8],
+    functions: &alloc::collections::BTreeMap<u32, (&'a str, u32, DisasmItem)>,
+    out: &mut alloc::string::String,
+    mut eca_handler: impl FnMut(&mut &[u8]),
+) -> Result<(), DisasmError<'a>> {
+    use {
+        self::instrs::Instr,
+        alloc::{
+            collections::btree_map::{BTreeMap, Entry},
+            vec::Vec,
+        },
+        core::{convert::TryInto, fmt::Write},
+    };
+
+    fn instr_from_byte(b: u8) -> Result<Instr, DisasmError<'static>> {
+        b.try_into().map_err(DisasmError::InvalidInstruction)
+    }
+
+    let mut labels = BTreeMap::<u32, u32>::default();
+    let mut buf = Vec::<instrs::Oper>::new();
+    let mut has_oob = false;
+
+    '_offset_pass: for (&off, &(name, len, kind)) in functions.iter() {
+        if matches!(kind, DisasmItem::Global) {
+            continue;
+        }
+
+        let prev = *binary;
+
+        *binary = &binary[off as usize..];
+
+        let mut label_count = 0;
+        while let Some(&byte) = binary.first() {
+            let offset: i32 = (prev.len() - binary.len()).try_into().unwrap();
+            if offset as u32 == off + len {
+                break;
+            }
+            let Ok(inst) = instr_from_byte(byte) else { break };
+            instrs::parse_args(binary, inst, &mut buf)
+                .ok_or(DisasmError::InstructionOutOfBounds(name))?;
+
+            for op in buf.drain(..) {
+                let rel = match op {
+                    instrs::Oper::O(rel) => rel,
+                    instrs::Oper::P(rel) => rel.into(),
+                    _ => continue,
+                };
+
+                let global_offset: u32 = (offset + rel).try_into().unwrap();
+                if functions.get(&global_offset).is_some() {
+                    continue;
+                }
+                label_count += match labels.entry(global_offset) {
+                    Entry::Occupied(_) => 0,
+                    Entry::Vacant(entry) => {
+                        entry.insert(label_count);
+                        1
+                    }
+                }
+            }
+
+            if matches!(inst, Instr::ECA) {
+                eca_handler(binary);
+            }
+        }
+
+        *binary = prev;
+    }
+
+    let mut ordered = functions.iter().collect::<Vec<_>>();
+    ordered.sort_unstable_by_key(|(_, (name, _, _))| name);
+
+    '_dump: for (&off, &(name, len, kind)) in ordered {
+        if matches!(kind, DisasmItem::Global) {
+            continue;
+        }
+        let prev = *binary;
+
+        writeln!(out, "{name}:")?;
+
+        *binary = &binary[off as usize..];
+        while let Some(&byte) = binary.first() {
+            let offset: i32 = (prev.len() - binary.len()).try_into().unwrap();
+            if offset as u32 == off + len {
+                break;
+            }
+            let Ok(inst) = instr_from_byte(byte) else {
+                writeln!(out, "invalid instr {byte}")?;
+                break;
+            };
+            instrs::parse_args(binary, inst, &mut buf).unwrap();
+
+            if let Some(label) = labels.get(&offset.try_into().unwrap()) {
+                write!(out, "{:>2}: ", label)?;
+            } else {
+                write!(out, "    ")?;
+            }
+
+            write!(out, "{inst:<8?} ")?;
+
+            'a: for (i, op) in buf.drain(..).enumerate() {
+                if i != 0 {
+                    write!(out, ", ")?;
+                }
+
+                let rel = 'b: {
+                    match op {
+                        instrs::Oper::O(rel) => break 'b rel,
+                        instrs::Oper::P(rel) => break 'b rel.into(),
+                        instrs::Oper::R(r) => write!(out, "r{r}")?,
+                        instrs::Oper::B(b) => write!(out, "{b}b")?,
+                        instrs::Oper::H(h) => write!(out, "{h}h")?,
+                        instrs::Oper::W(w) => write!(out, "{w}w")?,
+                        instrs::Oper::D(d) if (d as i64) < 0 => write!(out, "{}d", d as i64)?,
+                        instrs::Oper::D(d) => write!(out, "{d}d")?,
+                        instrs::Oper::A(a) => write!(out, "{a}a")?,
+                    }
+
+                    continue 'a;
+                };
+
+                let global_offset: u32 = (offset + rel).try_into().unwrap();
+                if let Some(&(name, ..)) = functions.get(&global_offset) {
+                    if name.contains('\0') {
+                        write!(out, ":{name:?}")?;
+                    } else {
+                        write!(out, ":{name}")?;
+                    }
+                } else {
+                    let local_has_oob = global_offset < off
+                        || global_offset > off + len
+                        || prev
+                            .get(global_offset as usize)
+                            .map_or(true, |&b| instr_from_byte(b).is_err())
+                        || prev[global_offset as usize] == 0;
+                    has_oob |= local_has_oob;
+                    let label = labels.get(&global_offset).unwrap();
+                    if local_has_oob {
+                        write!(out, "!!!!!!!!!{rel}")?;
+                    } else {
+                        write!(out, ":{label}")?;
+                    }
+                }
+            }
+
+            writeln!(out)?;
+
+            if matches!(inst, Instr::ECA) {
+                eca_handler(binary);
+            }
+        }
+
+        *binary = prev;
+    }
+
+    if has_oob {
+        return Err(DisasmError::HasOutOfBoundsJumps);
+    }
+
+    Ok(())
+}

depell/Cargo.toml

@@ -0,0 +1,23 @@
+[package]
+name = "depell"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]
+argon2 = "0.5.3"
+axum = "0.7.7"
+axum-server = { version = "0.7.1", optional = true, features = ["rustls", "tls-rustls"] }
+const_format = "0.2.33"
+getrandom = "0.2.15"
+hblang.workspace = true
+htmlm = "0.5.0"
+log = "0.4.22"
+rand_core = { version = "0.6.4", features = ["getrandom"] }
+rusqlite = { version = "0.32.1", features = ["bundled"] }
+serde = { version = "1.0.210", features = ["derive"] }
+time = "0.3.36"
+tokio = { version = "1.40.0", features = ["rt"] }
+
+[features]
+#default = ["tls"]
+tls = ["dep:axum-server"]

depell/README.md

@@ -0,0 +1,14 @@
+# Depell
+
+Depell is a website that allows users to import/post/run hblang code and create huge dependency graphs. Its currently hosted at https://depell.mlokis.tech.
+
+## Local Development
+
+Prerequirements:
+- rust nigthly toolchain: install rust from [here](https://www.rust-lang.org/tools/install)
+
+```bash
+rustup default nightly
+cargo xtask watch-depell-debug
+# browser http://localhost:8080
+```

depell/src/index.css

@@ -0,0 +1,179 @@
+* {
+	font-family: var(--font);
+}
+
+body {
+	--primary: white;
+	--secondary: #EFEFEF;
+	--timestamp: #777777;
+	--error: #ff3333;
+	--placeholder: #333333;
+}
+
+body {
+	--small-gap: 5px;
+	--font: system-ui, -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, 'Open Sans', 'Helvetica Neue', sans-serif;
+	--monospace: 'Courier New', Courier, monospace;
+
+	nav {
+		display: flex;
+		justify-content: space-between;
+		align-items: center;
+
+		section:last-child {
+			display: flex;
+			gap: var(--small-gap);
+		}
+	}
+
+	main {
+		margin-top: var(--small-gap);
+		display: flex;
+		flex-direction: column;
+		gap: var(--small-gap);
+	}
+}
+
+div.preview {
+	div.info {
+		display: flex;
+		gap: var(--small-gap);
+
+		span[apply=timestamp] {
+			color: var(--timestamp);
+		}
+	}
+
+	div.stats {
+		display: flex;
+		gap: var(--small-gap);
+	}
+}
+
+form {
+	display: flex;
+	flex-direction: column;
+	gap: var(--small-gap);
+
+
+	.error {
+		color: var(--error);
+		text-align: center;
+	}
+
+}
+
+textarea {
+	outline: none;
+	border: none;
+	background: var(--secondary);
+	padding: var(--small-gap);
+	padding-top: calc(var(--small-gap) * 1.5);
+	font-family: var(--monospace);
+	resize: none;
+	tab-size: 4;
+}
+
+pre {
+	background: var(--secondary);
+	padding: var(--small-gap);
+	padding-top: calc(var(--small-gap) * 1.5);
+	margin: 0px;
+	font-family: var(--monospace);
+	tab-size: 4;
+	overflow-x: auto;
+}
+
+input {
+	font-size: inherit;
+	outline: none;
+	border: none;
+	background: var(--secondary);
+	padding: var(--small-gap);
+}
+
+input:is(:hover, :focus) {
+	background: var(--primary);
+}
+
+button {
+	border: none;
+	outline: none;
+	font-size: inherit;
+	background: var(--secondary);
+}
+
+button:hover:not(:active) {
+	background: var(--primary);
+}
+
+div#code-editor {
+	display: flex;
+	position: relative;
+
+	textarea {
+		flex: 1;
+	}
+
+	span#code-size {
+		position: absolute;
+		right: 2px;
+		font-size: 12px;
+	}
+}
+
+div#dep-list {
+	display: flex;
+	flex-direction: column;
+	align-items: center;
+	gap: var(--small-gap);
+
+	section {
+		width: 100%;
+		display: flex;
+		flex-direction: column;
+		text-align: center;
+		gap: var(--small-gap);
+
+		div {
+			text-align: left;
+		}
+	}
+}
+
+.syn {
+	font-family: var(--monospace);
+
+	&.Comment {
+		color: #939f91;
+	}
+
+	&.Keyword {
+		color: #f85552;
+	}
+
+	&.Identifier,
+	&.Directive {
+		color: #3a94c5;
+	}
+
+	/* &.Number {} */
+
+	&.String {
+		color: #8da101;
+	}
+
+	&.Op,
+	&.Assign {
+		color: #f57d26;
+	}
+
+	&.Paren,
+	&.Bracket,
+	&.Comma,
+	&.Dor,
+	&.Ctor,
+	&.Colon {
+		color: #5c6a72;
+	}
+}

depell/src/index.js

@@ -0,0 +1,507 @@
+/// @ts-check
+
+/** @return {never} */
+function never() { throw new Error() }
+
+/**@type{WebAssembly.Instance}*/ let hbcInstance;
+/**@type{Promise<WebAssembly.WebAssemblyInstantiatedSource>}*/ let hbcInstaceFuture;
+async function getHbcInstance() {
+	hbcInstaceFuture ??= WebAssembly.instantiateStreaming(fetch("/hbc.wasm"), {});
+	return hbcInstance ??= (await hbcInstaceFuture).instance;
+}
+
+const stack_pointer_offset = 1 << 20;
+
+/** @param {WebAssembly.Instance} instance @param {Post[]} packages @param {number} fuel
+ * @returns {string} */
+function compileCode(instance, packages, fuel) {
+	let {
+		INPUT, INPUT_LEN,
+		LOG_MESSAGES, LOG_MESSAGES_LEN,
+		memory, compile_and_run,
+	} = instance.exports;
+
+	if (!(true
+		&& memory instanceof WebAssembly.Memory
+		&& INPUT instanceof WebAssembly.Global
+		&& INPUT_LEN instanceof WebAssembly.Global
+		&& LOG_MESSAGES instanceof WebAssembly.Global
+		&& LOG_MESSAGES_LEN instanceof WebAssembly.Global
+		&& typeof compile_and_run === "function"
+	)) never();
+
+	const codeLength = packPosts(packages, new DataView(memory.buffer, INPUT.value));
+	new DataView(memory.buffer).setUint32(INPUT_LEN.value, codeLength, true);
+
+	runWasmFunction(instance, compile_and_run, fuel);
+	return bufToString(memory, LOG_MESSAGES, LOG_MESSAGES_LEN);
+}
+
+/**@type{WebAssembly.Instance}*/ let fmtInstance;
+/**@type{Promise<WebAssembly.WebAssemblyInstantiatedSource>}*/ let fmtInstaceFuture;
+async function getFmtInstance() {
+	fmtInstaceFuture ??= WebAssembly.instantiateStreaming(fetch("/hbfmt.wasm"), {});
+	return fmtInstance ??= (await fmtInstaceFuture).instance;
+}
+
+/** @param {WebAssembly.Instance} instance @param {string} code @param {"tok" | "fmt" | "minify"} action
+ * @returns {string | Uint8Array | undefined} */
+function modifyCode(instance, code, action) {
+	let {
+		INPUT, INPUT_LEN,
+		OUTPUT, OUTPUT_LEN,
+		memory, fmt, tok, minify
+	} = instance.exports;
+
+	let funs = { fmt, tok, minify };
+	let fun = funs[action];
+	if (!(true
+		&& memory instanceof WebAssembly.Memory
+		&& INPUT instanceof WebAssembly.Global
+		&& INPUT_LEN instanceof WebAssembly.Global
+		&& OUTPUT instanceof WebAssembly.Global
+		&& OUTPUT_LEN instanceof WebAssembly.Global
+		&& funs.hasOwnProperty(action)
+		&& typeof fun === "function"
+	)) never();
+
+	if (action !== "fmt") {
+		INPUT = OUTPUT;
+		INPUT_LEN = OUTPUT_LEN;
+	}
+
+	let dw = new DataView(memory.buffer);
+	dw.setUint32(INPUT_LEN.value, code.length, true);
+	new Uint8Array(memory.buffer, INPUT.value).set(new TextEncoder().encode(code));
+
+	if (!runWasmFunction(instance, fun)) {
+		return undefined;
+	}
+	if (action === "tok") {
+		return bufSlice(memory, OUTPUT, OUTPUT_LEN);
+	} else {
+		return bufToString(memory, OUTPUT, OUTPUT_LEN);
+	}
+}
+
+
+/** @param {WebAssembly.Instance} instance @param {CallableFunction} func @param {any[]} args
+ * @returns {boolean} */
+function runWasmFunction(instance, func, ...args) {
+	const { PANIC_MESSAGE, PANIC_MESSAGE_LEN, memory, stack_pointer } = instance.exports;
+	if (!(true
+		&& memory instanceof WebAssembly.Memory
+		&& stack_pointer instanceof WebAssembly.Global
+	)) never();
+	const ptr = stack_pointer.value;
+	try {
+		func(...args);
+		return true;
+	} catch (error) {
+		if (error instanceof WebAssembly.RuntimeError
+			&& error.message == "unreachable"
+			&& PANIC_MESSAGE instanceof WebAssembly.Global
+			&& PANIC_MESSAGE_LEN instanceof WebAssembly.Global) {
+			console.error(bufToString(memory, PANIC_MESSAGE, PANIC_MESSAGE_LEN), error);
+		} else {
+			console.error(error);
+		}
+		stack_pointer.value = ptr;
+		return false;
+	}
+}
+
+/** @typedef {Object} Post
+ * @property {string} path 
+ * @property {string} code */
+
+/** @param {Post[]} posts @param {DataView} view  @returns {number} */
+function packPosts(posts, view) {
+	const enc = new TextEncoder(), buf = new Uint8Array(view.buffer, view.byteOffset);
+	let len = 0; for (const post of posts) {
+		view.setUint16(len, post.path.length, true); len += 2;
+		buf.set(enc.encode(post.path), len); len += post.path.length;
+		view.setUint16(len, post.code.length, true); len += 2;
+		buf.set(enc.encode(post.code), len); len += post.code.length;
+	}
+	return len;
+}
+
+/** @param {WebAssembly.Memory} mem
+ * @param {WebAssembly.Global} ptr
+ * @param {WebAssembly.Global} len
+ * @return {Uint8Array} */
+function bufSlice(mem, ptr, len) {
+	return new Uint8Array(mem.buffer, ptr.value,
+		new DataView(mem.buffer).getUint32(len.value, true));
+}
+
+/** @param {WebAssembly.Memory} mem
+ * @param {WebAssembly.Global} ptr
+ * @param {WebAssembly.Global} len
+ * @return {string} */
+function bufToString(mem, ptr, len) {
+	const res = new TextDecoder()
+		.decode(new Uint8Array(mem.buffer, ptr.value,
+			new DataView(mem.buffer).getUint32(len.value, true)));
+	new DataView(mem.buffer).setUint32(len.value, 0, true);
+	return res;
+}
+
+/** @param {HTMLElement} target */
+function wireUp(target) {
+	execApply(target);
+	cacheInputs(target);
+	bindCodeEdit(target);
+	bindTextareaAutoResize(target);
+}
+
+const importRe = /@use\s*\(\s*"(([^"]|\\")+)"\s*\)/g;
+
+/** @param {string} code
+ * @param {string[]} roots
+ * @param {Post[]} buf
+ * @param {Set<string>} prevRoots
+ * @returns {void} */
+function loadCachedPackages(code, roots, buf, prevRoots) {
+	buf[0].code = code;
+
+	roots.length = 0;
+	let changed = false;
+	for (const match of code.matchAll(importRe)) {
+		changed ||= !prevRoots.has(match[1]);
+		roots.push(match[1]);
+	}
+
+	if (!changed) return;
+	buf.length = 1;
+	prevRoots.clear();
+
+	for (let imp = roots.pop(); imp !== undefined; imp = roots.pop()) {
+		if (prevRoots.has(imp)) continue; prevRoots.add(imp);
+		buf.push({ path: imp, code: localStorage.getItem("package-" + imp) ?? never() });
+		for (const match of buf[buf.length - 1].code.matchAll(importRe)) {
+			roots.push(match[1]);
+		}
+	}
+}
+
+/**@type{Set<string>}*/ const prevRoots = new Set();
+
+/** @param {HTMLElement} target */
+async function bindCodeEdit(target) {
+	const edit = target.querySelector("#code-edit");
+	if (!(edit instanceof HTMLTextAreaElement)) return;
+
+	const codeSize = target.querySelector("#code-size");
+	const errors = target.querySelector("#compiler-output");
+	if (!(true
+		&& codeSize instanceof HTMLSpanElement
+		&& errors instanceof HTMLPreElement
+	)) never();
+
+	const MAX_CODE_SIZE = parseInt(codeSize.innerHTML);
+	if (Number.isNaN(MAX_CODE_SIZE)) never();
+
+	const hbc = await getHbcInstance(), fmt = await getFmtInstance();
+	let importDiff = new Set();
+	const keyBuf = [];
+	/**@type{Post[]}*/
+	const packages = [{ path: "local.hb", code: "" }];
+	const debounce = 100;
+	/**@type{AbortController|undefined}*/
+	let cancelation = undefined;
+	let timeout = 0;
+
+	prevRoots.clear();
+
+	const onInput = () => {
+		importDiff.clear();
+		for (const match of edit.value.matchAll(importRe)) {
+			if (localStorage["package-" + match[1]]) continue;
+			importDiff.add(match[1]);
+		}
+
+		if (importDiff.size !== 0) {
+			if (cancelation) cancelation.abort();
+			cancelation = new AbortController();
+
+			keyBuf.length = 0;
+			keyBuf.push(...importDiff.keys());
+
+			errors.textContent = "fetching: " + keyBuf.join(", ");
+
+			fetch(`/code`, {
+				method: "POST",
+				signal: cancelation.signal,
+				headers: { "Content-Type": "application/json" },
+				body: JSON.stringify(keyBuf),
+			}).then(async e => {
+				try {
+					const json = await e.json();
+					if (e.status == 200) {
+						for (const [key, value] of Object.entries(json)) {
+							localStorage["package-" + key] = value;
+						}
+						const missing = keyBuf.filter(i => json[i] === undefined);
+						if (missing.length !== 0) {
+							errors.textContent = "deps not found: " + missing.join(", ");
+						} else {
+							cancelation = undefined;
+							edit.dispatchEvent(new InputEvent("input"));
+						}
+					}
+				} catch (er) {
+					errors.textContent = "completely failed to fetch ("
+						+ e.status + "): " + keyBuf.join(", ");
+					console.error(e, er);
+				}
+			});
+		}
+
+		if (cancelation && importDiff.size !== 0) {
+			return;
+		}
+
+		loadCachedPackages(edit.value, keyBuf, packages, prevRoots);
+
+		errors.textContent = compileCode(hbc, packages, 1);
+		const minified_size = modifyCode(fmt, edit.value, "minify")?.length;
+		if (minified_size) {
+			codeSize.textContent = (MAX_CODE_SIZE - minified_size) + "";
+			const perc = Math.min(100, Math.floor(100 * (minified_size / MAX_CODE_SIZE)));
+			codeSize.style.color = `color-mix(in srgb, white, var(--error) ${perc}%)`;
+		}
+		timeout = 0;
+	};
+
+	edit.addEventListener("input", () => {
+		if (timeout) clearTimeout(timeout);
+		timeout = setTimeout(onInput, debounce)
+	});
+	edit.dispatchEvent(new InputEvent("input"));
+}
+
+/**
+ * @type {Array<string>}
+ * to be synched with `enum TokenGroup` in bytecode/src/fmt.rs */
+const TOK_CLASSES = [
+	'Blank',
+	'Comment',
+	'Keyword',
+	'Identifier',
+	'Directive',
+	'Number',
+	'String',
+	'Op',
+	'Assign',
+	'Paren',
+	'Bracket',
+	'Colon',
+	'Comma',
+	'Dot',
+	'Ctor',
+];
+
+/** @type {{ [key: string]: (el: HTMLElement) => void | Promise<void> }} */
+const applyFns = {
+	timestamp: (el) => {
+		const timestamp = el.innerText;
+		const date = new Date(parseInt(timestamp) * 1000);
+		el.innerText = date.toLocaleString();
+	},
+	fmt,
+};
+
+/**
+ * @param {HTMLElement} target */
+async function fmt(target) {
+	const code = target.innerText;
+	const instance = await getFmtInstance();
+	const decoder = new TextDecoder('utf-8');
+	const fmt = modifyCode(instance, code, 'fmt');
+	if (typeof fmt !== "string") never()
+	const codeBytes = new TextEncoder().encode(fmt);
+	const tok = modifyCode(instance, fmt, 'tok');
+	if (!(tok instanceof Uint8Array)) never();
+	target.innerHTML = '';
+	let start = 0;
+	let kind = tok[0];
+	for (let ii = 1; ii <= tok.length; ii += 1) {
+		// split over same tokens and buffer end
+		if (tok[ii] === kind && ii < tok.length) {
+			continue;
+		}
+		const text = decoder.decode(codeBytes.subarray(start, ii));
+		const textNode = document.createTextNode(text);;
+		if (kind === 0) {
+			target.appendChild(textNode);
+		} else {
+			const el = document.createElement('span');
+			el.classList.add('syn');
+			el.classList.add(TOK_CLASSES[kind]);
+			el.appendChild(textNode);
+			target.appendChild(el);
+		}
+		if (ii == tok.length) {
+			break;
+		}
+		start = ii;
+		kind = tok[ii];
+	}
+}
+
+/** @param {HTMLElement} target */
+async function execApply(target) {
+	for (const elem of target.querySelectorAll('[apply]')) {
+		if (!(elem instanceof HTMLElement)) continue;
+		const funcname = elem.getAttribute('apply') ?? never();
+		applyFns[funcname](elem);
+	}
+}
+
+/** @param {HTMLElement} target */
+function bindTextareaAutoResize(target) {
+	for (const textarea of target.querySelectorAll("textarea")) {
+		if (!(textarea instanceof HTMLTextAreaElement)) never();
+
+		const taCssMap = window.getComputedStyle(textarea);
+		const padding = parseInt(taCssMap.getPropertyValue('padding-top') ?? "0")
+			+ parseInt(taCssMap.getPropertyValue('padding-top') ?? "0");
+		textarea.style.height = "auto";
+		textarea.style.height = (textarea.scrollHeight - padding) + "px";
+		textarea.style.overflowY = "hidden";
+		textarea.addEventListener("input", function() {
+			let top = window.scrollY;
+			textarea.style.height = "auto";
+			textarea.style.height = (textarea.scrollHeight - padding) + "px";
+			window.scrollTo({ top });
+		});
+
+		textarea.onkeydown = (ev) => {
+			if (ev.key === "Tab") {
+				ev.preventDefault();
+				document.execCommand('insertText', false, "\t");
+			}
+		}
+	}
+}
+
+/** @param {HTMLElement} target */
+function cacheInputs(target) {
+	/**@type {HTMLFormElement}*/ let form;
+	for (form of target.querySelectorAll('form')) {
+		const path = form.getAttribute('hx-post') || form.getAttribute('hx-delete');
+		if (!path) {
+			console.warn('form does not have a hx-post or hx-delete attribute', form);
+			continue;
+		}
+
+		for (const input of form.elements) {
+			if (input instanceof HTMLInputElement || input instanceof HTMLTextAreaElement) {
+				if ('password submit button'.includes(input.type)) continue;
+				const key = path + input.name;
+				input.value = localStorage.getItem(key) ?? '';
+				input.addEventListener("input", () => localStorage.setItem(key, input.value));
+			} else {
+				console.warn("unhandled form element: ", input);
+			}
+		}
+	}
+}
+
+/** @param {string} [path]  */
+function updaetTab(path) {
+	for (const elem of document.querySelectorAll("button[hx-push-url]")) {
+		if (elem instanceof HTMLButtonElement)
+			elem.disabled = elem.getAttribute("hx-push-url") === (path ?? window.location.pathname);
+	}
+}
+
+if (window.location.hostname === 'localhost') {
+	let id; setInterval(async () => {
+		let new_id = await fetch('/hot-reload').then(reps => reps.text());
+		id ??= new_id;
+		if (id !== new_id) window.location.reload();
+	}, 300);
+
+	(async function test() {
+		{
+			const code = "main:=fn():void{return}";
+			const inst = await getFmtInstance()
+			const fmtd = modifyCode(inst, code, "fmt") ?? never();
+			if (typeof fmtd !== "string") never();
+			const prev = modifyCode(inst, fmtd, "minify") ?? never();
+			if (code != prev) console.error(code, prev);
+		}
+		{
+			const posts = [{
+				path: "foo.hb",
+				code: "main:=fn():int{return 42}",
+			}];
+			const res = compileCode(await getHbcInstance(), posts, 1) ?? never();
+			const expected = "exit code: 42\n";
+			if (expected != res) console.error(expected, res);
+		}
+	})()
+}
+
+document.body.addEventListener('htmx:afterSwap', (ev) => {
+	if (!(ev.target instanceof HTMLElement)) never();
+	wireUp(ev.target);
+	if (ev.target.tagName == "MAIN" || ev.target.tagName == "BODY")
+		updaetTab(ev['detail'].pathInfo.finalRequestPath);
+	console.log(ev);
+});
+
+getFmtInstance().then(inst => {
+	document.body.addEventListener('htmx:configRequest', (ev) => {
+		const details = ev['detail'];
+		if (details.path === "/post" && details.verb === "post") {
+			details.parameters['code'] = modifyCode(inst, details.parameters['code'], "minify");
+		}
+	});
+
+	/** @param {string} query @param {string} target @returns {number} */
+	function fuzzyCost(query, target) {
+		let qi = 0, bi = 0, cost = 0, matched = false;
+		while (qi < query.length) {
+			if (query.charAt(qi) === target.charAt(bi++)) {
+				matched = true;
+				qi++;
+			} else {
+				cost++;
+			}
+			if (bi === target.length) (bi = 0, qi++);
+		}
+		return cost + (matched ? 0 : 100 * target.length);
+	}
+
+	let deps = undefined;
+	/** @param {HTMLInputElement} input @returns {void} */
+	function filterCodeDeps(input) {
+		deps ??= document.getElementById("deps");
+		if (!(deps instanceof HTMLElement)) never();
+		if (input.value === "") {
+			deps.textContent = "results show here...";
+			return;
+		}
+		deps.innerHTML = "";
+		for (const root of [...prevRoots.keys()]
+			.sort((a, b) => fuzzyCost(input.value, a) - fuzzyCost(input.value, b))) {
+			const pane = document.createElement("div");
+			const code = modifyCode(inst, localStorage["package-" + root], "fmt");
+			pane.innerHTML = `<div>${root}</div><pre>${code}</pre>`;
+			deps.appendChild(pane);
+		}
+		if (deps.innerHTML === "") {
+			deps.textContent = "no results";
+		}
+	}
+
+	Object.assign(window, { filterCodeDeps });
+});
+
+updaetTab();
+wireUp(document.body);
+

depell/src/main.rs

@@ -0,0 +1,884 @@
+#![feature(iter_collect_into)]
+use {
+    argon2::{password_hash::SaltString, PasswordVerifier},
+    axum::{
+        body::Bytes,
+        extract::{DefaultBodyLimit, Path},
+        http::{header::COOKIE, request::Parts},
+        response::{AppendHeaders, Html},
+    },
+    const_format::formatcp,
+    core::fmt,
+    htmlm::{html, write_html},
+    rand_core::OsRng,
+    serde::{Deserialize, Serialize},
+    std::{
+        collections::{HashMap, HashSet},
+        fmt::{Display, Write},
+        net::Ipv4Addr,
+    },
+};
+
+const MAX_NAME_LENGTH: usize = 32;
+const MAX_POSTNAME_LENGTH: usize = 64;
+const MAX_CODE_LENGTH: usize = 1024 * 4;
+const SESSION_DURATION_SECS: u64 = 60 * 60;
+const MAX_FEED_SIZE: usize = 8 * 1024;
+
+type Redirect<const COUNT: usize = 1> = AppendHeaders<[(&'static str, &'static str); COUNT]>;
+
+macro_rules! static_asset {
+    ($mime:literal, $body:literal) => {
+        get(|| async {
+            axum::http::Response::builder()
+                .header("content-type", $mime)
+                .header("content-encoding", "gzip")
+                .body(axum::body::Body::from(Bytes::from_static(include_bytes!(concat!(
+                    $body, ".gz"
+                )))))
+                .unwrap()
+        })
+    };
+}
+
+async fn amain() {
+    use axum::routing::{delete, get, post};
+
+    let debug = cfg!(debug_assertions);
+
+    log::set_logger(&Logger).unwrap();
+    log::set_max_level(if debug { log::LevelFilter::Warn } else { log::LevelFilter::Error });
+
+    db::init();
+
+    let router = axum::Router::new()
+        .route("/", get(Index::page))
+        .route("/index.css", static_asset!("text/css", "index.css"))
+        .route("/index.js", static_asset!("text/javascript", "index.js"))
+        .route("/hbfmt.wasm", static_asset!("application/wasm", "hbfmt.wasm"))
+        .route("/hbc.wasm", static_asset!("application/wasm", "hbc.wasm"))
+        .route("/index-view", get(Index::get))
+        .route("/feed", get(Feed::page))
+        .route("/feed-view", get(Feed::get))
+        .route("/feed-more", post(Feed::more))
+        .route("/profile", get(Profile::page))
+        .route("/profile-view", get(Profile::get))
+        .route("/profile/:name", get(Profile::get_other_page))
+        .route("/profile/password", post(PasswordChange::post))
+        .route("/profile-view/:name", get(Profile::get_other))
+        .route("/post", get(Post::page))
+        .route("/post-view", get(Post::get))
+        .route("/post", post(Post::post))
+        .route("/code", post(fetch_code))
+        .route("/login", get(Login::page))
+        .route("/login-view", get(Login::get))
+        .route("/login", post(Login::post))
+        .route("/login", delete(Login::delete))
+        .route("/signup", get(Signup::page))
+        .route("/signup-view", get(Signup::get))
+        .route("/signup", post(Signup::post))
+        .route(
+            "/hot-reload",
+            get({
+                let id = std::time::SystemTime::now()
+                    .duration_since(std::time::SystemTime::UNIX_EPOCH)
+                    .unwrap()
+                    .as_millis();
+                move || async move { id.to_string() }
+            }),
+        )
+        .layer(DefaultBodyLimit::max(16 * 1024));
+
+    #[cfg(feature = "tls")]
+    {
+        let addr =
+            (Ipv4Addr::UNSPECIFIED, std::env::var("DEPELL_PORT").unwrap().parse::<u16>().unwrap());
+        let config = axum_server::tls_rustls::RustlsConfig::from_pem_file(
+            std::env::var("DEPELL_CERT_PATH").unwrap(),
+            std::env::var("DEPELL_KEY_PATH").unwrap(),
+        )
+        .await
+        .unwrap();
+
+        axum_server::bind_rustls(addr.into(), config)
+            .serve(router.into_make_service())
+            .await
+            .unwrap();
+    }
+    #[cfg(not(feature = "tls"))]
+    {
+        let addr = (Ipv4Addr::UNSPECIFIED, 8080);
+        let socket = tokio::net::TcpListener::bind(addr).await.unwrap();
+        axum::serve(socket, router).await.unwrap();
+    }
+}
+
+async fn fetch_code(
+    axum::Json(paths): axum::Json<Vec<String>>,
+) -> axum::Json<HashMap<String, String>> {
+    let mut deps = HashMap::<String, String>::new();
+    db::with(|db| {
+        for path in &paths {
+            let Some((author, name)) = path.split_once('/') else { continue };
+            db.fetch_deps
+                .query_map((name, author), |r| {
+                    Ok((
+                        r.get::<_, String>(1)? + "/" + r.get_ref(0)?.as_str()?,
+                        r.get::<_, String>(2)?,
+                    ))
+                })
+                .log("fetch deps query")
+                .into_iter()
+                .flatten()
+                .filter_map(|r| r.log("deps row"))
+                .collect_into(&mut deps);
+        }
+    });
+    axum::Json(deps)
+}
+
+#[derive(Deserialize)]
+#[serde(untagged)]
+enum Feed {
+    Before { before_timestamp: u64 },
+}
+
+#[derive(Deserialize)]
+struct Before {
+    before_timestamp: u64,
+}
+
+impl Feed {
+    async fn more(session: Session, axum::Form(data): axum::Form<Before>) -> Html<String> {
+        Self::Before { before_timestamp: data.before_timestamp }.render(&session)
+    }
+}
+
+impl Default for Feed {
+    fn default() -> Self {
+        Self::Before { before_timestamp: now() + 3600 }
+    }
+}
+
+impl Page for Feed {
+    fn render_to_buf(self, _: &Session, buf: &mut String) {
+        db::with(|db| {
+            let cursor = match self {
+                Feed::Before { before_timestamp } => db
+                    .get_pots_before
+                    .query_map((before_timestamp,), Post::from_row)
+                    .log("fetch before posts query")
+                    .into_iter()
+                    .flatten()
+                    .filter_map(|r| r.log("fetch before posts row")),
+            };
+
+            let base_len = buf.len();
+            let mut last_timestamp = None;
+            for post in cursor {
+                write!(buf, "{}", post).unwrap();
+                if buf.len() - base_len > MAX_FEED_SIZE {
+                    last_timestamp = Some(post.timestamp);
+                    break;
+                }
+            }
+
+            write_html!((*buf)
+                if let Some(last_timestamp) = last_timestamp {
+                    <div "hx-post"="/feed-more"
+                        "hx-trigger"="intersect once"
+                        "hx-swap"="outerHTML"
+                        "hx-vals"={format_args!("{{\"before_timestamp\":{last_timestamp}}}")}
+                    >"there might be more"</div>
+                } else {
+                    "no more stuff"
+                }
+            );
+        });
+    }
+}
+
+#[derive(Default)]
+struct Index;
+
+impl PublicPage for Index {
+    fn render_to_buf(self, buf: &mut String) {
+        buf.push_str(include_str!("welcome-page.html"));
+    }
+}
+
+#[derive(Deserialize, Default)]
+struct Post {
+    author: String,
+    name: String,
+    #[serde(skip)]
+    timestamp: u64,
+    #[serde(skip)]
+    imports: usize,
+    #[serde(skip)]
+    runs: usize,
+    #[serde(skip)]
+    dependencies: usize,
+    code: String,
+    #[serde(skip)]
+    error: Option<&'static str>,
+}
+
+impl Page for Post {
+    fn render_to_buf(self, session: &Session, buf: &mut String) {
+        let Self { name, code, error, .. } = self;
+        write_html! { (buf)
+            <form id="postForm" "hx-post"="/post" "hx-swap"="outerHTML">
+                if let Some(e) = error { <div class="error">e</div> }
+                <input name="author" type="text" value={session.name} hidden>
+                <input name="name" type="text" placeholder="name" value=name
+                    required maxlength=MAX_POSTNAME_LENGTH>
+                <div id="code-editor">
+                    <textarea id="code-edit" name="code" placeholder="code" rows=1
+                        required>code</textarea>
+                    <span id="code-size">MAX_CODE_LENGTH</span>
+                </div>
+                <input type="submit" value="submit">
+                <pre id="compiler-output"></pre>
+            </form>
+            !{include_str!("post-page.html")}
+        }
+    }
+}
+
+impl Post {
+    pub fn from_row(r: &rusqlite::Row) -> rusqlite::Result<Self> {
+        Ok(Post {
+            author: r.get(0)?,
+            name: r.get(1)?,
+            timestamp: r.get(2)?,
+            code: r.get(3)?,
+            ..Default::default()
+        })
+    }
+
+    async fn post(
+        session: Session,
+        axum::Form(mut data): axum::Form<Self>,
+    ) -> Result<Redirect, Html<String>> {
+        if data.name.len() > MAX_POSTNAME_LENGTH {
+            data.error = Some(formatcp!("name too long, max length is {MAX_POSTNAME_LENGTH}"));
+            return Err(data.render(&session));
+        }
+
+        if data.code.len() > MAX_CODE_LENGTH {
+            data.error = Some(formatcp!("code too long, max length is {MAX_CODE_LENGTH}"));
+            return Err(data.render(&session));
+        }
+
+        db::with(|db| {
+            if let Err(e) = db.create_post.insert((&data.name, &session.name, now(), &data.code)) {
+                if let rusqlite::Error::SqliteFailure(e, _) = e {
+                    if e.code == rusqlite::ErrorCode::ConstraintViolation {
+                        data.error = Some("this name is already used");
+                    }
+                }
+                data.error = data.error.or_else(|| {
+                    log::error!("create post error: {e}");
+                    Some("internal server error")
+                });
+                return;
+            }
+
+            for (author, name) in hblang::lexer::Lexer::uses(&data.code)
+                .filter_map(|v| v.split_once('/'))
+                .collect::<HashSet<_>>()
+            {
+                if db
+                    .create_import
+                    .insert((author, name, &session.name, &data.name))
+                    .log("create import query")
+                    .is_none()
+                {
+                    data.error = Some("internal server error");
+                    return;
+                };
+            }
+        });
+
+        if data.error.is_some() {
+            Err(data.render(&session))
+        } else {
+            Ok(redirect("/profile"))
+        }
+    }
+}
+
+impl fmt::Display for Post {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        let Self { author, name, timestamp, imports, runs, dependencies, code, .. } = self;
+        write_html! { f <div class="preview">
+            <div class="info">
+                <span>
+                    <a "hx-get"={format_args!("/profile-view/{author}")} href="" "hx-target"="main"
+                        "hx-push-url"={format_args!("/profile/{author}")}
+                        "hx-swam"="innerHTML">author</a>
+                    "/"
+                    name
+                </span>
+                <span apply="timestamp">timestamp</span>
+            </div>
+            <div class="stats">
+                for (name, count) in "inps runs deps".split(' ')
+                    .zip([imports, runs, dependencies])
+                    .filter(|(_, &c)| c != 0)
+                {
+                    name ": "<span>count</span>
+                }
+            </div>
+            <pre apply="fmt">code</pre>
+            if *timestamp == 0 {
+                <button "hx-get"="/post" "hx-swap"="outerHTML"
+                    "hx-target"="[preview]">"edit"</button>
+            }
+        </div> }
+        Ok(())
+    }
+}
+
+#[derive(Deserialize, Default)]
+struct PasswordChange {
+    old_password: String,
+    new_password: String,
+    #[serde(skip)]
+    error: Option<&'static str>,
+}
+
+impl PasswordChange {
+    async fn post(
+        session: Session,
+        axum::Form(mut change): axum::Form<PasswordChange>,
+    ) -> Html<String> {
+        db::with(|que| {
+            match que.authenticate.query_row((&session.name,), |r| r.get::<_, String>(1)) {
+                Ok(hash) if verify_password(&hash, &change.old_password).is_err() => {
+                    change.error = Some("invalid credentials");
+                }
+                Ok(_) => {
+                    let new_hashed = hash_password(&change.new_password);
+                    match que
+                        .change_passowrd
+                        .execute((new_hashed, &session.name))
+                        .log("execute update")
+                    {
+                        None => change.error = Some("intenal server error"),
+                        Some(0) => change.error = Some("password is incorrect"),
+                        Some(_) => {}
+                    }
+                }
+                Err(rusqlite::Error::QueryReturnedNoRows) => {
+                    change.error = Some("invalid credentials");
+                }
+                Err(e) => {
+                    log::error!("login queri failed: {e}");
+                    change.error = Some("internal server error");
+                }
+            }
+        });
+
+        if change.error.is_some() {
+            change.render(&session)
+        } else {
+            PasswordChange::default().render(&session)
+        }
+    }
+}
+
+impl Page for PasswordChange {
+    fn render_to_buf(self, _: &Session, buf: &mut String) {
+        let Self { old_password, new_password, error } = self;
+        write_html! { (buf)
+            <form "hx-post"="/profile/password" "hx-swap"="outerHTML">
+                if let Some(e) = error { <div class="error">e</div> }
+                <input name="old_password" type="password" autocomplete="old-password"
+                    placeholder="old password" value=old_password>
+                <input name="new_password" type="password" autocomplete="new-password" placeholder="new password"
+                    value=new_password>
+                <input type="submit" value="submit">
+            </form>
+        }
+    }
+}
+
+#[derive(Default)]
+struct Profile {
+    other: Option<String>,
+}
+
+impl Profile {
+    async fn get_other(session: Session, Path(name): Path<String>) -> Html<String> {
+        Profile { other: Some(name) }.render(&session)
+    }
+
+    async fn get_other_page(session: Session, Path(name): Path<String>) -> Html<String> {
+        base(|b| Profile { other: Some(name) }.render_to_buf(&session, b), Some(&session))
+    }
+}
+
+impl Page for Profile {
+    fn render_to_buf(self, session: &Session, buf: &mut String) {
+        db::with(|db| {
+            let name = self.other.as_ref().unwrap_or(&session.name);
+            let iter = db
+                .get_user_posts
+                .query_map((name,), Post::from_row)
+                .log("get user posts query")
+                .into_iter()
+                .flatten()
+                .filter_map(|p| p.log("user post row"));
+            write_html! { (*buf)
+                if name == &session.name {
+                    |b|{PasswordChange::default().render_to_buf(session, b)}
+                }
+
+                for post in iter {
+                    !{post}
+                } else {
+                    "no posts"
+                }
+                !{include_str!("profile-page.html")}
+            }
+        })
+    }
+}
+
+fn hash_password(password: &str) -> String {
+    use argon2::PasswordHasher;
+    argon2::Argon2::default()
+        .hash_password(password.as_bytes(), &SaltString::generate(&mut OsRng))
+        .unwrap()
+        .to_string()
+}
+
+fn verify_password(hash: &str, password: &str) -> Result<(), argon2::password_hash::Error> {
+    argon2::Argon2::default()
+        .verify_password(password.as_bytes(), &argon2::PasswordHash::new(hash)?)
+}
+
+#[derive(Serialize, Deserialize, Default, Debug)]
+struct Login {
+    name: String,
+    password: String,
+    #[serde(skip)]
+    error: Option<&'static str>,
+}
+
+impl PublicPage for Login {
+    fn render_to_buf(self, buf: &mut String) {
+        let Self { name, password, error } = self;
+        write_html! { (buf)
+            <form "hx-post"="/login" "hx-swap"="outerHTML">
+                if let Some(e) = error { <div class="error">e</div> }
+                <input name="name" type="text" autocomplete="name" placeholder="name" value=name
+                    required maxlength=MAX_NAME_LENGTH>
+                <input name="password" type="password" autocomplete="current-password" placeholder="password"
+                    value=password>
+                <input type="submit" value="submit">
+            </form>
+        }
+    }
+}
+
+impl Login {
+    async fn post(
+        axum::Form(mut data): axum::Form<Self>,
+    ) -> Result<AppendHeaders<[(&'static str, String); 2]>, Html<String>> {
+        // TODO: hash password
+        let mut id = [0u8; 32];
+        db::with(|db| match db.authenticate.query_row((&data.name,), |r| r.get::<_, String>(1)) {
+            Ok(hash) => {
+                if verify_password(&hash, &data.password).is_err() {
+                    data.error = Some("invalid credentials");
+                } else {
+                    getrandom::getrandom(&mut id).unwrap();
+                    if db
+                        .login
+                        .insert((id, &data.name, now() + SESSION_DURATION_SECS))
+                        .log("create session query")
+                        .is_none()
+                    {
+                        data.error = Some("internal server error");
+                    }
+                }
+            }
+            Err(rusqlite::Error::QueryReturnedNoRows) => {
+                data.error = Some("invalid credentials");
+            }
+            Err(e) => {
+                log::error!("login queri failed: {e}");
+                data.error = Some("internal server error");
+            }
+        });
+
+        if data.error.is_some() {
+            Err(data.render())
+        } else {
+            Ok(AppendHeaders([
+                ("hx-location", "/feed".into()),
+                (
+                    "set-cookie",
+                    format!(
+                        "id={}; SameSite=Strict; Secure; Max-Age={SESSION_DURATION_SECS}",
+                        to_hex(&id)
+                    ),
+                ),
+            ]))
+        }
+    }
+
+    async fn delete(session: Session) -> Redirect {
+        _ = db::with(|q| q.logout.execute((session.id,)).log("delete session query"));
+        redirect("/login")
+    }
+}
+
+#[derive(Serialize, Deserialize, Default)]
+struct Signup {
+    name: String,
+    new_password: String,
+    confirm_password: String,
+    #[serde(default)]
+    confirm_no_password: bool,
+    #[serde(skip)]
+    error: Option<&'static str>,
+}
+
+impl PublicPage for Signup {
+    fn render_to_buf(self, buf: &mut String) {
+        let Signup { name, new_password, confirm_password, confirm_no_password, error } = self;
+        let vals = if confirm_no_password { "{\"confirm_no_password\":true}" } else { "{}" };
+        write_html! { (buf)
+            <form "hx-post"="/signup" "hx-swap"="outerHTML" "hx-vals"=vals>
+                if let Some(e) = error { <div class="error">e</div> }
+                <input name="name" type="text" autocomplete="name" placeholder="name" value=name
+                    maxlength=MAX_NAME_LENGTH required>
+                <input name="new_password" type="password" autocomplete="new-password" placeholder="new password"
+                    value=new_password>
+                <input name="confirm_password" type="password" autocomplete="confirm-password"
+                    placeholder="confirm password" value=confirm_password>
+                <input type="submit" value="submit">
+            </form>
+        }
+    }
+}
+
+impl Signup {
+    async fn post(axum::Form(mut data): axum::Form<Self>) -> Result<Redirect, Html<String>> {
+        if data.name.len() > MAX_NAME_LENGTH {
+            data.error = Some(formatcp!("name too long, max length is {MAX_NAME_LENGTH}"));
+            return Err(data.render());
+        }
+
+        if !data.confirm_no_password && data.new_password.is_empty() {
+            data.confirm_no_password = true;
+            data.error = Some("Are you sure you don't want to use a password? (then submit again)");
+            return Err(data.render());
+        }
+
+        db::with(|db| {
+            // TODO: hash passwords
+            match db.register.insert((&data.name, hash_password(&data.new_password))) {
+                Ok(_) => {}
+                Err(rusqlite::Error::SqliteFailure(e, _))
+                    if e.code == rusqlite::ErrorCode::ConstraintViolation =>
+                {
+                    data.error = Some("username already taken");
+                }
+                Err(e) => {
+                    log::error!("create user query: {e}");
+                    data.error = Some("internal server error");
+                }
+            };
+        });
+
+        if data.error.is_some() {
+            Err(data.render())
+        } else {
+            Ok(redirect("/login"))
+        }
+    }
+}
+
+fn base(body: impl FnOnce(&mut String), session: Option<&Session>) -> Html<String> {
+    let username = session.map(|s| &s.name);
+
+    let nav_button = |f: &mut String, name: &str| {
+        write_html! {(f)
+            <button "hx-push-url"={format_args!("/{name}")}
+                "hx-get"={format_args!("/{name}-view")}
+                "hx-target"="main"
+                "hx-swap"="innerHTML">name</button>
+        }
+    };
+
+    Html(html! {
+        "<!DOCTYPE html>"
+        <html lang="en">
+            <head>
+                <meta name="charset" content="UTF-8">
+                <meta name="viewport" content="width=device-width, initial-scale=1">
+                <link rel="stylesheet" href="/index.css">
+            </head>
+            <body>
+                <nav>
+                    <button "hx-push-url"="/" "hx-get"="/index-view" "hx-target"="main" "hx-swap"="innerHTML">"depell"</button>
+                    <section>
+                        if let Some(username) = username {
+                            <button "hx-push-url"={format_args!("/profile/{username}")} "hx-get"="/profile-view" "hx-target"="main"
+                                "hx-swap"="innerHTML">username</button>
+                            |f|{nav_button(f, "feed"); nav_button(f, "post")}
+                            <button "hx-delete"="/login">"logout"</button>
+                        } else {
+                            |f|{nav_button(f, "login"); nav_button(f, "signup")}
+                        }
+                    </section>
+                </nav>
+                <section id="post-form"></section>
+                <main>|f|{body(f)}</main>
+            </body>
+            <script src="https://unpkg.com/htmx.org@2.0.3/dist/htmx.min.js" integrity="sha384-0895/pl2MU10Hqc6jd4RvrthNlDiE9U1tWmX7WRESftEDRosgxNsQG/Ze9YMRzHq" crossorigin="anonymous"></script>
+            <script type="module" src="/index.js"></script>
+        </html>
+    })
+}
+
+struct Session {
+    name: String,
+    id: [u8; 32],
+}
+
+#[axum::async_trait]
+impl<S> axum::extract::FromRequestParts<S> for Session {
+    /// If the extractor fails it'll use this "rejection" type. A rejection is
+    /// a kind of error that can be converted into a response.
+    type Rejection = Redirect;
+
+    /// Perform the extraction.
+    async fn from_request_parts(parts: &mut Parts, _: &S) -> Result<Self, Self::Rejection> {
+        let err = redirect("/login");
+
+        let value = parts
+            .headers
+            .get_all(COOKIE)
+            .into_iter()
+            .find_map(|c| c.to_str().ok()?.trim().strip_prefix("id="))
+            .map(|c| c.split_once(';').unwrap_or((c, "")).0)
+            .ok_or(err)?;
+        let mut id = [0u8; 32];
+        parse_hex(value, &mut id).ok_or(err)?;
+
+        let (name, expiration) = db::with(|db| {
+            db.get_session
+                .query_row((id,), |r| Ok((r.get::<_, String>(0)?, r.get::<_, u64>(1)?)))
+                .log("fetching session")
+                .ok_or(err)
+        })?;
+
+        if expiration < now() {
+            return Err(err);
+        }
+
+        Ok(Self { name, id })
+    }
+}
+
+fn now() -> u64 {
+    std::time::SystemTime::now()
+        .duration_since(std::time::SystemTime::UNIX_EPOCH)
+        .unwrap()
+        .as_secs()
+}
+
+fn parse_hex(hex: &str, dst: &mut [u8]) -> Option<()> {
+    fn hex_to_nibble(b: u8) -> Option<u8> {
+        Some(match b {
+            b'a'..=b'f' => b - b'a' + 10,
+            b'A'..=b'F' => b - b'A' + 10,
+            b'0'..=b'9' => b - b'0',
+            _ => return None,
+        })
+    }
+
+    if hex.len() != dst.len() * 2 {
+        return None;
+    }
+
+    for (d, p) in dst.iter_mut().zip(hex.as_bytes().chunks_exact(2)) {
+        *d = (hex_to_nibble(p[0])? << 4) | hex_to_nibble(p[1])?;
+    }
+
+    Some(())
+}
+
+fn to_hex(src: &[u8]) -> String {
+    use std::fmt::Write;
+    let mut buf = String::new();
+    for &b in src {
+        write!(buf, "{b:02x}").unwrap()
+    }
+    buf
+}
+
+fn main() {
+    tokio::runtime::Builder::new_current_thread().enable_all().build().unwrap().block_on(amain());
+}
+
+mod db {
+    use std::cell::RefCell;
+
+    macro_rules! gen_queries {
+        ($vis:vis struct $name:ident {
+            $($qname:ident: $code:expr,)*
+        }) => {
+            $vis struct $name<'a> {
+                $($vis $qname: rusqlite::Statement<'a>,)*
+            }
+
+            impl<'a> $name<'a> {
+                fn new(db: &'a rusqlite::Connection) -> Self {
+                    Self {
+                        $($qname: db.prepare($code).unwrap(),)*
+                    }
+                }
+            }
+        };
+    }
+
+    gen_queries! {
+        pub struct Queries {
+            register: "INSERT INTO user (name, password_hash) VALUES(?, ?)",
+            change_passowrd: "UPDATE user SET password_hash = ? WHERE name = ?",
+            authenticate: "SELECT name, password_hash FROM user WHERE name = ?",
+            login: "INSERT OR REPLACE INTO session (id, username, expiration) VALUES(?, ?, ?)",
+            logout: "DELETE FROM session WHERE id = ?",
+            get_session: "SELECT username, expiration FROM session WHERE id = ?",
+            get_user_posts: "SELECT author, name, timestamp, code FROM post WHERE author = ?
+                ORDER BY timestamp DESC",
+            get_pots_before: "SELECT author, name, timestamp, code FROM post WHERE timestamp < ?",
+            create_post: "INSERT INTO post (name, author, timestamp, code) VALUES(?, ?, ?, ?)",
+            fetch_deps: "
+                WITH RECURSIVE roots(name, author, code) AS (
+                    SELECT name, author, code FROM post WHERE name = ? AND author = ?
+                    UNION
+                    SELECT post.name, post.author, post.code FROM
+                        post JOIN import ON post.name = import.to_name
+                            AND post.author = import.to_author
+                        JOIN roots ON import.from_name = roots.name
+                            AND import.from_author = roots.author
+                ) SELECT * FROM roots;
+            ",
+            create_import: "INSERT INTO import(to_author, to_name, from_author, from_name)
+                VALUES(?, ?, ?, ?)",
+        }
+    }
+
+    struct Db {
+        queries: Queries<'static>,
+        _db: Box<rusqlite::Connection>,
+    }
+
+    impl Db {
+        fn new() -> Self {
+            let db = Box::new(rusqlite::Connection::open("db.sqlite").unwrap());
+            Self {
+                queries: Queries::new(unsafe {
+                    std::mem::transmute::<&rusqlite::Connection, &rusqlite::Connection>(&db)
+                }),
+                _db: db,
+            }
+        }
+    }
+
+    pub fn with<T>(with: impl FnOnce(&mut Queries) -> T) -> T {
+        thread_local! { static DB_CONN: RefCell<Db> = RefCell::new(Db::new()); }
+        DB_CONN.with_borrow_mut(|q| with(&mut q.queries))
+    }
+
+    pub fn init() {
+        let db = rusqlite::Connection::open("db.sqlite").unwrap();
+        db.execute_batch(include_str!("schema.sql")).unwrap();
+        Queries::new(&db);
+    }
+}
+
+fn redirect(to: &'static str) -> Redirect {
+    AppendHeaders([("hx-location", to)])
+}
+
+trait PublicPage: Default {
+    fn render_to_buf(self, buf: &mut String);
+
+    fn render(self) -> Html<String> {
+        let mut str = String::new();
+        self.render_to_buf(&mut str);
+        Html(str)
+    }
+
+    async fn get() -> Html<String> {
+        Self::default().render()
+    }
+
+    async fn page(session: Option<Session>) -> Html<String> {
+        base(|s| Self::default().render_to_buf(s), session.as_ref())
+    }
+}
+
+trait Page: Default {
+    fn render_to_buf(self, session: &Session, buf: &mut String);
+
+    fn render(self, session: &Session) -> Html<String> {
+        let mut str = String::new();
+        self.render_to_buf(session, &mut str);
+        Html(str)
+    }
+
+    async fn get(session: Session) -> Html<String> {
+        Self::default().render(&session)
+    }
+
+    async fn page(session: Option<Session>) -> Result<Html<String>, axum::response::Redirect> {
+        match session {
+            Some(session) => {
+                Ok(base(|f| Self::default().render_to_buf(&session, f), Some(&session)))
+            }
+            None => Err(axum::response::Redirect::permanent("/login")),
+        }
+    }
+}
+
+trait ResultExt<O, E> {
+    fn log(self, prefix: impl Display) -> Option<O>;
+}
+
+impl<O, E: Display> ResultExt<O, E> for Result<O, E> {
+    fn log(self, prefix: impl Display) -> Option<O> {
+        match self {
+            Ok(v) => Some(v),
+            Err(e) => {
+                log::error!("{prefix}: {e}");
+                None
+            }
+        }
+    }
+}
+
+struct Logger;
+
+impl log::Log for Logger {
+    fn enabled(&self, _: &log::Metadata) -> bool {
+        true
+    }
+
+    fn log(&self, record: &log::Record) {
+        if self.enabled(record.metadata()) {
+            eprintln!("{} - {}", record.module_path().unwrap_or("=="), record.args());
+        }
+    }
+
+    fn flush(&self) {}
+}

depell/src/post-page.html

@@ -0,0 +1,21 @@
+<div id="dep-list">
+	<input placeholder="search impoted deps.." oninput="filterCodeDeps(this, event)">
+	<section id="deps">
+		results show here...
+	</section>
+</div>
+
+<div>
+
+	<h3>About posting code</h3>
+	<p>
+		If you are unfammiliar with <a href="https://git.ablecorp.us/AbleOS/holey-bytes">hblang</a>, refer to the
+		<strong>hblang/README.md</strong> or
+		vizit <a href="/profile/mlokis">mlokis'es posts</a>. Preferably don't edit the code here.
+	</p>
+
+	<h3>Extra textarea features</h3>
+	<ul>
+		<li>proper tab behaviour</li>
+		<li>snap to previous tab boundary on "empty" lines</li>
+	</ul>

depell/src/schema.sql

@@ -0,0 +1,55 @@
+PRAGMA foreign_keys = ON;
+
+CREATE TABLE IF NOT EXISTS user(
+	name TEXT NOT NULL,
+	password_hash TEXT NOT NULL,
+	PRIMARY KEY (name)
+) WITHOUT ROWID;
+
+CREATE TABLE IF NOT EXISTS session(
+	id BLOB NOT NULL,
+	username TEXT NOT NULL,
+	expiration INTEGER NOT NULL,
+	FOREIGN KEY (username) REFERENCES user (name)
+	PRIMARY KEY (username)
+) WITHOUT ROWID;
+
+CREATE UNIQUE INDEX IF NOT EXISTS
+	session_id ON session (id);
+
+CREATE TABLE IF NOT EXISTS post(
+	name TEXT NOT NULL,
+	author TEXT,
+	timestamp INTEGER,
+	code TEXT NOT NULL,
+	FOREIGN KEY (author) REFERENCES user(name) ON DELETE SET NULL,
+	PRIMARY KEY (author, name)
+);
+
+CREATE INDEX IF NOT EXISTS
+	post_timestamp ON post(timestamp DESC);
+
+CREATE TABLE IF NOT EXISTS import(
+	from_name TEXT NOT NULL,
+	from_author TEXT,
+	to_name TEXT NOT NULL,
+	to_author TEXT,
+	FOREIGN KEY (from_name, from_author) REFERENCES post(name, author),
+	FOREIGN KEY (to_name, to_author) REFERENCES post(name, author)
+);
+
+CREATE INDEX IF NOT EXISTS
+	dependencies ON import(from_name, from_author);
+
+CREATE INDEX IF NOT EXISTS
+	dependants ON import(to_name, to_author);
+
+CREATE TABLE IF NOT EXISTS run(
+	code_name TEXT NOT NULL,
+	code_author TEXT NOT NULL,
+	runner TEXT NOT NULL,
+	FOREIGN KEY (code_name, code_author) REFERENCES post(name, author),
+	FOREIGN KEY (runner) REFERENCES user(name),
+	PRIMARY KEY (code_name, code_author, runner)
+);
+

depell/src/welcome-page.html

@@ -0,0 +1,17 @@
+<h1>Welcome to depell</h1>
+<p>
+	Depell (dependency hell) is a simple "social" media site best compared to twitter, except that all you can post is
+	<a href="https://git.ablecorp.us/AbleOS/holey-bytes">hblang</a> code with no comments allowed. Instead of likes you
+	run the program, and instead of retweets you import the program as dependency. Run counts even when ran indirectly.
+</p>
+
+<p>
+	The backend only serves the code and frontend compiles and runs it locally. All posts are immutable.
+</p>
+
+<h2>Security?</h2>
+<p>
+	All code runs in WASM (inside a holey-bytes VM until hblang compiles to wasm) and is controlled by JavaScript. WASM
+	cant do any form of IO without going trough JavaScript so as long as JS import does not allow wasm to execute
+	arbitrary JS code, WASM can act as a container inside the JS.
+</p>

depell/wasm-fmt/Cargo.toml

@@ -0,0 +1,11 @@
+[package]
+name = "wasm-hbfmt"
+version = "0.1.0"
+edition = "2021"
+
+[lib]
+crate-type = ["cdylib"]
+
+[dependencies]
+hblang = { workspace = true, features = ["no_log"] }
+wasm-rt = { version = "0.1.0", path = "../wasm-rt" }

depell/wasm-fmt/src/lib.rs

@@ -0,0 +1,42 @@
+#![no_std]
+#![feature(str_from_raw_parts)]
+#![feature(alloc_error_handler)]
+
+use hblang::{fmt, parser};
+
+wasm_rt::decl_runtime!(128 * 1024, 1024 * 4);
+
+const MAX_OUTPUT_SIZE: usize = 1024 * 10;
+wasm_rt::decl_buffer!(MAX_OUTPUT_SIZE, MAX_OUTPUT, OUTPUT, OUTPUT_LEN);
+
+const MAX_INPUT_SIZE: usize = 1024 * 4;
+wasm_rt::decl_buffer!(MAX_INPUT_SIZE, MAX_INPUT, INPUT, INPUT_LEN);
+
+#[no_mangle]
+unsafe extern "C" fn fmt() {
+    ALLOCATOR.reset();
+
+    let code = core::str::from_raw_parts(core::ptr::addr_of!(INPUT).cast(), INPUT_LEN);
+
+    let arena = parser::Arena::with_capacity(code.len() * parser::SOURCE_TO_AST_FACTOR);
+    let mut ctx = parser::Ctx::default();
+    let exprs = parser::Parser::parse(&mut ctx, code, "source.hb", &mut parser::no_loader, &arena);
+
+    let mut f = wasm_rt::Write(&mut OUTPUT[..]);
+    fmt::fmt_file(exprs, code, &mut f).unwrap();
+    OUTPUT_LEN = MAX_OUTPUT_SIZE - f.0.len();
+}
+
+#[no_mangle]
+unsafe extern "C" fn tok() {
+    let code = core::slice::from_raw_parts_mut(
+        core::ptr::addr_of_mut!(OUTPUT).cast(), OUTPUT_LEN);
+    OUTPUT_LEN = fmt::get_token_kinds(code);
+}
+
+#[no_mangle]
+unsafe extern "C" fn minify() {
+    let code = core::str::from_raw_parts_mut(
+        core::ptr::addr_of_mut!(OUTPUT).cast(), OUTPUT_LEN);
+    OUTPUT_LEN = fmt::minify(code);
+}

depell/wasm-hbc/Cargo.toml

@@ -0,0 +1,14 @@
+[package]
+name = "wasm-hbc"
+version = "0.1.0"
+edition = "2021"
+
+[lib]
+crate-type = ["cdylib"]
+
+[dependencies]
+hblang = { workspace = true, features = [] }
+hbvm.workspace = true
+log = { version = "0.4.22", features = ["release_max_level_error"] }
+wasm-rt = { version = "0.1.0", path = "../wasm-rt", features = ["log"] }
+

depell/wasm-hbc/src/lib.rs

@@ -0,0 +1,119 @@
+#![feature(alloc_error_handler)]
+#![feature(slice_take)]
+#![no_std]
+
+use {
+    alloc::{string::String, vec::Vec},
+    hblang::{
+        son::{hbvm::HbvmBackend, Codegen, CodegenCtx},
+        ty::Module,
+        Ent,
+    },
+};
+
+extern crate alloc;
+
+const ARENA_CAP: usize = 128 * 16 * 1024;
+wasm_rt::decl_runtime!(ARENA_CAP, 1024 * 4);
+
+const MAX_INPUT_SIZE: usize = 32 * 4 * 1024;
+wasm_rt::decl_buffer!(MAX_INPUT_SIZE, MAX_INPUT, INPUT, INPUT_LEN);
+
+#[no_mangle]
+unsafe fn compile_and_run(mut fuel: usize) {
+    ALLOCATOR.reset();
+
+    _ = log::set_logger(&wasm_rt::Logger);
+    log::set_max_level(log::LevelFilter::Error);
+
+    struct File<'a> {
+        path: &'a str,
+        code: &'a mut str,
+    }
+
+    let mut root = 0;
+
+    let files = {
+        let mut input_bytes =
+            core::slice::from_raw_parts_mut(core::ptr::addr_of_mut!(INPUT).cast::<u8>(), INPUT_LEN);
+
+        let mut files = Vec::with_capacity(32);
+        while let Some((&mut path_len, rest)) = input_bytes.split_first_chunk_mut() {
+            let (path, rest) = rest.split_at_mut(u16::from_le_bytes(path_len) as usize);
+            let (&mut code_len, rest) = rest.split_first_chunk_mut().unwrap();
+            let (code, rest) = rest.split_at_mut(u16::from_le_bytes(code_len) as usize);
+            files.push(File {
+                path: core::str::from_utf8_unchecked(path),
+                code: core::str::from_utf8_unchecked_mut(code),
+            });
+            input_bytes = rest;
+        }
+
+        let root_path = files[root].path;
+        hblang::quad_sort(&mut files, |a, b| a.path.cmp(b.path));
+        root = files.binary_search_by_key(&root_path, |p| p.path).unwrap();
+
+        files
+    };
+
+    let mut ctx = CodegenCtx::default();
+
+    let files = {
+        let paths = files.iter().map(|f| f.path).collect::<Vec<_>>();
+        let mut loader = |path: &str, _: &str, kind| match kind {
+            hblang::parser::FileKind::Module => Ok(paths.binary_search(&path).unwrap()),
+            hblang::parser::FileKind::Embed => Err("embeds are not supported".into()),
+        };
+        files
+            .into_iter()
+            .map(|f| {
+                hblang::parser::Ast::new(
+                    f.path,
+                    // since 'free' does nothing this is fine
+                    String::from_raw_parts(f.code.as_mut_ptr(), f.code.len(), f.code.len()),
+                    &mut ctx.parser,
+                    &mut loader,
+                )
+            })
+            .collect::<Vec<_>>()
+    };
+
+    let mut ct = {
+        let mut backend = HbvmBackend::default();
+        Codegen::new(&mut backend, &files, &mut ctx).generate(Module::new(root));
+
+        if !ctx.parser.errors.borrow().is_empty() {
+            log::error!("{}", ctx.parser.errors.borrow());
+            return;
+        }
+
+        let mut c = Codegen::new(&mut backend, &files, &mut ctx);
+        c.assemble_comptime()
+    };
+
+    while fuel != 0 {
+        match ct.vm.run() {
+            Ok(hbvm::VmRunOk::End) => {
+                log::error!("exit code: {}", ct.vm.read_reg(1).0 as i64);
+                break;
+            }
+            Ok(hbvm::VmRunOk::Ecall) => {
+                let unknown = ct.vm.read_reg(2).0;
+                log::error!("unknown ecall: {unknown}")
+            }
+            Ok(hbvm::VmRunOk::Timer) => {
+                fuel -= 1;
+                if fuel == 0 {
+                    log::error!("program timed out");
+                }
+            }
+            Ok(hbvm::VmRunOk::Breakpoint) => todo!(),
+            Err(e) => {
+                log::error!("vm error: {e}");
+                break;
+            }
+        }
+    }
+
+    //log::error!("memory consumption: {}b / {}b", ALLOCATOR.used(), ARENA_CAP);
+}

depell/wasm-rt/Cargo.toml

@@ -0,0 +1,7 @@
+[package]
+name = "wasm-rt"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]
+log = { version = "0.4.22", optional = true }

depell/wasm-rt/src/lib.rs

@@ -0,0 +1,162 @@
+#![feature(alloc_error_handler)]
+#![feature(pointer_is_aligned_to)]
+#![feature(slice_take)]
+#![no_std]
+
+use core::{
+    alloc::{GlobalAlloc, Layout},
+    cell::UnsafeCell,
+};
+
+extern crate alloc;
+
+#[macro_export]
+macro_rules! decl_buffer {
+    ($cap:expr, $export_cap:ident, $export_base:ident, $export_len:ident) => {
+        #[no_mangle]
+        static $export_cap: usize = $cap;
+        #[no_mangle]
+        static mut $export_base: [u8; $cap] = [0; $cap];
+        #[no_mangle]
+        static mut $export_len: usize = 0;
+    };
+}
+
+#[macro_export]
+macro_rules! decl_runtime {
+    ($memory_size:expr, $max_panic_size:expr) => {
+        #[cfg(debug_assertions)]
+        #[no_mangle]
+        static mut PANIC_MESSAGE: [u8; $max_panic_size] = [0; $max_panic_size];
+        #[cfg(debug_assertions)]
+        #[no_mangle]
+        static mut PANIC_MESSAGE_LEN: usize = 0;
+
+        #[cfg(target_arch = "wasm32")]
+        #[panic_handler]
+        pub fn handle_panic(_info: &core::panic::PanicInfo) -> ! {
+            #[cfg(debug_assertions)]
+            {
+                unsafe {
+                    use core::fmt::Write;
+                    let mut f = $crate::Write(&mut PANIC_MESSAGE[..]);
+                    _ = writeln!(f, "{}", _info);
+                    PANIC_MESSAGE_LEN = $max_panic_size - f.0.len();
+                }
+            }
+
+            core::arch::wasm32::unreachable();
+        }
+
+        #[global_allocator]
+        static ALLOCATOR: $crate::ArenaAllocator<{ $memory_size }> = $crate::ArenaAllocator::new();
+
+        #[cfg(target_arch = "wasm32")]
+        #[alloc_error_handler]
+        fn alloc_error(_: core::alloc::Layout) -> ! {
+            #[cfg(debug_assertions)]
+            {
+                unsafe {
+                    use core::fmt::Write;
+                    let mut f = $crate::Write(&mut PANIC_MESSAGE[..]);
+                    _ = writeln!(f, "out of memory");
+                    PANIC_MESSAGE_LEN = $max_panic_size - f.0.len();
+                }
+            }
+
+            core::arch::wasm32::unreachable()
+        }
+    };
+}
+
+#[cfg(feature = "log")]
+pub struct Logger;
+
+#[cfg(feature = "log")]
+impl log::Log for Logger {
+    fn enabled(&self, _: &log::Metadata) -> bool {
+        true
+    }
+
+    fn log(&self, record: &log::Record) {
+        if self.enabled(record.metadata()) {
+            const MAX_LOG_MESSAGE: usize = 1024 * 8;
+            #[no_mangle]
+            static mut LOG_MESSAGES: [u8; MAX_LOG_MESSAGE] = [0; MAX_LOG_MESSAGE];
+            #[no_mangle]
+            static mut LOG_MESSAGES_LEN: usize = 0;
+
+            unsafe {
+                use core::fmt::Write;
+                let mut f = Write(&mut LOG_MESSAGES[LOG_MESSAGES_LEN..]);
+                _ = writeln!(f, "{}", record.args());
+                LOG_MESSAGES_LEN = MAX_LOG_MESSAGE - f.0.len();
+            }
+        }
+    }
+
+    fn flush(&self) {}
+}
+
+pub struct ArenaAllocator<const SIZE: usize> {
+    arena: UnsafeCell<[u8; SIZE]>,
+    head: UnsafeCell<*mut u8>,
+}
+
+impl<const SIZE: usize> ArenaAllocator<SIZE> {
+    #[expect(clippy::new_without_default)]
+    pub const fn new() -> Self {
+        ArenaAllocator {
+            arena: UnsafeCell::new([0; SIZE]),
+            head: UnsafeCell::new(core::ptr::null_mut()),
+        }
+    }
+
+    #[expect(clippy::missing_safety_doc)]
+    pub unsafe fn reset(&self) {
+        (*self.head.get()) = self.arena.get().cast::<u8>().add(SIZE);
+    }
+
+    pub fn used(&self) -> usize {
+        unsafe { self.arena.get() as usize + SIZE - (*self.head.get()) as usize }
+    }
+}
+
+unsafe impl<const SIZE: usize> Sync for ArenaAllocator<SIZE> {}
+
+unsafe impl<const SIZE: usize> GlobalAlloc for ArenaAllocator<SIZE> {
+    unsafe fn alloc(&self, layout: Layout) -> *mut u8 {
+        let size = layout.size();
+        let align = layout.align();
+
+        let until = self.arena.get() as *mut u8;
+
+        let new_head = (*self.head.get()).sub(size);
+        let aligned_head = (new_head as usize & !(align - 1)) as *mut u8;
+        debug_assert!(aligned_head.is_aligned_to(align));
+
+        if until > aligned_head {
+            return core::ptr::null_mut();
+        }
+
+        *self.head.get() = aligned_head;
+        aligned_head
+    }
+
+    unsafe fn dealloc(&self, _ptr: *mut u8, _layout: Layout) {
+        /* lol */
+    }
+}
+
+pub struct Write<'a>(pub &'a mut [u8]);
+
+impl core::fmt::Write for Write<'_> {
+    fn write_str(&mut self, s: &str) -> core::fmt::Result {
+        if let Some(m) = self.0.take_mut(..s.len()) {
+            m.copy_from_slice(s.as_bytes());
+            Ok(())
+        } else {
+            Err(core::fmt::Error)
+        }
+    }
+}

examples/asm/hello-name.hba

@@ -0,0 +1,100 @@
+jmp entry
+
+puts:
+-- Write string to console
+--  r2: [IN] *const u8  String pointer
+--  r3: [IN] usize      String length
+
+    li8     r1, 0x1         -- Write syscall
+    brc     r2, r3, 2       -- Copy parameters
+    li8     r2, 0x1         -- STDOUT
+    eca
+
+    jal     r0, r31, 0
+
+gets:
+-- Read string until end of buffer or LF
+--  r2: [IN] *mut u8    Buffer
+--  r3: [IN] usize      Buffer length
+
+    -- Register allocations:
+    --  r33: *mut u8    Buffer end
+    --  r34: u8         Immediate char
+    --  r35: u8         Const [0x0A = LF]
+
+    li8     r35, 0x0A
+    add64   r33, r2, r3
+
+    -- Setup syscall
+    li8     r2, 0x1         -- Stdin
+    cp      r3, r2
+    li8     r4, 0x1         -- Read one char
+
+    jeq     r3, r33, end
+    loop:
+        li8     r1, 0x1     -- Read syscall
+        eca
+        addi64  r3, r3, 1
+        ld      r34, r3, 0, 1
+        jeq     r34, r35, end
+        jne     r3, r33, loop
+    
+    end:
+        -- Set copied amount
+        sub64   r1, r33, r3
+        addi64  r1, -1
+        jal     r0, r31, 0
+
+alloc-pages:
+-- Allocate pages
+--  r1: [OUT] *mut u8   Pointer to page
+--  r2: [IN]  u16       Page count
+
+    muli16  r3, r2, 4096    -- page count
+    li8     r1, 0x9         -- mmap syscall
+    li8     r2, 0x0         -- no address set, kernel chosen
+    li8     r4, 0x2         -- PROT_WRITE
+    li8     r5, 0x20        -- MAP_ANONYMOUS
+    li64    r6, -1          -- Doesn't map file
+    li8     r7, 0x0         -- Doesn't map file
+    eca
+
+    jal     r0, r31, 0
+
+entry:
+-- Program entrypoint
+
+    -- Register allocations:
+    -- r32: *mut u8     Buffer
+    -- r36: usize       Read buffer length
+
+    -- Allocate one page (4096 KiB)
+    li8     r2, 1
+    jal     r31, 0, alloc-pages
+    cp      r32, r1
+
+    -- Print message
+    lra16   r2, r0, #enter-your-name
+    li8     r3, 17
+    jal     r31, r0, puts
+
+    -- Read name
+    cp      r2, r32
+    li16    r3, 4096
+    jal     r31, r0, gets
+    cp      r36, r1
+
+    -- Print your name is
+    lra16   r2, r0, #your-name-is
+    li8     r3, 15
+    jal     r31, r0, puts
+
+    -- And now print the name
+    cp      r2, r32
+    cp      r3, r36
+    jal     r31, r0, puts
+
+    tx
+
+#enter-your-name: "Enter your name: "
+#your-name-is   : "\nYour name is: "

hbasm/Cargo.toml

@@ -1,20 +0,0 @@
-[package]
-name = "hbasm"
-version = "0.1.0"
-edition = "2021"
-
-[dependencies]
-hbbytecode = { path = "../hbbytecode" }
-paste = "1.0"
-hashbrown = "0.14.0"
-ariadne = "0.3.0"
-
-[dependencies.lasso]
-version = "0.7"
-default-features = false
-features = ["no-std"]
-
-[dependencies.logos]
-version = "0.13"
-default-features = false
-features = ["export_derive"]

hbasm/assets/serial_driver.hbasm

@@ -1,18 +0,0 @@
-jmp r0, start
-start:
-    jmp r0, init_serial_port
-
--- Uses r20 to set the port
-init_serial_port:
-    add r20, r30, r10
-    li r20, 00
-
-   -- outb(PORT + 1, 0x00);    // Disable all interrupts
-   -- outb(PORT + 3, 0x80);    // Enable DLAB (set baud rate divisor)
-   -- outb(PORT + 0, 0x03);    // Set divisor to 3 (lo byte) 38400 baud
-   -- outb(PORT + 1, 0x00);    //                  (hi byte)
-   -- outb(PORT + 3, 0x03);    // 8 bits, no parity, one stop bit
-   -- outb(PORT + 2, 0xC7);    // Enable FIFO, clear them, with 14-byte threshold
-   -- outb(PORT + 4, 0x0B);    // IRQs enabled, RTS/DSR set
-   -- outb(PORT + 4, 0x1E);    // Set in loopback mode, test the serial chip
-   -- outb(PORT + 0, 0xAE);    // Test serial chip (send byte 0xAE and check if serial returns same byte)

hbasm/src/lib.rs

@@ -1,65 +0,0 @@
-#![no_std]
-#![feature(error_in_core)]
-
-extern crate alloc;
-
-pub mod text;
-
-mod macros;
-
-use {alloc::vec::Vec, hashbrown::HashSet};
-
-#[derive(Default)]
-pub struct Assembler {
-    pub buf: Vec<u8>,
-    sub:     HashSet<usize>,
-}
-
-impl Assembler {
-    macros::impl_asm!(
-        bbbb(p0: u8, p1: u8, p2: u8, p3: u8)
-            => [DIR, DIRF, FMAF],
-        bbb(p0: u8, p1: u8, p2: u8)
-            => [ADD, SUB, MUL, AND, OR, XOR, SL, SR, SRS, CMP, CMPU, BRC, ADDF, SUBF, MULF],
-        bbdh(p0: u8, p1: u8, p2: impl Imm, p3: u16)
-            => [LD, ST],
-        bbd(p0: u8, p1: u8, p2: impl Imm)
-            => [ADDI, MULI, ANDI, ORI, XORI, SLI, SRI, SRSI, CMPI, CMPUI,
-                BMC, JEQ, JNE, JLT, JGT, JLTU, JGTU, ADDFI, MULFI],
-        bb(p0: u8, p1: u8)
-            => [NEG, NOT, CP, SWA, NEGF, ITF, FTI],
-        bd(p0: u8, p1: impl Imm)
-            => [LI, JMP],
-        n()
-            => [NOP, ECALL],
-    );
-}
-
-pub trait Imm {
-    fn insert(&self, asm: &mut Assembler);
-}
-
-macro_rules! impl_imm_le_bytes {
-    ($($ty:ty),* $(,)?) => {
-        $(
-            impl Imm for $ty {
-                #[inline(always)]
-                fn insert(&self, asm: &mut Assembler) {
-                    asm.buf.extend(self.to_le_bytes());
-                }
-            }
-        )*
-    };
-}
-
-impl_imm_le_bytes!(u64, i64, f64);
-
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
-pub struct Symbol(pub u64);
-impl Imm for Symbol {
-    #[inline(always)]
-    fn insert(&self, asm: &mut Assembler) {
-        asm.sub.insert(asm.buf.len());
-        asm.buf.extend(self.0.to_le_bytes());
-    }
-}

hbasm/src/macros.rs

@@ -1,72 +0,0 @@
-macro_rules! impl_asm_opcodes {
-    (
-        $generic:ident
-        ($($param_i:ident: $param_ty:ty),*)
-        => []
-    ) => {};
-
-    (
-        $generic:ident
-        ($($param_i:ident: $param_ty:ty),*)
-        => [$opcode:ident, $($rest:tt)*]
-    ) => {
-        paste::paste! {
-            #[allow(dead_code)]
-            #[inline(always)]
-            pub fn [<i_ $opcode:lower>](&mut self, $($param_i: $param_ty),*) {
-                self.$generic(hbbytecode::opcode::$opcode, $($param_i),*)
-            }
-        }
-
-        macros::impl_asm_opcodes!(
-            $generic($($param_i: $param_ty),*)
-            => [$($rest)*]
-        );
-    };
-}
-
-macro_rules! gen_impl_asm_insert {
-    ($($ty:ident),* $(,)?) => {
-        macro_rules! impl_asm_insert {
-            $(($self:expr, $id:ident, $ty) => {
-                $self.buf.extend($id.to_le_bytes())
-            };)*
-
-            ($self:expr, $id:ident, $_:ty) => {
-                Imm::insert(&$id, $self)
-            };
-        }
-    };
-}
-
-gen_impl_asm_insert!(u8, u16, u64);
-
-macro_rules! impl_asm {
-    (
-        $(
-            $ityn:ident
-            ($($param_i:ident: $param_ty:ty),* $(,)?)
-            => [$($opcode:ident),* $(,)?],
-        )*
-    ) => {
-        paste::paste! {
-            $(
-                #[allow(dead_code)]
-                fn [<i_param_ $ityn>](&mut self, opcode: u8, $($param_i: $param_ty),*) {
-                    self.buf.push(opcode);
-                    $(macros::impl_asm_insert!(self, $param_i, $param_ty);)*
-                }
-
-                macros::impl_asm_opcodes!(
-                    [<i_param_ $ityn>]($($param_i: $param_ty),*)
-                    => [$($opcode,)*]
-                );
-            )*
-        }
-    };
-}
-
-pub(super) use {impl_asm, impl_asm_opcodes};
-
-#[allow(clippy::single_component_path_imports)]
-pub(super) use impl_asm_insert;

hbasm/src/main.rs

@@ -1,46 +0,0 @@
-use {
-    ariadne::{ColorGenerator, Label, Report, ReportKind, Source},
-    std::{
-        error::Error,
-        io::{stdin, Read},
-    },
-};
-
-fn main() -> Result<(), Box<dyn Error>> {
-    let mut code = String::new();
-    stdin().read_to_string(&mut code)?;
-
-    let mut buf = vec![];
-
-    if let Err(e) = hbasm::text::assembly(&code, &mut buf) {
-        let mut colors = ColorGenerator::new();
-
-        let e_code = match e.kind {
-            hbasm::text::ErrorKind::UnexpectedToken => 1,
-            hbasm::text::ErrorKind::InvalidToken => 2,
-            hbasm::text::ErrorKind::UnexpectedEnd => 3,
-            hbasm::text::ErrorKind::InvalidSymbol => 4,
-        };
-        let message = match e.kind {
-            hbasm::text::ErrorKind::UnexpectedToken => "This token is not expected!",
-            hbasm::text::ErrorKind::InvalidToken => "The token is not valid!",
-            hbasm::text::ErrorKind::UnexpectedEnd => "The assembler reached the end of input unexpectedly!",
-            hbasm::text::ErrorKind::InvalidSymbol => "This referenced symbol doesn't have a corresponding label!",
-        };
-        let a = colors.next();
-
-        Report::build(ReportKind::Error, "engine_internal", e.span.clone().start)
-        .with_code(e_code)
-        .with_message(format!("{:?}", e.kind))
-        .with_label(
-            Label::new(("engine_internal", e.span.clone()))
-                .with_message(message)
-                .with_color(a),
-        )
-        .finish()
-        .eprint(("engine_internal", Source::from(&code)))
-        .unwrap();
-    }
-
-    Ok(())
-}

hbasm/src/text.rs

@@ -1,273 +0,0 @@
-extern crate alloc;
-use alloc::vec::Vec;
-
-use {
-    core::fmt::{Display, Formatter},
-    hashbrown::HashMap,
-    lasso::{Rodeo, Spur},
-    logos::{Lexer, Logos, Span},
-};
-
-macro_rules! tokendef {
-    ($($opcode:literal),* $(,)?) => {
-        paste::paste! {
-            #[derive(Clone, Copy, Debug, PartialEq, Eq, Logos)]
-            #[logos(extras = Rodeo)]
-            #[logos(skip r"[ \t\f]+")]
-            #[logos(skip r"-- .*")]
-            pub enum Token {
-                $(#[token($opcode, |_| hbbytecode::opcode::[<$opcode:upper>])])*
-                OpCode(u8),
-
-                #[regex("[0-9]+", |l| l.slice().parse().ok())]
-                #[regex(
-                    "-[0-9]+",
-                    |lexer| {
-                        Some(u64::from_ne_bytes(lexer.slice().parse::<i64>().ok()?.to_ne_bytes()))
-                    },
-                )] Integer(u64),
-
-                #[regex(
-                    "r[0-9]+",
-                    |lexer| match lexer.slice()[1..].parse() {
-                        Ok(n) => Some(n),
-                        _ => None
-                    },
-                )] Register(u8),
-
-                #[regex(
-                    r"\p{XID_Start}\p{XID_Continue}*:",
-                    |lexer| lexer.extras.get_or_intern(&lexer.slice()[..lexer.slice().len() - 1]),
-                )] Label(Spur),
-
-                #[regex(
-                    r"\p{XID_Start}\p{XID_Continue}*",
-                    |lexer| lexer.extras.get_or_intern(lexer.slice()),
-                )] Symbol(Spur),
-
-                #[token("\n")]
-                #[token(";")] ISep,
-                #[token(",")] PSep,
-            }
-        }
-    };
-}
-
-#[rustfmt::skip]
-tokendef![
-    "nop", "add", "sub", "mul", "and", "or", "xor", "sl", "sr", "srs", "cmp", "cmpu",
-    "dir", "neg", "not", "addi", "muli", "andi", "ori", "xori", "sli", "sri", "srsi",
-    "cmpi", "cmpui", "cp", "swa", "li", "ld", "st", "bmc", "brc", "jmp", "jeq", "jne",
-    "jlt", "jgt", "jltu", "jgtu", "ecall", "addf", "subf", "mulf", "dirf", "fmaf", "negf",
-    "itf", "fti", "addfi", "mulfi",
-];
-
-#[derive(Copy, Clone, Debug, PartialEq, Eq)]
-pub enum ErrorKind {
-    UnexpectedToken,
-    InvalidToken,
-    UnexpectedEnd,
-    InvalidSymbol,
-}
-
-#[derive(Clone, Debug, PartialEq, Eq)]
-pub struct Error {
-    pub kind: ErrorKind,
-    pub span: Span,
-}
-
-impl Display for Error {
-    fn fmt(&self, f: &mut Formatter<'_>) -> core::fmt::Result {
-        write!(f, "Error {:?} at {:?}", self.kind, self.span)
-    }
-}
-
-impl core::error::Error for Error {}
-
-macro_rules! expect_matches {
-    ($self:expr, $($pat:pat),* $(,)?) => {$(
-        let $pat = $self.next()?
-            else { return Err(ErrorKind::UnexpectedToken) };
-    )*}
-}
-
-pub fn assembly(code: &str, buf: &mut Vec<u8>) -> Result<(), Error> {
-    struct Assembler<'a> {
-        lexer: Lexer<'a, Token>,
-        buf: &'a mut Vec<u8>,
-        label_map: HashMap<Spur, u64>,
-        to_sub_label: HashMap<usize, Spur>,
-    }
-
-    impl<'a> Assembler<'a> {
-        fn next(&mut self) -> Result<Token, ErrorKind> {
-            match self.lexer.next() {
-                Some(Ok(t)) => Ok(t),
-                Some(Err(())) => Err(ErrorKind::InvalidToken),
-                None => Err(ErrorKind::UnexpectedEnd),
-            }
-        }
-
-        fn assemble(&mut self) -> Result<(), ErrorKind> {
-            use hbbytecode::opcode::*;
-            loop {
-                match self.lexer.next() {
-                    Some(Ok(Token::OpCode(op))) => {
-                        self.buf.push(op);
-                        match op {
-                            NOP | ECALL => Ok(()),
-                            DIR | DIRF => {
-                                expect_matches!(
-                                    self,
-                                    Token::Register(r0),
-                                    Token::PSep,
-                                    Token::Register(r1),
-                                    Token::PSep,
-                                    Token::Register(r2),
-                                    Token::PSep,
-                                    Token::Register(r3),
-                                );
-                                self.buf.extend([r0, r1, r2, r3]);
-                                Ok(())
-                            }
-                            ADD..=CMPU | ADDF..=MULF => {
-                                expect_matches!(
-                                    self,
-                                    Token::Register(r0),
-                                    Token::PSep,
-                                    Token::Register(r1),
-                                    Token::PSep,
-                                    Token::Register(r2),
-                                );
-                                self.buf.extend([r0, r1, r2]);
-                                Ok(())
-                            }
-                            BRC => {
-                                expect_matches!(
-                                    self,
-                                    Token::Register(r0),
-                                    Token::PSep,
-                                    Token::Register(r1),
-                                    Token::PSep,
-                                    Token::Integer(count),
-                                );
-                                self.buf.extend([
-                                    r0,
-                                    r1,
-                                    u8::try_from(count).map_err(|_| ErrorKind::UnexpectedToken)?,
-                                ]);
-                                Ok(())
-                            }
-                            NEG..=NOT | CP..=SWA | NEGF..=FTI => {
-                                expect_matches!(
-                                    self,
-                                    Token::Register(r0),
-                                    Token::PSep,
-                                    Token::Register(r1),
-                                );
-                                self.buf.extend([r0, r1]);
-                                Ok(())
-                            }
-                            LI | JMP => {
-                                expect_matches!(self, Token::Register(r0), Token::PSep);
-                                self.buf.push(r0);
-                                self.insert_imm()?;
-                                Ok(())
-                            }
-                            ADDI..=CMPUI | BMC | JEQ..=JGTU | ADDFI..=MULFI => {
-                                expect_matches!(
-                                    self,
-                                    Token::Register(r0),
-                                    Token::PSep,
-                                    Token::Register(r1),
-                                    Token::PSep,
-                                );
-                                self.buf.extend([r0, r1]);
-                                self.insert_imm()?;
-                                Ok(())
-                            }
-                            LD..=ST => {
-                                expect_matches!(
-                                    self,
-                                    Token::Register(r0),
-                                    Token::PSep,
-                                    Token::Register(r1),
-                                    Token::PSep,
-                                    Token::Integer(offset),
-                                    Token::PSep,
-                                    Token::Integer(len),
-                                );
-                                self.buf.extend([r0, r1]);
-                                self.buf.extend(offset.to_le_bytes());
-                                self.buf.extend(
-                                    u16::try_from(len)
-                                        .map_err(|_| ErrorKind::InvalidToken)?
-                                        .to_le_bytes(),
-                                );
-                                Ok(())
-                            }
-                            _ => unreachable!(),
-                        }?;
-                        match self.next() {
-                            Ok(Token::ISep) => (),
-                            Ok(_) => return Err(ErrorKind::UnexpectedToken),
-                            Err(ErrorKind::UnexpectedEnd) => return Ok(()),
-                            Err(e) => return Err(e),
-                        }
-                    }
-                    Some(Ok(Token::Label(lbl))) => {
-                        self.label_map.insert(lbl, self.buf.len() as u64);
-                    }
-                    Some(Ok(Token::ISep)) => (),
-                    Some(Ok(_)) => return Err(ErrorKind::UnexpectedToken),
-                    Some(Err(())) => return Err(ErrorKind::InvalidToken),
-                    None => return Ok(()),
-                }
-            }
-        }
-
-        fn link_local_syms(&mut self) -> Result<(), ErrorKind> {
-            for (ix, sym) in &self.to_sub_label {
-                self.label_map
-                    .get(sym)
-                    .ok_or(ErrorKind::InvalidSymbol)?
-                    .to_le_bytes()
-                    .iter()
-                    .enumerate()
-                    .for_each(|(i, b)| {
-                        self.buf[ix + i] = *b;
-                    });
-            }
-
-            Ok(())
-        }
-
-        fn insert_imm(&mut self) -> Result<(), ErrorKind> {
-            let imm = match self.next()? {
-                Token::Integer(i) => i.to_le_bytes(),
-                Token::Symbol(s) => {
-                    self.to_sub_label.insert(self.buf.len(), s);
-                    [0; 8]
-                }
-                _ => return Err(ErrorKind::UnexpectedToken),
-            };
-            self.buf.extend(imm);
-            Ok(())
-        }
-    }
-
-    let mut asm = Assembler {
-        lexer: Token::lexer(code),
-        label_map: Default::default(),
-        to_sub_label: Default::default(),
-        buf,
-    };
-
-    asm.assemble().map_err(|kind| Error {
-        kind,
-        span: asm.lexer.span(),
-    })?;
-
-    asm.link_local_syms()
-        .map_err(|kind| Error { kind, span: 0..0 })
-}

hbbytecode/hbbytecode.h

@@ -1,60 +0,0 @@
-/* HoleyBytes Bytecode representation in C
- * Requires C23 compiler or better
- */
-
-#pragma once
-#include <assert.h>
-#include <limits.h>
-#include <stdint.h>
-
-static_assert(CHAR_BIT == 8, "Cursed architectures are not supported");
-
-enum hbbc_Opcode: uint8_t {
-    hbbc_Op_NOP   , hbbc_Op_ADD   , hbbc_Op_SUB  , hbbc_Op_MUL   , hbbc_Op_AND  , hbbc_Op_OR   ,
-    hbbc_Op_XOR   , hbbc_Op_SL    , hbbc_Op_SR   , hbbc_Op_SRS   , hbbc_Op_CMP  , hbbc_Op_CMPU ,
-    hbbc_Op_DIR   , hbbc_Op_NEG   , hbbc_Op_NOT  , hbbc_Op_ADDI  , hbbc_Op_MULI , hbbc_Op_ANDI ,
-    hbbc_Op_ORI   , hbbc_Op_XORI  , hbbc_Op_SLI  , hbbc_Op_SRI   , hbbc_Op_SRSI , hbbc_Op_CMPI ,
-    hbbc_Op_CMPUI , hbbc_Op_CP    , hbbc_Op_SWA  , hbbc_Op_LI    , hbbc_Op_LD   , hbbc_Op_ST   ,
-    hbbc_Op_BMC   , hbbc_Op_BRC   , hbbc_Op_JMP  , hbbc_Op_JEQ   , hbbc_Op_JNE  , hbbc_Op_JLT  ,
-    hbbc_Op_JGT   , hbbc_Op_JLTU  , hbbc_Op_JGTU , hbbc_Op_ECALL , hbbc_Op_ADDF , hbbc_Op_SUBF ,
-    hbbc_Op_MULF  , hbbc_Op_DIRF  , hbbc_Op_FMAF , hbbc_Op_NEGF  , hbbc_Op_ITF  , hbbc_Op_FTI  ,
-    hbbc_Op_ADDFI , hbbc_Op_MULFI ,
-} typedef hbbc_Opcode;
-
-static_assert(sizeof(hbbc_Opcode) == 1);
-
-#pragma pack(push, 1)
-struct hbbc_ParamBBBB
-    { uint8_t _0; uint8_t _1; uint8_t _2; uint8_t _3; }
-    typedef hbbc_ParamBBBB;
-    static_assert(sizeof(hbbc_ParamBBBB) == 32 / 8);
-
-struct hbbc_ParamBBB
-    { uint8_t _0; uint8_t _1; uint8_t _2; }
-    typedef hbbc_ParamBBB;
-    static_assert(sizeof(hbbc_ParamBBB) == 24 / 8);
-
-struct hbbc_ParamBBDH
-    { uint8_t _0; uint8_t _1; uint64_t _2; uint16_t _3; }
-    typedef hbbc_ParamBBDH;
-    static_assert(sizeof(hbbc_ParamBBDH) == 96 / 8);
-
-struct hbbc_ParamBBD
-    { uint8_t _0; uint8_t _1; uint64_t _2; }
-    typedef hbbc_ParamBBD;
-    static_assert(sizeof(hbbc_ParamBBD) == 80 / 8);
-
-struct hbbc_ParamBB
-    { uint8_t _0; uint8_t _1; }
-    typedef hbbc_ParamBB;
-    static_assert(sizeof(hbbc_ParamBB) == 16 / 8);
-
-struct hbbc_ParamBD
-    { uint8_t _0; uint64_t _1; }
-    typedef hbbc_ParamBD;
-    static_assert(sizeof(hbbc_ParamBD) == 72 / 8);
-
-typedef uint64_t hbbc_ParamD;
-    static_assert(sizeof(hbbc_ParamD) == 64 / 8);
-
-#pragma pack(pop)

hbbytecode/src/lib.rs

@@ -1,107 +0,0 @@
-#![no_std]
-
-macro_rules! constmod {
-    ($vis:vis $mname:ident($repr:ty) {
-        $(#![doc = $mdoc:literal])?
-        $($cname:ident = $val:expr $(,$doc:literal)?;)*
-    }) => {
-        $(#[doc = $mdoc])?
-        $vis mod $mname {
-            $(
-                $(#[doc = $doc])?
-                pub const $cname: $repr = $val;
-            )*
-        }
-    };
-}
-
-constmod!(pub opcode(u8) {
-    //! Opcode constant module
-
-    NOP = 0, "N; Do nothing";
-
-    ADD  = 1,  "BBB; #0 ← #1 + #2";
-    SUB  = 2,  "BBB; #0 ← #1 - #2";
-    MUL  = 3,  "BBB; #0 ← #1 × #2";
-    AND  = 4,  "BBB; #0 ← #1 & #2";
-    OR   = 5,  "BBB; #0 ← #1 | #2";
-    XOR  = 6,  "BBB; #0 ← #1 ^ #2";
-    SL   = 7,  "BBB; #0 ← #1 « #2";
-    SR   = 8,  "BBB; #0 ← #1 » #2";
-    SRS  = 9,  "BBB; #0 ← #1 » #2 (signed)";
-    CMP  = 10, "BBB; #0 ← #1 <=> #2";
-    CMPU = 11, "BBB; #0 ← #1 <=> #2 (unsigned)";
-    DIR  = 12, "BBBB; #0 ← #2 / #3, #1 ← #2 % #3";
-    NEG  = 13, "BB; #0 ← -#1";
-    NOT  = 14, "BB; #0 ← !#1";
-
-    ADDI  = 15, "BBD; #0 ← #1 + imm #2";
-    MULI  = 16, "BBD; #0 ← #1 × imm #2";
-    ANDI  = 17, "BBD; #0 ← #1 & imm #2";
-    ORI   = 18, "BBD; #0 ← #1 | imm #2";
-    XORI  = 19, "BBD; #0 ← #1 ^ imm #2";
-    SLI   = 20, "BBD; #0 ← #1 « imm #2";
-    SRI   = 21, "BBD; #0 ← #1 » imm #2";
-    SRSI  = 22, "BBD; #0 ← #1 » imm #2 (signed)";
-    CMPI  = 23, "BBD; #0 ← #1 <=> imm #2";
-    CMPUI = 24, "BBD; #0 ← #1 <=> imm #2 (unsigned)";
-
-    CP  = 25, "BB; Copy #0 ← #1";
-    SWA = 26, "BB; Swap #0 and #1";
-    LI  = 27, "BD; #0 ← imm #1";
-    LD  = 28, "BBDB; #0 ← [#1 + imm #3], imm #4 bytes, overflowing";
-    ST  = 29, "BBDB; [#1 + imm #3] ← #0, imm #4 bytes, overflowing";
-    BMC = 30, "BBD; [#0] ← [#1], imm #2 bytes";
-    BRC = 31, "BBB; #0 ← #1, imm #2 registers";
-
-    JMP   = 32, "BD;  Unconditional jump [#0 + imm #1]";
-    JEQ   = 33, "BBD; if #0 = #1 → jump imm #2";
-    JNE   = 34, "BBD; if #0 ≠ #1 → jump imm #2";
-    JLT   = 35, "BBD; if #0 < #1 → jump imm #2";
-    JGT   = 36, "BBD; if #0 > #1 → jump imm #2";
-    JLTU  = 37, "BBD; if #0 < #1 → jump imm #2 (unsigned)";
-    JGTU  = 38, "BBD; if #0 > #1 → jump imm #2 (unsigned)";
-    ECALL = 39, "N; Issue system call";
-
-    ADDF = 40, "BBB; #0 ← #1 +. #2";
-    SUBF = 41, "BBB; #0 ← #1 -. #2";
-    MULF = 42, "BBB; #0 ← #1 +. #2";
-    DIRF = 43, "BBBB; #0 ← #2 / #3, #1 ← #2 % #3";
-    FMAF = 44, "BBBB; #0 ← (#1 * #2) + #3";
-    NEGF = 45, "BB; #0 ← -#1";
-    ITF  = 46, "BB; #0 ← #1 as float";
-    FTI  = 47, "BB; #0 ← #1 as int";
-
-    ADDFI = 48, "BBD; #0 ← #1 +. imm #2";
-    MULFI = 49, "BBD; #0 ← #1 *. imm #2";
-});
-
-#[repr(packed)]
-pub struct ParamBBBB(pub u8, pub u8, pub u8, pub u8);
-
-#[repr(packed)]
-pub struct ParamBBB(pub u8, pub u8, pub u8);
-
-#[repr(packed)]
-pub struct ParamBBDH(pub u8, pub u8, pub u64, pub u16);
-
-#[repr(packed)]
-pub struct ParamBBD(pub u8, pub u8, pub u64);
-
-#[repr(packed)]
-pub struct ParamBB(pub u8, pub u8);
-
-#[repr(packed)]
-pub struct ParamBD(pub u8, pub u64);
-
-/// # Safety
-/// Has to be valid to be decoded from bytecode.
-pub unsafe trait OpParam {}
-unsafe impl OpParam for ParamBBBB {}
-unsafe impl OpParam for ParamBBB {}
-unsafe impl OpParam for ParamBBDH {}
-unsafe impl OpParam for ParamBBD {}
-unsafe impl OpParam for ParamBB {}
-unsafe impl OpParam for ParamBD {}
-unsafe impl OpParam for u64 {}
-unsafe impl OpParam for () {}

hbvm/Cargo.toml

@@ -1,16 +0,0 @@
-[package]
-name = "hbvm"
-version = "0.1.0"
-edition = "2021"
-
-[profile.release]
-lto = true
-
-[dependencies]
-delegate          = "0.9"
-derive_more       = "0.99"
-hashbrown         = "0.13"
-hbbytecode.path   = "../hbbytecode"
-log               = "0.4"
-paste             = "1.0"
-static_assertions = "1.0"

hbvm/src/lib.rs

@@ -1,7 +0,0 @@
-#![doc = include_str!("../README.md")]
-#![no_std]
-
-extern crate alloc;
-
-pub mod validate;
-pub mod vm;

hbvm/src/main.rs

@@ -1,59 +0,0 @@
-use hbvm::vm::{
-    mem::{Memory, MemoryAccessReason, PageSize},
-    trap::HandleTrap,
-    value::Value,
-};
-
-use {
-    hbvm::{validate::validate, vm::Vm},
-    std::io::{stdin, Read},
-};
-
-fn main() -> Result<(), Box<dyn std::error::Error>> {
-    let mut prog = vec![];
-    stdin().read_to_end(&mut prog)?;
-
-    if let Err(e) = validate(&prog) {
-        eprintln!("Program validation error: {e:?}");
-        return Ok(());
-    } else {
-        unsafe {
-            let mut vm = Vm::<_, 0>::new_unchecked(&prog, TestTrapHandler);
-            vm.memory.insert_test_page();
-            println!("Program interrupt: {:?}", vm.run());
-            println!("{:?}", vm.registers);
-        }
-    }
-    Ok(())
-}
-
-pub fn time() -> u32 {
-    9
-}
-
-struct TestTrapHandler;
-impl HandleTrap for TestTrapHandler {
-    fn page_fault(
-        &mut self,
-        _: MemoryAccessReason,
-        _: &mut Memory,
-        _: u64,
-        _: PageSize,
-        _: *mut u8,
-    ) -> bool {
-        false
-    }
-
-    fn invalid_op(&mut self, _: &mut [Value; 256], _: &mut usize, _: &mut Memory, _: u8) -> bool
-    where
-        Self: Sized,
-    {
-        false
-    }
-
-    fn ecall(&mut self, _: &mut [Value; 256], _: &mut usize, _: &mut Memory)
-    where
-        Self: Sized,
-    {
-    }
-}

hbvm/src/validate.rs

@@ -1,64 +0,0 @@
-//! Validate if program is sound to execute
-
-/// Program validation error kind
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
-pub enum ErrorKind {
-    /// Unknown opcode
-    InvalidInstruction,
-    /// VM doesn't implement this valid opcode
-    Unimplemented,
-    /// Attempted to copy over register boundary
-    RegisterArrayOverflow,
-}
-
-/// Error
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
-pub struct Error {
-    /// Kind
-    pub kind:  ErrorKind,
-    /// Location in bytecode
-    pub index: usize,
-}
-
-/// Perform bytecode validation. If it passes, the program should be
-/// sound to execute.
-pub fn validate(mut program: &[u8]) -> Result<(), Error> {
-    use hbbytecode::opcode::*;
-
-    let start = program;
-    loop {
-        // Match on instruction types and perform necessary checks
-        program = match program {
-            [] => return Ok(()),
-            [LD..=ST, reg, _, _, _, _, _, _, _, _, _, count, ..]
-                if usize::from(*reg) * 8 + usize::from(*count) > 2048 =>
-            {
-                return Err(Error {
-                    kind:  ErrorKind::RegisterArrayOverflow,
-                    index: (program.as_ptr() as usize) - (start.as_ptr() as usize),
-                })
-            }
-            [BRC, src, dst, count, ..]
-                if src.checked_add(*count).is_none() || dst.checked_add(*count).is_none() =>
-            {
-                return Err(Error {
-                    kind:  ErrorKind::RegisterArrayOverflow,
-                    index: (program.as_ptr() as usize) - (start.as_ptr() as usize),
-                })
-            }
-            [NOP | ECALL, rest @ ..]
-            | [DIR | DIRF, _, _, _, _, rest @ ..]
-            | [ADD..=CMPU | BRC | ADDF..=MULF, _, _, _, rest @ ..]
-            | [NEG..=NOT | CP..=SWA | NEGF..=FTI, _, _, rest @ ..]
-            | [LI | JMP, _, _, _, _, _, _, _, _, _, rest @ ..]
-            | [ADDI..=CMPUI | BMC | JEQ..=JGTU | ADDFI..=MULFI, _, _, _, _, _, _, _, _, _, _, rest @ ..]
-            | [LD..=ST, _, _, _, _, _, _, _, _, _, _, _, _, rest @ ..] => rest,
-            _ => {
-                return Err(Error {
-                    kind:  ErrorKind::InvalidInstruction,
-                    index: (program.as_ptr() as usize) - (start.as_ptr() as usize),
-                })
-            }
-        }
-    }
-}

hbvm/src/vm/mem/mod.rs

@@ -1,430 +0,0 @@
-//! Program memory implementation
-
-pub mod paging;
-
-use {
-    self::paging::{PageTable, Permission, PtEntry},
-    super::{trap::HandleTrap, VmRunError},
-    alloc::boxed::Box,
-    core::mem::MaybeUninit,
-    derive_more::Display,
-};
-
-/// HoleyBytes virtual memory
-#[derive(Clone, Debug)]
-pub struct Memory {
-    /// Root page table
-    root_pt: *mut PageTable,
-}
-
-impl Default for Memory {
-    fn default() -> Self {
-        Self {
-            root_pt: Box::into_raw(Box::default()),
-        }
-    }
-}
-
-impl Drop for Memory {
-    fn drop(&mut self) {
-        let _ = unsafe { Box::from_raw(self.root_pt) };
-    }
-}
-
-impl Memory {
-    // HACK: Just for allocation testing, will be removed when proper memory interfaces
-    // implemented.
-    pub fn insert_test_page(&mut self) {
-        unsafe {
-            let mut entry = PtEntry::new(
-                {
-                    let layout = alloc::alloc::Layout::from_size_align_unchecked(4096, 4096);
-                    let ptr = alloc::alloc::alloc_zeroed(layout);
-                    if ptr.is_null() {
-                        alloc::alloc::handle_alloc_error(layout);
-                    }
-
-                    core::ptr::write_bytes(ptr, 69, 10);
-                    ptr.cast()
-                },
-                Permission::Write,
-            );
-
-            for _ in 0..4 {
-                let mut pt = Box::<PageTable>::default();
-                pt[0] = entry;
-                entry = PtEntry::new(Box::into_raw(pt) as _, Permission::Node);
-            }
-
-            (*self.root_pt)[0] = entry;
-        }
-    }
-
-    /// Load value from an address
-    ///
-    /// # Safety
-    /// Applies same conditions as for [`core::ptr::copy_nonoverlapping`]
-    pub unsafe fn load(
-        &mut self,
-        addr: u64,
-        target: *mut u8,
-        count: usize,
-        traph: &mut impl HandleTrap,
-    ) -> Result<(), LoadError> {
-        self.memory_access(
-            MemoryAccessReason::Load,
-            addr,
-            target,
-            count,
-            |perm| {
-                matches!(
-                    perm,
-                    Permission::Readonly | Permission::Write | Permission::Exec
-                )
-            },
-            |src, dst, count| core::ptr::copy_nonoverlapping(src, dst, count),
-            traph,
-        )
-        .map_err(LoadError)
-    }
-
-    /// Store value to an address
-    ///
-    /// # Safety
-    /// Applies same conditions as for [`core::ptr::copy_nonoverlapping`]
-    pub unsafe fn store(
-        &mut self,
-        addr: u64,
-        source: *const u8,
-        count: usize,
-        traph: &mut impl HandleTrap,
-    ) -> Result<(), StoreError> {
-        self.memory_access(
-            MemoryAccessReason::Store,
-            addr,
-            source.cast_mut(),
-            count,
-            |perm| perm == Permission::Write,
-            |dst, src, count| core::ptr::copy_nonoverlapping(src, dst, count),
-            traph,
-        )
-        .map_err(StoreError)
-    }
-
-    /// Copy a block of memory
-    ///
-    /// # Safety
-    /// - Same as for [`Self::load`] and [`Self::store`]
-    /// - Your faith in the gods of UB
-    ///     - Addr-san claims it's fine but who knows is she isn't lying :ferrisSus:
-    pub unsafe fn block_copy(
-        &mut self,
-        src: u64,
-        dst: u64,
-        count: usize,
-        traph: &mut impl HandleTrap,
-    ) -> Result<(), BlkCopyError> {
-        // Yea, i know it is possible to do this more efficiently, but I am too lazy.
-
-        const STACK_BUFFER_SIZE: usize = 512;
-
-        // Decide if to use stack-allocated buffer or to heap allocate
-        // Deallocation is again decided on size at the end of the function
-        let mut buf = MaybeUninit::<[u8; STACK_BUFFER_SIZE]>::uninit();
-        let buf = if count <= STACK_BUFFER_SIZE {
-            buf.as_mut_ptr().cast()
-        } else {
-            unsafe {
-                let layout = core::alloc::Layout::from_size_align_unchecked(count, 1);
-                let ptr = alloc::alloc::alloc(layout);
-                if ptr.is_null() {
-                    alloc::alloc::handle_alloc_error(layout);
-                }
-
-                ptr
-            }
-        };
-
-        // Perform memory block transfer
-        let status = (|| {
-            // Load to buffer
-            self.memory_access(
-                MemoryAccessReason::Load,
-                src,
-                buf,
-                count,
-                |perm| {
-                    matches!(
-                        perm,
-                        Permission::Readonly | Permission::Write | Permission::Exec
-                    )
-                },
-                |src, dst, count| core::ptr::copy(src, dst, count),
-                traph,
-            )
-            .map_err(|addr| BlkCopyError {
-                access_reason: MemoryAccessReason::Load,
-                addr,
-            })?;
-
-            // Store from buffer
-            self.memory_access(
-                MemoryAccessReason::Store,
-                dst,
-                buf,
-                count,
-                |perm| perm == Permission::Write,
-                |dst, src, count| core::ptr::copy(src, dst, count),
-                traph,
-            )
-            .map_err(|addr| BlkCopyError {
-                access_reason: MemoryAccessReason::Store,
-                addr,
-            })?;
-
-            Ok::<_, BlkCopyError>(())
-        })();
-
-        // Deallocate if used heap-allocated array
-        if count > STACK_BUFFER_SIZE {
-            alloc::alloc::dealloc(
-                buf,
-                core::alloc::Layout::from_size_align_unchecked(count, 1),
-            );
-        }
-
-        status
-    }
-
-    /// Split address to pages, check their permissions and feed pointers with offset
-    /// to a specified function.
-    ///
-    /// If page is not found, execute page fault trap handler.
-    #[allow(clippy::too_many_arguments)] // Silence peasant
-    fn memory_access(
-        &mut self,
-        reason: MemoryAccessReason,
-        src: u64,
-        mut dst: *mut u8,
-        len: usize,
-        permission_check: fn(Permission) -> bool,
-        action: fn(*mut u8, *mut u8, usize),
-        traph: &mut impl HandleTrap,
-    ) -> Result<(), u64> {
-        let mut pspl = AddrSplitter::new(src, len, self.root_pt);
-        loop {
-            match pspl.next() {
-                // Page found
-                Some(Ok(AddrSplitOk {
-                    vaddr,
-                    ptr,
-                    size,
-                    perm,
-                })) => {
-                    if !permission_check(perm) {
-                        return Err(vaddr);
-                    }
-
-                    // Perform memory action and bump dst pointer
-                    action(ptr, dst, size);
-                    dst = unsafe { dst.add(size) };
-                }
-                Some(Err(AddrSplitError { addr, size })) => {
-                    // Execute page fault handler
-                    if traph.page_fault(reason, self, addr, size, dst) {
-                        // Shift the splitter address
-                        pspl.bump(size);
-
-                        // Bump dst pointer
-                        dst = unsafe { dst.add(size as _) };
-                    } else {
-                        return Err(addr); // Unhandleable
-                    }
-                }
-                None => return Ok(()),
-            }
-        }
-    }
-}
-
-/// Result from address split
-struct AddrSplitOk {
-    /// Virtual address
-    vaddr: u64,
-
-    /// Pointer to the start for perform operation
-    ptr: *mut u8,
-
-    /// Size to the end of page / end of desired size
-    size: usize,
-
-    /// Page permission
-    perm: Permission,
-}
-
-struct AddrSplitError {
-    /// Address of failure
-    addr: u64,
-
-    /// Requested page size
-    size: PageSize,
-}
-
-/// Address splitter into pages
-struct AddrSplitter {
-    /// Current address
-    addr: u64,
-
-    /// Size left
-    size: usize,
-
-    /// Page table
-    pagetable: *const PageTable,
-}
-
-impl AddrSplitter {
-    /// Create a new page splitter
-    pub const fn new(addr: u64, size: usize, pagetable: *const PageTable) -> Self {
-        Self {
-            addr,
-            size,
-            pagetable,
-        }
-    }
-
-    /// Bump address by size X
-    fn bump(&mut self, page_size: PageSize) {
-        self.addr += page_size as u64;
-        self.size = self.size.saturating_sub(page_size as _);
-    }
-}
-
-impl Iterator for AddrSplitter {
-    type Item = Result<AddrSplitOk, AddrSplitError>;
-
-    fn next(&mut self) -> Option<Self::Item> {
-        // The end, everything is fine
-        if self.size == 0 {
-            return None;
-        }
-
-        let (base, perm, size, offset) = 'a: {
-            let mut current_pt = self.pagetable;
-
-            // Walk the page table
-            for lvl in (0..5).rev() {
-                // Get an entry
-                unsafe {
-                    let entry = (*current_pt).get_unchecked(
-                        usize::try_from((self.addr >> (lvl * 9 + 12)) & ((1 << 9) - 1))
-                            .expect("?conradluget a better CPU"),
-                    );
-
-                    let ptr = entry.ptr();
-                    match entry.permission() {
-                        // No page → page fault
-                        Permission::Empty => {
-                            return Some(Err(AddrSplitError {
-                                addr: self.addr,
-                                size: PageSize::from_lvl(lvl)?,
-                            }))
-                        }
-
-                        // Node → proceed waking
-                        Permission::Node => current_pt = ptr as _,
-
-                        // Leaft → return relevant data
-                        perm => {
-                            break 'a (
-                                // Pointer in host memory
-                                ptr as *mut u8,
-                                perm,
-                                PageSize::from_lvl(lvl)?,
-                                // In-page offset
-                                self.addr as usize & ((1 << (lvl * 9 + 12)) - 1),
-                            );
-                        }
-                    }
-                }
-            }
-            return None; // Reached the end (should not happen)
-        };
-
-        // Get available byte count in the selected page with offset
-        let avail = (size as usize - offset).clamp(0, self.size);
-        self.bump(size);
-
-        Some(Ok(AddrSplitOk {
-            vaddr: self.addr,
-            ptr: unsafe { base.add(offset) }, // Return pointer to the start of region
-            size: avail,
-            perm,
-        }))
-    }
-}
-
-/// Page size
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
-pub enum PageSize {
-    /// 4 KiB page (on level 0)
-    Size4K = 4096,
-
-    /// 2 MiB page (on level 1)
-    Size2M = 1024 * 1024 * 2,
-
-    /// 1 GiB page (on level 2)
-    Size1G = 1024 * 1024 * 1024,
-}
-
-impl PageSize {
-    /// Convert page table level to size of page
-    fn from_lvl(lvl: u8) -> Option<Self> {
-        match lvl {
-            0 => Some(PageSize::Size4K),
-            1 => Some(PageSize::Size2M),
-            2 => Some(PageSize::Size1G),
-            _ => None,
-        }
-    }
-}
-
-/// Unhandled load access trap
-#[derive(Clone, Copy, Display, Debug, PartialEq, Eq)]
-pub struct LoadError(u64);
-
-/// Unhandled store access trap
-#[derive(Clone, Copy, Display, Debug, PartialEq, Eq)]
-pub struct StoreError(u64);
-
-#[derive(Clone, Copy, Display, Debug, PartialEq, Eq)]
-pub enum MemoryAccessReason {
-    Load,
-    Store,
-}
-
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
-pub struct BlkCopyError {
-    access_reason: MemoryAccessReason,
-    addr: u64,
-}
-
-impl From<BlkCopyError> for VmRunError {
-    fn from(value: BlkCopyError) -> Self {
-        match value.access_reason {
-            MemoryAccessReason::Load => Self::LoadAccessEx(value.addr),
-            MemoryAccessReason::Store => Self::StoreAccessEx(value.addr),
-        }
-    }
-}
-
-impl From<LoadError> for VmRunError {
-    fn from(value: LoadError) -> Self {
-        Self::LoadAccessEx(value.0)
-    }
-}
-
-impl From<StoreError> for VmRunError {
-    fn from(value: StoreError) -> Self {
-        Self::StoreAccessEx(value.0)
-    }
-}

hbvm/src/vm/mem/paging.rs

@@ -1,152 +0,0 @@
-//! Page table and associated structures implementation
-
-use {
-    core::{
-        fmt::Debug,
-        mem::MaybeUninit,
-        ops::{Index, IndexMut},
-        slice::SliceIndex,
-    },
-    delegate::delegate,
-};
-
-/// Page entry permission
-#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
-#[repr(u8)]
-pub enum Permission {
-    /// No page present
-    #[default]
-    Empty,
-    /// Points to another pagetable
-    Node,
-    /// Page is read only
-    Readonly,
-    /// Page is readable and writable
-    Write,
-    /// Page is readable and executable
-    Exec,
-}
-
-/// Page table entry
-#[derive(Clone, Copy, Default, PartialEq, Eq)]
-pub struct PtEntry(u64);
-impl PtEntry {
-    /// Create new
-    ///
-    /// # Safety
-    /// - `ptr` has to point to valid data and shall not be deallocated
-    ///    troughout the entry lifetime
-    #[inline]
-    pub unsafe fn new(ptr: *mut PtPointedData, permission: Permission) -> Self {
-        Self(ptr as u64 | permission as u64)
-    }
-
-    /// Get permission
-    #[inline]
-    pub fn permission(&self) -> Permission {
-        unsafe { core::mem::transmute(self.0 as u8 & 0b111) }
-    }
-
-    /// Get pointer to the data (leaf) or next page table (node)
-    #[inline]
-    pub fn ptr(&self) -> *mut PtPointedData {
-        (self.0 & !((1 << 12) - 1)) as _
-    }
-}
-
-impl Debug for PtEntry {
-    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
-        f.debug_struct("PtEntry")
-            .field("ptr", &self.ptr())
-            .field("permission", &self.permission())
-            .finish()
-    }
-}
-
-/// Page table
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
-#[repr(align(4096))]
-pub struct PageTable([PtEntry; 512]);
-
-impl PageTable {
-    delegate!(to self.0 {
-        /// Returns a reference to an element or subslice depending on the type of
-        /// index.
-        ///
-        /// - If given a position, returns a reference to the element at that
-        ///   position or `None` if out of bounds.
-        /// - If given a range, returns the subslice corresponding to that range,
-        ///   or `None` if out of bounds.
-        ///
-        pub fn get<I>(&self, ix: I) -> Option<&I::Output>
-            where I: SliceIndex<[PtEntry]>;
-
-        /// Returns a mutable reference to an element or subslice depending on the
-        /// type of index (see [`get`]) or `None` if the index is out of bounds.
-        pub fn get_mut<I>(&mut self, ix: I) -> Option<&mut I::Output>
-            where I: SliceIndex<[PtEntry]>;
-
-        /// Returns a reference to an element or subslice, without doing bounds
-        /// checking.
-        ///
-        /// For a safe alternative see [`get`].
-        ///
-        /// # Safety
-        ///
-        /// Calling this method with an out-of-bounds index is *[undefined behavior]*
-        /// even if the resulting reference is not used.
-        pub unsafe fn get_unchecked<I>(&self, index: I) -> &I::Output
-            where I: SliceIndex<[PtEntry]>;
-
-        /// Returns a mutable reference to an element or subslice, without doing
-        /// bounds checking.
-        ///
-        /// For a safe alternative see [`get_mut`].
-        ///
-        /// # Safety
-        ///
-        /// Calling this method with an out-of-bounds index is *[undefined behavior]*
-        /// even if the resulting reference is not used.
-        pub unsafe fn get_unchecked_mut<I>(&mut self, index: I) -> &mut I::Output
-            where I: SliceIndex<[PtEntry]>;
-    });
-}
-
-impl<Idx> Index<Idx> for PageTable
-where
-    Idx: SliceIndex<[PtEntry]>,
-{
-    type Output = Idx::Output;
-
-    #[inline(always)]
-    fn index(&self, index: Idx) -> &Self::Output {
-        &self.0[index]
-    }
-}
-
-impl<Idx> IndexMut<Idx> for PageTable
-where
-    Idx: SliceIndex<[PtEntry]>,
-{
-    #[inline(always)]
-    fn index_mut(&mut self, index: Idx) -> &mut Self::Output {
-        &mut self.0[index]
-    }
-}
-
-impl Default for PageTable {
-    fn default() -> Self {
-        // SAFETY: It's fine, zeroed page table entry is valid (= empty)
-        Self(unsafe { MaybeUninit::zeroed().assume_init() })
-    }
-}
-
-/// Data page table entry can possibly point to
-#[derive(Clone, Copy)]
-#[repr(C, align(4096))]
-pub union PtPointedData {
-    /// Node - next page table
-    pub pt:   PageTable,
-    /// Leaf
-    pub page: u8,
-}

hbvm/src/vm/mod.rs

@@ -1,373 +0,0 @@
-//! HoleyBytes Virtual Machine
-//!
-//! All unsafe code here should be sound, if input bytecode passes validation.
-
-// # General safety notice:
-// - Validation has to assure there is 256 registers (r0 - r255)
-// - Instructions have to be valid as specified (values and sizes)
-// - Mapped pages should be at least 4 KiB
-// - Yes, I am aware of the UB when jumping in-mid of instruction where
-//   the read byte corresponds to an instruction whose lenght exceets the
-//   program size. If you are (rightfully) worried about the UB, for now just
-//   append your program with 11 zeroes.
-
-use self::trap::HandleTrap;
-
-pub mod mem;
-pub mod trap;
-pub mod value;
-
-use {
-    crate::validate,
-    core::ops,
-    hbbytecode::{OpParam, ParamBB, ParamBBB, ParamBBBB, ParamBBD, ParamBBDH, ParamBD},
-    mem::Memory,
-    static_assertions::assert_impl_one,
-    value::Value,
-};
-
-/// Extract a parameter from program
-macro_rules! param {
-    ($self:expr, $ty:ty) => {{
-        assert_impl_one!($ty: OpParam);
-        let data = $self
-            .program
-            .as_ptr()
-            .add($self.pc + 1)
-            .cast::<$ty>()
-            .read();
-        $self.pc += 1 + core::mem::size_of::<$ty>();
-        data
-    }};
-}
-
-/// Perform binary operation `#0 ← #1 OP #2`
-macro_rules! binary_op {
-    ($self:expr, $ty:ident, $handler:expr) => {{
-        let ParamBBB(tg, a0, a1) = param!($self, ParamBBB);
-        $self.write_reg(
-            tg,
-            $handler(
-                Value::$ty(&$self.read_reg(a0)),
-                Value::$ty(&$self.read_reg(a1)),
-            ),
-        );
-    }};
-}
-
-/// Perform binary operation with immediate `#0 ← #1 OP imm #2`
-macro_rules! binary_op_imm {
-    ($self:expr, $ty:ident, $handler:expr) => {{
-        let ParamBBD(tg, a0, imm) = param!($self, ParamBBD);
-        $self.write_reg(
-            tg,
-            $handler(Value::$ty(&$self.read_reg(a0)), Value::$ty(&imm.into())),
-        );
-    }};
-}
-
-/// Jump at `#3` if ordering on `#0 <=> #1` is equal to expected
-macro_rules! cond_jump {
-    ($self:expr, $ty:ident, $expected:ident) => {{
-        let ParamBBD(a0, a1, jt) = param!($self, ParamBBD);
-        if core::cmp::Ord::cmp(&$self.read_reg(a0).as_u64(), &$self.read_reg(a1).as_u64())
-            == core::cmp::Ordering::$expected
-        {
-            $self.pc = jt as usize;
-        }
-    }};
-}
-
-/// HoleyBytes Virtual Machine
-pub struct Vm<'a, T, const TIMER_QUOTIENT: usize> {
-    /// Holds 256 registers
-    ///
-    /// Writing to register 0 is considered undefined behaviour
-    /// in terms of HoleyBytes program execution
-    pub registers: [Value; 256],
-
-    /// Memory implementation
-    pub memory: Memory,
-
-    /// Trap handler
-    pub traph: T,
-
-    // Program counter
-    pc: usize,
-
-    /// Program
-    program: &'a [u8],
-
-    /// Program timer
-    timer: usize,
-}
-
-impl<'a, T: HandleTrap, const TIMER_QUOTIENT: usize> Vm<'a, T, TIMER_QUOTIENT> {
-    /// Create a new VM with program and trap handler
-    ///
-    /// # Safety
-    /// Program code has to be validated
-    pub unsafe fn new_unchecked(program: &'a [u8], traph: T) -> Self {
-        Self {
-            registers: [Value::from(0_u64); 256],
-            memory: Default::default(),
-            traph,
-            pc: 0,
-            program,
-            timer: 0,
-        }
-    }
-
-    /// Create a new VM with program and trap handler only if it passes validation
-    pub fn new_validated(program: &'a [u8], traph: T) -> Result<Self, validate::Error> {
-        validate::validate(program)?;
-        Ok(unsafe { Self::new_unchecked(program, traph) })
-    }
-
-    /// Execute program
-    ///
-    /// Program can return [`VmRunError`] if a trap handling failed
-    pub fn run(&mut self) -> Result<VmRunOk, VmRunError> {
-        use hbbytecode::opcode::*;
-        loop {
-            // Fetch instruction
-            let Some(&opcode) = self.program.get(self.pc)
-                else { return Ok(VmRunOk::End) };
-
-            // Big match
-            unsafe {
-                match opcode {
-                    NOP => param!(self, ()),
-                    ADD => binary_op!(self, as_u64, u64::wrapping_add),
-                    SUB => binary_op!(self, as_u64, u64::wrapping_sub),
-                    MUL => binary_op!(self, as_u64, u64::wrapping_mul),
-                    AND => binary_op!(self, as_u64, ops::BitAnd::bitand),
-                    OR => binary_op!(self, as_u64, ops::BitOr::bitor),
-                    XOR => binary_op!(self, as_u64, ops::BitXor::bitxor),
-                    SL => binary_op!(self, as_u64, ops::Shl::shl),
-                    SR => binary_op!(self, as_u64, ops::Shr::shr),
-                    SRS => binary_op!(self, as_i64, ops::Shr::shr),
-                    CMP => {
-                        let ParamBBB(tg, a0, a1) = param!(self, ParamBBB);
-                        self.write_reg(
-                            tg,
-                            self.read_reg(a0).as_i64().cmp(&self.read_reg(a1).as_i64()) as i64,
-                        );
-                    }
-                    CMPU => {
-                        let ParamBBB(tg, a0, a1) = param!(self, ParamBBB);
-                        self.write_reg(
-                            tg,
-                            self.read_reg(a0).as_u64().cmp(&self.read_reg(a1).as_u64()) as i64,
-                        );
-                    }
-                    NOT => {
-                        let param = param!(self, ParamBB);
-                        self.write_reg(param.0, !self.read_reg(param.1).as_u64());
-                    }
-                    NEG => {
-                        let param = param!(self, ParamBB);
-                        self.write_reg(
-                            param.0,
-                            match self.read_reg(param.1).as_u64() {
-                                0 => 1_u64,
-                                _ => 0,
-                            },
-                        );
-                    }
-                    DIR => {
-                        let ParamBBBB(dt, rt, a0, a1) = param!(self, ParamBBBB);
-                        let a0 = self.read_reg(a0).as_u64();
-                        let a1 = self.read_reg(a1).as_u64();
-                        self.write_reg(dt, a0.checked_div(a1).unwrap_or(u64::MAX));
-                        self.write_reg(rt, a0.checked_rem(a1).unwrap_or(u64::MAX));
-                    }
-                    ADDI => binary_op_imm!(self, as_u64, ops::Add::add),
-                    MULI => binary_op_imm!(self, as_u64, ops::Mul::mul),
-                    ANDI => binary_op_imm!(self, as_u64, ops::BitAnd::bitand),
-                    ORI => binary_op_imm!(self, as_u64, ops::BitOr::bitor),
-                    XORI => binary_op_imm!(self, as_u64, ops::BitXor::bitxor),
-                    SLI => binary_op_imm!(self, as_u64, ops::Shl::shl),
-                    SRI => binary_op_imm!(self, as_u64, ops::Shr::shr),
-                    SRSI => binary_op_imm!(self, as_i64, ops::Shr::shr),
-                    CMPI => {
-                        let ParamBBD(tg, a0, imm) = param!(self, ParamBBD);
-                        self.write_reg(
-                            tg,
-                            self.read_reg(a0).as_i64().cmp(&Value::from(imm).as_i64()) as i64,
-                        );
-                    }
-                    CMPUI => {
-                        let ParamBBD(tg, a0, imm) = param!(self, ParamBBD);
-                        self.write_reg(tg, self.read_reg(a0).as_u64().cmp(&imm) as i64);
-                    }
-                    CP => {
-                        let param = param!(self, ParamBB);
-                        self.write_reg(param.0, self.read_reg(param.1));
-                    }
-                    SWA => {
-                        let ParamBB(src, dst) = param!(self, ParamBB);
-                        if src + dst != 0 {
-                            core::ptr::swap(
-                                self.registers.get_unchecked_mut(usize::from(src)),
-                                self.registers.get_unchecked_mut(usize::from(dst)),
-                            );
-                        }
-                    }
-                    LI => {
-                        let param = param!(self, ParamBD);
-                        self.write_reg(param.0, param.1);
-                    }
-                    LD => {
-                        let ParamBBDH(dst, base, off, count) = param!(self, ParamBBDH);
-                        let n: usize = match dst {
-                            0 => 1,
-                            _ => 0,
-                        };
-
-                        self.memory.load(
-                            self.read_reg(base).as_u64() + off + n as u64,
-                            self.registers.as_mut_ptr().add(usize::from(dst) + n).cast(),
-                            usize::from(count).saturating_sub(n),
-                            &mut self.traph,
-                        )?;
-                    }
-                    ST => {
-                        let ParamBBDH(dst, base, off, count) = param!(self, ParamBBDH);
-                        self.memory.store(
-                            self.read_reg(base).as_u64() + off,
-                            self.registers.as_ptr().add(usize::from(dst)).cast(),
-                            count.into(),
-                            &mut self.traph,
-                        )?;
-                    }
-                    BMC => {
-                        let ParamBBD(src, dst, count) = param!(self, ParamBBD);
-                        self.memory.block_copy(
-                            self.read_reg(src).as_u64(),
-                            self.read_reg(dst).as_u64(),
-                            count as _,
-                            &mut self.traph,
-                        )?;
-                    }
-                    BRC => {
-                        let ParamBBB(src, dst, count) = param!(self, ParamBBB);
-                        core::ptr::copy(
-                            self.registers.get_unchecked(usize::from(src)),
-                            self.registers.get_unchecked_mut(usize::from(dst)),
-                            usize::from(count * 8),
-                        );
-                    }
-                    JMP => {
-                        let ParamBD(reg, offset) = param!(self, ParamBD);
-                        self.pc = (self.read_reg(reg).as_u64() + offset) as usize;
-                    }
-                    JEQ => cond_jump!(self, int, Equal),
-                    JNE => {
-                        let ParamBBD(a0, a1, jt) = param!(self, ParamBBD);
-                        if self.read_reg(a0).as_u64() != self.read_reg(a1).as_u64() {
-                            self.pc = jt as usize;
-                        }
-                    }
-                    JLT => cond_jump!(self, int, Less),
-                    JGT => cond_jump!(self, int, Greater),
-                    JLTU => cond_jump!(self, sint, Less),
-                    JGTU => cond_jump!(self, sint, Greater),
-                    ECALL => {
-                        param!(self, ());
-                        self.traph
-                            .ecall(&mut self.registers, &mut self.pc, &mut self.memory);
-                    }
-                    ADDF => binary_op!(self, as_f64, ops::Add::add),
-                    SUBF => binary_op!(self, as_f64, ops::Sub::sub),
-                    MULF => binary_op!(self, as_f64, ops::Mul::mul),
-                    DIRF => {
-                        let ParamBBBB(dt, rt, a0, a1) = param!(self, ParamBBBB);
-                        let a0 = self.read_reg(a0).as_f64();
-                        let a1 = self.read_reg(a1).as_f64();
-                        self.write_reg(dt, a0 / a1);
-                        self.write_reg(rt, a0 % a1);
-                    }
-                    FMAF => {
-                        let ParamBBBB(dt, a0, a1, a2) = param!(self, ParamBBBB);
-                        self.write_reg(
-                            dt,
-                            self.read_reg(a0).as_f64() * self.read_reg(a1).as_f64()
-                                + self.read_reg(a2).as_f64(),
-                        );
-                    }
-                    NEGF => {
-                        let ParamBB(dt, a0) = param!(self, ParamBB);
-                        self.write_reg(dt, -self.read_reg(a0).as_f64());
-                    }
-                    ITF => {
-                        let ParamBB(dt, a0) = param!(self, ParamBB);
-                        self.write_reg(dt, self.read_reg(a0).as_i64() as f64);
-                    }
-                    FTI => {
-                        let ParamBB(dt, a0) = param!(self, ParamBB);
-                        self.write_reg(dt, self.read_reg(a0).as_f64() as i64);
-                    }
-                    ADDFI => binary_op_imm!(self, as_f64, ops::Add::add),
-                    MULFI => binary_op_imm!(self, as_f64, ops::Mul::mul),
-                    op => {
-                        if !self.traph.invalid_op(
-                            &mut self.registers,
-                            &mut self.pc,
-                            &mut self.memory,
-                            op,
-                        ) {
-                            return Err(VmRunError::InvalidOpcodeEx(op));
-                        }
-                    }
-                }
-            }
-
-            if TIMER_QUOTIENT != 0 {
-                self.timer = self.timer.wrapping_add(1);
-                if self.timer % TIMER_QUOTIENT == 0 {
-                    return Ok(VmRunOk::Timer);
-                }
-            }
-        }
-    }
-
-    /// Read register
-    #[inline]
-    unsafe fn read_reg(&self, n: u8) -> Value {
-        *self.registers.get_unchecked(n as usize)
-    }
-
-    /// Write a register.
-    /// Writing to register 0 is no-op.
-    #[inline]
-    unsafe fn write_reg(&mut self, n: u8, value: impl Into<Value>) {
-        if n != 0 {
-            *self.registers.get_unchecked_mut(n as usize) = value.into();
-        }
-    }
-}
-
-/// Virtual machine halt error
-#[derive(Copy, Clone, Debug, PartialEq, Eq)]
-#[repr(u8)]
-pub enum VmRunError {
-    /// Unhandled invalid opcode exceptions
-    InvalidOpcodeEx(u8),
-
-    /// Unhandled load access exception
-    LoadAccessEx(u64),
-
-    /// Unhandled store access exception
-    StoreAccessEx(u64),
-}
-
-/// Virtual machine halt ok
-#[derive(Copy, Clone, Debug, PartialEq, Eq)]
-pub enum VmRunOk {
-    /// Program has eached its end
-    End,
-
-    /// Program was interrupted by a timer
-    Timer,
-}

hbvm/src/vm/trap.rs

@@ -1,35 +0,0 @@
-//! Program trap handling interfaces
-
-use super::{
-    mem::{Memory, MemoryAccessReason, PageSize},
-    value::Value,
-};
-
-/// Handle VM traps
-pub trait HandleTrap {
-    /// Handle page fault
-    fn page_fault(
-        &mut self,
-        reason: MemoryAccessReason,
-        memory: &mut Memory,
-        vaddr: u64,
-        size: PageSize,
-        dataptr: *mut u8,
-    ) -> bool;
-
-    /// Handle invalid opcode exception
-    fn invalid_op(
-        &mut self,
-        regs: &mut [Value; 256],
-        pc: &mut usize,
-        memory: &mut Memory,
-        op: u8,
-    ) -> bool
-    where
-        Self: Sized;
-
-    /// Handle environment calls
-    fn ecall(&mut self, regs: &mut [Value; 256], pc: &mut usize, memory: &mut Memory)
-    where
-        Self: Sized;
-}

hbvm/src/vm/value.rs

@@ -1,48 +0,0 @@
-//! HoleyBytes register value definition
-
-use core::fmt::Debug;
-
-/// Define [`Value`] union
-///
-/// # Safety
-/// Union variants have to be sound to byte-reinterpretate
-/// between each other. Otherwise the behaviour is undefined.
-macro_rules! value_def {
-    ($($ty:ident),* $(,)?) => {
-        /// HBVM register value
-        #[derive(Copy, Clone)]
-        #[repr(packed)]
-        pub union Value {
-            $(pub $ty: $ty),*
-        }
-
-        paste::paste! {
-            impl Value {$(
-                #[doc = "Byte-reinterpret [`Value`] as [`" $ty "`]"]
-                #[inline]
-                pub fn [<as_ $ty>](&self) -> $ty {
-                    unsafe { self.$ty }
-                }
-            )*}
-        }
-
-        $(
-            impl From<$ty> for Value {
-                #[inline]
-                fn from(value: $ty) -> Self {
-                    Self { $ty: value }
-                }
-            }
-        )*
-    };
-}
-
-value_def!(u64, i64, f64);
-static_assertions::const_assert_eq!(core::mem::size_of::<Value>(), 8);
-
-impl Debug for Value {
-    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
-        // Print formatted as hexadecimal, unsigned integer
-        write!(f, "{:x}", self.as_u64())
-    }
-}

lang/Cargo.toml

@@ -0,0 +1,23 @@
+[package]
+name = "hblang"
+version = "0.1.0"
+edition = "2021"
+
+[[bin]]
+name = "hbc"
+path = "src/main.rs"
+
+[[bin]]
+name = "fuzz"
+path = "src/fuzz_main.rs"
+
+[dependencies]
+hbbytecode = { workspace = true, features = ["disasm"] }
+hbvm = { workspace = true, features = ["nightly"] }
+hashbrown = { version = "0.15.0", default-features = false, features = ["raw-entry"] }
+log = "0.4.22"
+
+[features]
+default = ["std"]
+std = []
+no_log = ["log/max_level_off"]

lang/command-help.txt

@@ -0,0 +1,4 @@
+--fmt            - format all imported source files
+--fmt-stdout     - dont write the formatted file but print it
+--dump-asm       - output assembly instead of raw code, (the assembly is more for debugging the compiler)
+--threads <1...> - number of extra threads compiler can use [default: 0]

lang/src/fmt.rs

@@ -0,0 +1,614 @@
+use {
+    crate::{
+        lexer::{self, Lexer, TokenKind},
+        parser::{self, CommentOr, CtorField, EnumField, Expr, Poser, Radix, StructField},
+    },
+    core::fmt::{self},
+};
+
+pub fn display_radix(radix: Radix, mut value: u64, buf: &mut [u8; 64]) -> &str {
+    fn conv_radix(d: u8) -> u8 {
+        match d {
+            0..=9 => d + b'0',
+            _ => d - 10 + b'A',
+        }
+    }
+
+    for (i, b) in buf.iter_mut().enumerate().rev() {
+        let d = (value % radix as u64) as u8;
+        value /= radix as u64;
+        *b = conv_radix(d);
+        if value == 0 {
+            return unsafe { core::str::from_utf8_unchecked(&buf[i..]) };
+        }
+    }
+
+    unreachable!()
+}
+
+#[repr(u8)]
+enum TokenGroup {
+    Blank,
+    Comment,
+    Keyword,
+    Identifier,
+    Directive,
+    Number,
+    String,
+    Op,
+    Assign,
+    Paren,
+    Bracket,
+    Colon,
+    Comma,
+    Dot,
+    Ctor,
+}
+
+fn token_group(kind: TokenKind) -> TokenGroup {
+    use {crate::lexer::TokenKind::*, TokenGroup as TG};
+    match kind {
+        BSlash | Pound | Eof | Ct => TG::Blank,
+        Comment => TG::Comment,
+        Directive => TG::Directive,
+        Colon => TG::Colon,
+        Semi | Comma => TG::Comma,
+        Dot => TG::Dot,
+        Ctor | Tupl | TArrow => TG::Ctor,
+        LParen | RParen => TG::Paren,
+        LBrace | RBrace | LBrack | RBrack => TG::Bracket,
+        Number | Float => TG::Number,
+        Under | CtIdent | Ident => TG::Identifier,
+        Tick | Tilde | Que | Not | Mod | Band | Bor | Xor | Mul | Add | Sub | Div | Shl | Shr
+        | Or | And | Lt | Gt | Eq | Le | Ge | Ne => TG::Op,
+        Decl | Assign | BorAss | XorAss | BandAss | AddAss | SubAss | MulAss | DivAss | ModAss
+        | ShrAss | ShlAss => TG::Assign,
+        DQuote | Quote => TG::String,
+        Return | If | Else | Loop | Break | Continue | Fn | Idk | Die | Struct | Packed | True
+        | False | Null | Match | Enum => TG::Keyword,
+    }
+}
+
+pub fn get_token_kinds(mut source: &mut [u8]) -> usize {
+    let len = source.len();
+    loop {
+        let src = unsafe { core::str::from_utf8_unchecked(source) };
+        let mut token = lexer::Lexer::new(src).eat();
+        match token.kind {
+            TokenKind::Eof => break,
+            // ???
+            TokenKind::CtIdent | TokenKind::Directive => token.start -= 1,
+            _ => {}
+        }
+        let start = token.start as usize;
+        let end = token.end as usize;
+        source[..start].fill(0);
+        source[start..end].fill(token_group(token.kind) as u8);
+        source = &mut source[end..];
+    }
+    len
+}
+
+pub fn minify(source: &mut str) -> usize {
+    fn needs_space(c: u8) -> bool {
+        matches!(c, b'a'..=b'z' | b'A'..=b'Z' | b'0'..=b'9' | 127..)
+    }
+
+    let mut writer = source.as_mut_ptr();
+    let mut reader = &source[..];
+    let mut prev_needs_whitecpace = false;
+    let mut prev_needs_newline = false;
+    loop {
+        let mut token = lexer::Lexer::new(reader).eat();
+        match token.kind {
+            TokenKind::Eof => break,
+            TokenKind::CtIdent | TokenKind::Directive => token.start -= 1,
+            _ => {}
+        }
+
+        let cpy_len = token.range().len();
+
+        let mut prefix = 0;
+        if prev_needs_whitecpace && needs_space(reader.as_bytes()[token.start as usize]) {
+            prefix = b' ';
+            debug_assert!(token.start != 0, "{reader}");
+        }
+        prev_needs_whitecpace = needs_space(reader.as_bytes()[token.end as usize - 1]);
+
+        let inbetween_new_lines =
+            reader[..token.start as usize].bytes().filter(|&b| b == b'\n').count()
+                + token.kind.precedence().is_some() as usize;
+        let extra_prefix_new_lines = if inbetween_new_lines > 1 {
+            1 + token.kind.precedence().is_none() as usize
+        } else {
+            prev_needs_newline as usize
+        };
+
+        if token.kind == TokenKind::Comment && reader.as_bytes()[token.end as usize - 1] != b'/' {
+            prev_needs_newline = true;
+            prev_needs_whitecpace = false;
+        } else {
+            prev_needs_newline = false;
+        }
+
+        let sstr = reader[token.start as usize..].as_ptr();
+        reader = &reader[token.end as usize..];
+        unsafe {
+            if extra_prefix_new_lines != 0 {
+                for _ in 0..extra_prefix_new_lines {
+                    writer.write(b'\n');
+                    writer = writer.add(1);
+                }
+            } else if prefix != 0 {
+                writer.write(prefix);
+                writer = writer.add(1);
+            }
+            writer.copy_from(sstr, cpy_len);
+            writer = writer.add(cpy_len);
+        }
+    }
+
+    unsafe { writer.sub_ptr(source.as_mut_ptr()) }
+}
+
+pub struct Formatter<'a> {
+    source: &'a str,
+    depth: usize,
+}
+
+// we exclusively use `write_str` to reduce bloat
+impl<'a> Formatter<'a> {
+    pub fn new(source: &'a str) -> Self {
+        Self { source, depth: 0 }
+    }
+
+    fn fmt_list<T: Poser, F: core::fmt::Write>(
+        &mut self,
+        f: &mut F,
+        trailing: bool,
+        end: &str,
+        sep: &str,
+        list: &[T],
+        fmt: impl Fn(&mut Self, &T, &mut F) -> fmt::Result,
+    ) -> fmt::Result {
+        self.fmt_list_low(f, trailing, end, sep, list, |s, v, f| {
+            fmt(s, v, f)?;
+            Ok(true)
+        })
+    }
+
+    fn fmt_list_low<T: Poser, F: core::fmt::Write>(
+        &mut self,
+        f: &mut F,
+        trailing: bool,
+        end: &str,
+        sep: &str,
+        list: &[T],
+        fmt: impl Fn(&mut Self, &T, &mut F) -> Result<bool, fmt::Error>,
+    ) -> fmt::Result {
+        if !trailing {
+            let mut first = true;
+            for expr in list {
+                if !core::mem::take(&mut first) {
+                    f.write_str(sep)?;
+                    f.write_str(" ")?;
+                }
+                first = !fmt(self, expr, f)?;
+            }
+            return f.write_str(end);
+        }
+
+        writeln!(f)?;
+        self.depth += 1;
+        let res = (|| {
+            for (i, stmt) in list.iter().enumerate() {
+                for _ in 0..self.depth {
+                    f.write_str("\t")?;
+                }
+                let add_sep = fmt(self, stmt, f)?;
+                if add_sep {
+                    f.write_str(sep)?;
+                }
+                if let Some(expr) = list.get(i + 1)
+                    && let Some(rest) = self.source.get(expr.posi() as usize..)
+                {
+                    if sep.is_empty() && insert_needed_semicolon(rest) {
+                        f.write_str(";")?;
+                    }
+                    if preserve_newlines(&self.source[..expr.posi() as usize]) > 1 {
+                        f.write_str("\n")?;
+                    }
+                }
+                if add_sep {
+                    f.write_str("\n")?;
+                }
+            }
+            Ok(())
+        })();
+        self.depth -= 1;
+
+        for _ in 0..self.depth {
+            f.write_str("\t")?;
+        }
+        f.write_str(end)?;
+        res
+    }
+
+    fn fmt_paren<F: core::fmt::Write>(
+        &mut self,
+        expr: &Expr,
+        f: &mut F,
+        cond: impl FnOnce(&Expr) -> bool,
+    ) -> fmt::Result {
+        if cond(expr) {
+            f.write_str("(")?;
+            self.fmt(expr, f)?;
+            f.write_str(")")
+        } else {
+            self.fmt(expr, f)
+        }
+    }
+
+    pub fn fmt<F: core::fmt::Write>(&mut self, expr: &Expr, f: &mut F) -> fmt::Result {
+        macro_rules! impl_parenter {
+            ($($name:ident => $pat:pat,)*) => {
+                $(
+                    let $name = |e: &Expr| matches!(e, $pat);
+                )*
+            };
+        }
+
+        impl_parenter! {
+            unary => Expr::BinOp { .. },
+            postfix => Expr::UnOp { .. } | Expr::BinOp { .. },
+            consecutive => Expr::UnOp { .. },
+        }
+
+        match *expr {
+            Expr::Ct { value, .. } => {
+                f.write_str("$: ")?;
+                self.fmt(value, f)
+            }
+            Expr::String { literal, .. } => f.write_str(literal),
+            Expr::Comment { literal, .. } => f.write_str(literal),
+            Expr::Mod { path, .. } => write!(f, "@use(\"{path}\")"),
+            Expr::Embed { path, .. } => write!(f, "@embed(\"{path}\")"),
+            Expr::Field { target, name: field, .. } => {
+                self.fmt_paren(target, f, postfix)?;
+                f.write_str(".")?;
+                f.write_str(field)
+            }
+            Expr::Directive { name, args, .. } => {
+                f.write_str("@")?;
+                f.write_str(name)?;
+                f.write_str("(")?;
+                self.fmt_list(f, false, ")", ",", args, Self::fmt)
+            }
+            Expr::Struct { fields, trailing_comma, packed, .. } => {
+                if packed {
+                    f.write_str("packed ")?;
+                }
+
+                f.write_str("struct {")?;
+                self.fmt_list_low(f, trailing_comma, "}", ",", fields, |s, field, f| {
+                    match field {
+                        CommentOr::Or(StructField { name, ty, .. }) => {
+                            f.write_str(name)?;
+                            f.write_str(": ")?;
+                            s.fmt(ty, f)?
+                        }
+                        CommentOr::Comment { literal, .. } => {
+                            f.write_str(literal)?;
+                            f.write_str("\n")?;
+                        }
+                    }
+                    Ok(field.or().is_some())
+                })
+            }
+            Expr::Enum { variants, trailing_comma, .. } => {
+                f.write_str("enum {")?;
+                self.fmt_list_low(f, trailing_comma, "}", ",", variants, |_, var, f| {
+                    match var {
+                        CommentOr::Or(EnumField { name, .. }) => {
+                            f.write_str(name)?;
+                        }
+                        CommentOr::Comment { literal, .. } => {
+                            f.write_str(literal)?;
+                            f.write_str("\n")?;
+                        }
+                    }
+                    Ok(var.or().is_some())
+                })
+            }
+            Expr::Ctor { ty, fields, trailing_comma, .. } => {
+                if let Some(ty) = ty {
+                    self.fmt_paren(ty, f, unary)?;
+                }
+                f.write_str(".{")?;
+                self.fmt_list(
+                    f,
+                    trailing_comma,
+                    "}",
+                    ",",
+                    fields,
+                    |s: &mut Self, CtorField { name, value, .. }: &_, f| {
+                        f.write_str(name)?;
+                        if !matches!(value, &Expr::Ident { id, .. } if *name == &self.source[id.range()]) {
+                            f.write_str(": ")?;
+                            s.fmt(value, f)?;
+                        }
+                        Ok(())
+                    },
+                )
+            }
+            Expr::Tupl {
+                pos,
+                ty: Some(&Expr::Slice { pos: spos, size: Some(&Expr::Number { value, .. }), item }),
+                fields,
+                trailing_comma,
+            } if value as usize == fields.len() => self.fmt(
+                &Expr::Tupl {
+                    pos,
+                    ty: Some(&Expr::Slice { pos: spos, size: None, item }),
+                    fields,
+                    trailing_comma,
+                },
+                f,
+            ),
+            Expr::Tupl { ty, fields, trailing_comma, .. } => {
+                if let Some(ty) = ty {
+                    self.fmt_paren(ty, f, unary)?;
+                }
+                f.write_str(".(")?;
+                self.fmt_list(f, trailing_comma, ")", ",", fields, Self::fmt)
+            }
+            Expr::Slice { item, size, .. } => {
+                f.write_str("[")?;
+                self.fmt(item, f)?;
+                if let Some(size) = size {
+                    f.write_str("; ")?;
+                    self.fmt(size, f)?;
+                }
+                f.write_str("]")
+            }
+            Expr::Index { base, index } => {
+                self.fmt(base, f)?;
+                f.write_str("[")?;
+                self.fmt(index, f)?;
+                f.write_str("]")
+            }
+            Expr::UnOp { op, val, .. } => {
+                f.write_str(op.name())?;
+                self.fmt_paren(val, f, unary)
+            }
+            Expr::Break { .. } => f.write_str("break"),
+            Expr::Continue { .. } => f.write_str("continue"),
+            Expr::If { cond, then, else_, .. } => {
+                f.write_str("if ")?;
+                self.fmt(cond, f)?;
+                f.write_str(" ")?;
+                self.fmt_paren(then, f, consecutive)?;
+                if let Some(e) = else_ {
+                    f.write_str(" else ")?;
+                    self.fmt(e, f)?;
+                }
+                Ok(())
+            }
+            Expr::Match { value, branches, .. } => {
+                f.write_str("match ")?;
+                self.fmt(value, f)?;
+                f.write_str(" {")?;
+                self.fmt_list(f, true, "}", ",", branches, |s, br, f| {
+                    s.fmt(&br.pat, f)?;
+                    f.write_str(" => ")?;
+                    s.fmt(&br.body, f)
+                })
+            }
+            Expr::Loop { body, .. } => {
+                f.write_str("loop ")?;
+                self.fmt(body, f)
+            }
+            Expr::Closure { ret, body, args, .. } => {
+                f.write_str("fn(")?;
+                self.fmt_list(f, false, "", ",", args, |s, arg, f| {
+                    if arg.is_ct {
+                        f.write_str("$")?;
+                    }
+                    f.write_str(arg.name)?;
+                    f.write_str(": ")?;
+                    s.fmt(&arg.ty, f)
+                })?;
+                f.write_str("): ")?;
+                self.fmt(ret, f)?;
+                f.write_str(" ")?;
+                self.fmt_paren(body, f, consecutive)?;
+                Ok(())
+            }
+            Expr::Call { func, args, trailing_comma } => {
+                self.fmt_paren(func, f, postfix)?;
+                f.write_str("(")?;
+                self.fmt_list(f, trailing_comma, ")", ",", args, Self::fmt)
+            }
+            Expr::Return { val: Some(val), .. } => {
+                f.write_str("return ")?;
+                self.fmt(val, f)
+            }
+            Expr::Return { val: None, .. } => f.write_str("return"),
+            Expr::Wildcard { .. } => f.write_str("_"),
+            Expr::Ident { pos, is_ct, .. } => {
+                if is_ct {
+                    f.write_str("$")?;
+                }
+                f.write_str(&self.source[Lexer::restore(self.source, pos).eat().range()])
+            }
+            Expr::Block { stmts, .. } => {
+                f.write_str("{")?;
+                self.fmt_list(f, true, "}", "", stmts, Self::fmt)
+            }
+            Expr::Number { value, radix, .. } => {
+                f.write_str(match radix {
+                    Radix::Decimal => "",
+                    Radix::Hex => "0x",
+                    Radix::Octal => "0o",
+                    Radix::Binary => "0b",
+                })?;
+                let mut buf = [0u8; 64];
+                f.write_str(display_radix(radix, value as u64, &mut buf))
+            }
+            Expr::Float { pos, .. } => {
+                f.write_str(&self.source[Lexer::restore(self.source, pos).eat().range()])
+            }
+            Expr::Bool { value, .. } => f.write_str(if value { "true" } else { "false" }),
+            Expr::Idk { .. } => f.write_str("idk"),
+            Expr::Die { .. } => f.write_str("die"),
+            Expr::Null { .. } => f.write_str("null"),
+            Expr::BinOp {
+                left,
+                op: TokenKind::Assign,
+                right: &Expr::BinOp { left: lleft, op, right, .. },
+                ..
+            } if left.pos() == lleft.pos() => {
+                self.fmt(left, f)?;
+                f.write_str(" ")?;
+                f.write_str(op.name())?;
+                f.write_str("= ")?;
+                self.fmt(right, f)
+            }
+            Expr::BinOp { right, op, left, .. } => {
+                let prec_miss_left = |e: &Expr| {
+                    matches!(
+                        e, Expr::BinOp { op: lop, .. } if op.precedence() > lop.precedence()
+                    )
+                };
+                let prec_miss_right = |e: &Expr| {
+                    matches!(
+                        e, Expr::BinOp { op: lop, .. }
+                            if (op.precedence() == lop.precedence() && !op.is_comutative())
+                                || op.precedence() > lop.precedence()
+                    )
+                };
+
+                self.fmt_paren(left, f, prec_miss_left)?;
+                if let Some(mut prev) = self.source.get(..right.pos() as usize) {
+                    prev = prev.trim_end();
+                    let estimate_bound =
+                        prev.rfind(|c: char| c.is_ascii_whitespace()).map_or(prev.len(), |i| i + 1);
+                    let exact_bound = lexer::Lexer::new(&prev[estimate_bound..]).last().start;
+                    prev = &prev[..exact_bound as usize + estimate_bound];
+                    if preserve_newlines(prev) > 0 {
+                        f.write_str("\n")?;
+                        for _ in 0..self.depth + 1 {
+                            f.write_str("\t")?;
+                        }
+                        f.write_str(op.name())?;
+                        f.write_str(" ")?;
+                    } else {
+                        f.write_str(" ")?;
+                        f.write_str(op.name())?;
+                        f.write_str(" ")?;
+                    }
+                } else {
+                    f.write_str(" ")?;
+                    f.write_str(op.name())?;
+                    f.write_str(" ")?;
+                }
+                self.fmt_paren(right, f, prec_miss_right)
+            }
+        }
+    }
+}
+
+pub fn preserve_newlines(source: &str) -> usize {
+    source[source.trim_end().len()..].bytes().filter(|&c| c == b'\n').count()
+}
+
+pub fn insert_needed_semicolon(source: &str) -> bool {
+    let kind = lexer::Lexer::new(source).eat().kind;
+    kind.precedence().is_some() || matches!(kind, TokenKind::Ctor | TokenKind::Tupl)
+}
+
+impl core::fmt::Display for parser::Ast {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        fmt_file(self.exprs(), &self.file, f)
+    }
+}
+
+pub fn fmt_file(exprs: &[Expr], file: &str, f: &mut impl fmt::Write) -> fmt::Result {
+    for (i, expr) in exprs.iter().enumerate() {
+        Formatter::new(file).fmt(expr, f)?;
+        if let Some(expr) = exprs.get(i + 1)
+            && let Some(rest) = file.get(expr.pos() as usize..)
+        {
+            if insert_needed_semicolon(rest) {
+                write!(f, ";")?;
+            }
+
+            if preserve_newlines(&file[..expr.pos() as usize]) > 1 {
+                writeln!(f)?;
+            }
+        }
+
+        if i + 1 != exprs.len() {
+            writeln!(f)?;
+        }
+    }
+    Ok(())
+}
+
+#[cfg(test)]
+pub mod test {
+    use {
+        crate::parser::{self, Ctx},
+        alloc::borrow::ToOwned,
+        std::{fmt::Write, string::String},
+    };
+
+    pub fn format(ident: &str, input: &str) {
+        let mut minned = input.to_owned();
+        let len = crate::fmt::minify(&mut minned);
+        minned.truncate(len);
+
+        let mut ctx = Ctx::default();
+        let ast = parser::Ast::new(ident, minned, &mut ctx, &mut parser::no_loader);
+        let mut output = String::new();
+        write!(output, "{ast}").unwrap();
+
+        let input_path = format!("formatter_{ident}.expected");
+        let output_path = format!("formatter_{ident}.actual");
+        std::fs::write(&input_path, input).unwrap();
+        std::fs::write(&output_path, output).unwrap();
+
+        let success = std::process::Command::new("diff")
+            .arg("-u")
+            .arg("--color")
+            .arg(&input_path)
+            .arg(&output_path)
+            .status()
+            .unwrap()
+            .success();
+        std::fs::remove_file(&input_path).unwrap();
+        std::fs::remove_file(&output_path).unwrap();
+        assert!(success, "test failed");
+    }
+
+    macro_rules! test {
+        ($($name:ident => $input:expr;)*) => {$(
+            #[test]
+            fn $name() {
+                format(stringify!($name), $input);
+            }
+        )*};
+    }
+
+    test! {
+        comments => "// comment\n// comment\n\n// comment\n\n\
+            /* comment */\n/* comment */\n\n/* comment */";
+        some_ordinary_code => "loft := fn(): int return loft(1, 2, 3)";
+        some_arg_per_line_code => "loft := fn(): int return loft(\
+            \n\t1,\n\t2,\n\t3,\n)";
+        some_ordinary_struct => "loft := fn(): int return loft.{a: 1, b: 2}";
+        some_ordinary_fild_per_lin_struct => "loft := fn(): int return loft.{\
+            \n\ta: 1,\n\tb: 2,\n}";
+        code_block => "loft := fn(): int {\n\tloft()\n\treturn 1\n}";
+    }
+}

lang/src/fs.rs

@@ -0,0 +1,384 @@
+use {
+    crate::{
+        parser::{Ast, Ctx, FileKind},
+        son::{self, hbvm::HbvmBackend},
+        ty, FnvBuildHasher,
+    },
+    alloc::{string::String, vec::Vec},
+    core::{fmt::Write, num::NonZeroUsize, ops::Deref},
+    hashbrown::hash_map,
+    std::{
+        collections::VecDeque,
+        eprintln,
+        ffi::OsStr,
+        io::{self, Write as _},
+        path::{Path, PathBuf},
+        string::ToString,
+        sync::Mutex,
+    },
+};
+
+type HashMap<K, V> = hashbrown::HashMap<K, V, FnvBuildHasher>;
+
+pub struct Logger;
+
+impl log::Log for Logger {
+    fn enabled(&self, _: &log::Metadata) -> bool {
+        true
+    }
+
+    fn log(&self, record: &log::Record) {
+        if self.enabled(record.metadata()) {
+            eprintln!("{}", record.args())
+        }
+    }
+
+    fn flush(&self) {}
+}
+
+#[derive(Default)]
+pub struct Options {
+    pub fmt: bool,
+    pub fmt_stdout: bool,
+    pub dump_asm: bool,
+    pub in_house_regalloc: bool,
+    pub extra_threads: usize,
+}
+
+impl Options {
+    pub fn from_args(args: &[&str], out: &mut Vec<u8>) -> std::io::Result<Self> {
+        if args.contains(&"--help") || args.contains(&"-h") {
+            writeln!(out, "Usage: hbc [OPTIONS...] <FILE>")?;
+            writeln!(out, include_str!("../command-help.txt"))?;
+            return Err(std::io::ErrorKind::Other.into());
+        }
+
+        Ok(Options {
+            fmt: args.contains(&"--fmt"),
+            fmt_stdout: args.contains(&"--fmt-stdout"),
+            dump_asm: args.contains(&"--dump-asm"),
+            in_house_regalloc: args.contains(&"--in-house-regalloc"),
+            extra_threads: args
+                .iter()
+                .position(|&a| a == "--threads")
+                .map(|i| {
+                    args[i + 1].parse::<NonZeroUsize>().map_err(|e| {
+                        writeln!(out, "--threads expects non zero integer: {e}")
+                            .err()
+                            .unwrap_or(std::io::ErrorKind::Other.into())
+                    })
+                })
+                .transpose()?
+                .map_or(1, NonZeroUsize::get)
+                - 1,
+        })
+    }
+}
+
+pub fn run_compiler(
+    root_file: &str,
+    options: Options,
+    out: &mut Vec<u8>,
+    warnings: &mut String,
+) -> std::io::Result<()> {
+    let parsed = parse_from_fs(options.extra_threads, root_file)?;
+
+    if (options.fmt || options.fmt_stdout) && !parsed.errors.is_empty() {
+        *out = parsed.errors.into_bytes();
+        return Err(std::io::Error::other("fmt fialed (errors are in out)"));
+    }
+
+    if options.fmt {
+        let mut output = String::new();
+        for ast in parsed.ast {
+            write!(output, "{ast}").unwrap();
+            if ast.file.deref().trim() != output.as_str().trim() {
+                std::fs::write(&*ast.path, &output)?;
+            }
+            output.clear();
+        }
+    } else if options.fmt_stdout {
+        write!(out, "{}", &parsed.ast[0])?;
+    } else {
+        let mut backend = HbvmBackend::default();
+        backend.use_in_house_regalloc = options.in_house_regalloc;
+
+        let mut ctx = crate::son::CodegenCtx::default();
+        *ctx.parser.errors.get_mut() = parsed.errors;
+        let mut codegen = son::Codegen::new(&mut backend, &parsed.ast, &mut ctx);
+        codegen.push_embeds(parsed.embeds);
+        codegen.generate(ty::Module::MAIN);
+
+        *warnings = core::mem::take(&mut *codegen.warnings.borrow_mut());
+
+        if !codegen.errors.borrow().is_empty() {
+            drop(codegen);
+            *out = ctx.parser.errors.into_inner().into_bytes();
+            return Err(std::io::Error::other("compilation faoled (errors are in out)"));
+        }
+
+        codegen.assemble(out);
+
+        if options.dump_asm {
+            let mut disasm = String::new();
+            codegen.disasm(&mut disasm, out).map_err(|e| io::Error::other(e.to_string()))?;
+            *out = disasm.into_bytes();
+        }
+    }
+
+    Ok(())
+}
+
+struct TaskQueue<T> {
+    inner: Mutex<TaskQueueInner<T>>,
+}
+
+impl<T> TaskQueue<T> {
+    fn new(max_waiters: usize) -> Self {
+        Self { inner: Mutex::new(TaskQueueInner::new(max_waiters)) }
+    }
+
+    pub fn push(&self, message: T) {
+        self.extend([message]);
+    }
+
+    pub fn extend(&self, messages: impl IntoIterator<Item = T>) {
+        self.inner.lock().unwrap().push(messages);
+    }
+
+    pub fn pop(&self) -> Option<T> {
+        TaskQueueInner::pop(&self.inner)
+    }
+}
+
+enum TaskSlot<T> {
+    Waiting,
+    Delivered(T),
+    Closed,
+}
+
+struct TaskQueueInner<T> {
+    max_waiters: usize,
+    messages: VecDeque<T>,
+    parked: VecDeque<(*mut TaskSlot<T>, std::thread::Thread)>,
+}
+
+unsafe impl<T: Send> Send for TaskQueueInner<T> {}
+unsafe impl<T: Send + Sync> Sync for TaskQueueInner<T> {}
+
+impl<T> TaskQueueInner<T> {
+    fn new(max_waiters: usize) -> Self {
+        Self { max_waiters, messages: Default::default(), parked: Default::default() }
+    }
+
+    fn push(&mut self, messages: impl IntoIterator<Item = T>) {
+        for msg in messages {
+            if let Some((dest, thread)) = self.parked.pop_front() {
+                unsafe { *dest = TaskSlot::Delivered(msg) };
+                thread.unpark();
+            } else {
+                self.messages.push_back(msg);
+            }
+        }
+    }
+
+    fn pop(s: &Mutex<Self>) -> Option<T> {
+        let mut res = TaskSlot::Waiting;
+        {
+            let mut s = s.lock().unwrap();
+            if let Some(msg) = s.messages.pop_front() {
+                return Some(msg);
+            }
+
+            if s.max_waiters == s.parked.len() + 1 {
+                for (dest, thread) in s.parked.drain(..) {
+                    unsafe { *dest = TaskSlot::Closed };
+                    thread.unpark();
+                }
+                return None;
+            }
+
+            s.parked.push_back((&mut res, std::thread::current()));
+        }
+
+        loop {
+            std::thread::park();
+
+            let _s = s.lock().unwrap();
+            match core::mem::replace(&mut res, TaskSlot::Waiting) {
+                TaskSlot::Delivered(msg) => return Some(msg),
+                TaskSlot::Closed => return None,
+                TaskSlot::Waiting => {}
+            }
+        }
+    }
+}
+
+pub struct Loaded {
+    ast: Vec<Ast>,
+    embeds: Vec<Vec<u8>>,
+    errors: String,
+}
+
+pub fn parse_from_fs(extra_threads: usize, root: &str) -> io::Result<Loaded> {
+    fn resolve(path: &str, from: &str, tmp: &mut PathBuf) -> Result<PathBuf, CantLoadFile> {
+        tmp.clear();
+        match Path::new(from).parent() {
+            Some(parent) => tmp.extend([parent, Path::new(path)]),
+            None => tmp.push(path),
+        };
+
+        tmp.canonicalize().map_err(|source| CantLoadFile { path: std::mem::take(tmp), source })
+    }
+
+    #[derive(Debug)]
+    struct CantLoadFile {
+        path: PathBuf,
+        source: io::Error,
+    }
+
+    impl core::fmt::Display for CantLoadFile {
+        fn fmt(&self, f: &mut core::fmt::Formatter) -> core::fmt::Result {
+            write!(f, "can't load file: {}", display_rel_path(&self.path),)
+        }
+    }
+
+    impl core::error::Error for CantLoadFile {
+        fn source(&self) -> Option<&(dyn core::error::Error + 'static)> {
+            Some(&self.source)
+        }
+    }
+
+    impl From<CantLoadFile> for io::Error {
+        fn from(e: CantLoadFile) -> Self {
+            io::Error::new(io::ErrorKind::InvalidData, e)
+        }
+    }
+
+    type Task = (usize, PathBuf);
+
+    let seen_modules = Mutex::new(HashMap::<PathBuf, usize>::default());
+    let seen_embeds = Mutex::new(HashMap::<PathBuf, usize>::default());
+    let tasks = TaskQueue::<Task>::new(extra_threads + 1);
+    let ast = Mutex::new(Vec::<io::Result<Ast>>::new());
+    let embeds = Mutex::new(Vec::<Vec<u8>>::new());
+
+    let loader = |path: &str, from: &str, kind: FileKind, tmp: &mut _| {
+        let mut physiscal_path = resolve(path, from, tmp)?;
+
+        match kind {
+            FileKind::Module => {
+                let id = {
+                    let mut seen = seen_modules.lock().unwrap();
+                    let len = seen.len();
+                    match seen.entry(physiscal_path) {
+                        hash_map::Entry::Occupied(entry) => {
+                            return Ok(*entry.get());
+                        }
+                        hash_map::Entry::Vacant(entry) => {
+                            physiscal_path = entry.insert_entry(len as _).key().clone();
+                            len
+                        }
+                    }
+                };
+
+                if !physiscal_path.exists() {
+                    return Err(io::Error::new(
+                        io::ErrorKind::NotFound,
+                        format!("can't find file: {}", display_rel_path(&physiscal_path)),
+                    ));
+                }
+
+                tasks.push((id, physiscal_path));
+                Ok(id)
+            }
+            FileKind::Embed => {
+                let id = {
+                    let mut seen = seen_embeds.lock().unwrap();
+                    let len = seen.len();
+                    match seen.entry(physiscal_path) {
+                        hash_map::Entry::Occupied(entry) => {
+                            return Ok(*entry.get());
+                        }
+                        hash_map::Entry::Vacant(entry) => {
+                            physiscal_path = entry.insert_entry(len as _).key().clone();
+                            len
+                        }
+                    }
+                };
+
+                let content = std::fs::read(&physiscal_path).map_err(|e| {
+                    io::Error::new(
+                        e.kind(),
+                        format!(
+                            "can't load embed file: {}: {e}",
+                            display_rel_path(&physiscal_path)
+                        ),
+                    )
+                })?;
+                let mut embeds = embeds.lock().unwrap();
+                if id as usize >= embeds.len() {
+                    embeds.resize(id as usize + 1, Default::default());
+                }
+                embeds[id as usize] = content;
+                Ok(id)
+            }
+        }
+    };
+
+    let execute_task = |ctx: &mut _, (_, path): Task, tmp: &mut _| {
+        let path = path.to_str().ok_or_else(|| {
+            io::Error::new(
+                io::ErrorKind::InvalidData,
+                format!("path contains invalid characters: {}", display_rel_path(&path)),
+            )
+        })?;
+        Ok(Ast::new(path, std::fs::read_to_string(path)?, ctx, &mut |path, from, kind| {
+            loader(path, from, kind, tmp).map_err(|e| e.to_string())
+        }))
+    };
+
+    let thread = || {
+        let mut ctx = Ctx::default();
+        let mut tmp = PathBuf::new();
+        while let Some(task @ (indx, ..)) = tasks.pop() {
+            let res = execute_task(&mut ctx, task, &mut tmp);
+            let mut ast = ast.lock().unwrap();
+            let len = ast.len().max(indx + 1);
+            ast.resize_with(len, || Err(io::ErrorKind::InvalidData.into()));
+            ast[indx] = res;
+        }
+        ctx.errors.into_inner()
+    };
+
+    let path = Path::new(root).canonicalize().map_err(|e| {
+        io::Error::new(e.kind(), format!("can't canonicalize root file path ({root})"))
+    })?;
+    seen_modules.lock().unwrap().insert(path.clone(), 0);
+    tasks.push((0, path));
+
+    let errors = if extra_threads == 0 {
+        thread()
+    } else {
+        std::thread::scope(|s| {
+            (0..extra_threads + 1)
+                .map(|_| s.spawn(thread))
+                .collect::<Vec<_>>()
+                .into_iter()
+                .map(|t| t.join().unwrap())
+                .collect::<String>()
+        })
+    };
+
+    Ok(Loaded {
+        ast: ast.into_inner().unwrap().into_iter().collect::<io::Result<Vec<_>>>()?,
+        embeds: embeds.into_inner().unwrap(),
+        errors,
+    })
+}
+
+pub fn display_rel_path(path: &(impl AsRef<OsStr> + ?Sized)) -> std::path::Display {
+    static CWD: std::sync::LazyLock<PathBuf> =
+        std::sync::LazyLock::new(|| std::env::current_dir().unwrap_or_default());
+    std::path::Path::new(path).strip_prefix(&*CWD).unwrap_or(std::path::Path::new(path)).display()
+}

lang/src/fuzz.rs

@@ -0,0 +1,141 @@
+use {
+    crate::{
+        lexer::TokenKind,
+        parser,
+        son::{hbvm::HbvmBackend, Codegen, CodegenCtx},
+        ty::Module,
+    },
+    alloc::string::String,
+    core::{fmt::Write, hash::BuildHasher, ops::Range},
+};
+
+#[derive(Default)]
+struct Rand(pub u64);
+
+impl Rand {
+    pub fn next(&mut self) -> u64 {
+        self.0 = crate::FnvBuildHasher::default().hash_one(self.0);
+        self.0
+    }
+
+    pub fn range(&mut self, min: u64, max: u64) -> u64 {
+        self.next() % (max - min) + min
+    }
+
+    fn bool(&mut self) -> bool {
+        self.next() % 2 == 0
+    }
+}
+
+#[derive(Default)]
+struct FuncGen {
+    rand: Rand,
+    buf: String,
+    vars: u64,
+}
+
+impl FuncGen {
+    fn gen(&mut self, seed: u64) -> &str {
+        self.rand = Rand(seed);
+        self.buf.clear();
+        self.buf.push_str("main := fn(): void ");
+        self.block().unwrap();
+        &self.buf
+    }
+
+    fn block(&mut self) -> core::fmt::Result {
+        let prev_vars = self.vars;
+        self.buf.push('{');
+        for _ in 0..self.rand.range(1, 10) {
+            self.stmt()?;
+        }
+        self.buf.push('}');
+        self.vars = prev_vars;
+
+        Ok(())
+    }
+
+    fn stmt(&mut self) -> core::fmt::Result {
+        match self.rand.range(0, 100) {
+            0..4 => _ = self.block(),
+            4..10 => {
+                write!(self.buf, "var{} := ", self.vars)?;
+                self.expr()?;
+                self.vars += 1;
+            }
+
+            10..20 if self.vars != 0 => {
+                write!(self.buf, "var{} = ", self.rand.range(0, self.vars))?;
+                self.expr()?;
+            }
+            20..23 => {
+                self.buf.push_str("if ");
+                self.expr()?;
+                self.block()?;
+                if self.rand.bool() {
+                    self.buf.push_str(" else ");
+                    self.block()?;
+                }
+            }
+            _ => {
+                self.buf.push_str("return ");
+                self.expr()?;
+            }
+        }
+
+        self.buf.push(';');
+        Ok(())
+    }
+
+    fn expr(&mut self) -> core::fmt::Result {
+        match self.rand.range(0, 100) {
+            0..80 => {
+                write!(self.buf, "{}", self.rand.next())
+            }
+            80..90 if self.vars != 0 => {
+                write!(self.buf, "var{}", self.rand.range(0, self.vars))
+            }
+            80..100 => {
+                self.expr()?;
+                let ops = [
+                    TokenKind::Add,
+                    TokenKind::Sub,
+                    TokenKind::Mul,
+                    TokenKind::Div,
+                    TokenKind::Shl,
+                    TokenKind::Eq,
+                    TokenKind::Ne,
+                    TokenKind::Lt,
+                    TokenKind::Gt,
+                    TokenKind::Le,
+                    TokenKind::Ge,
+                    TokenKind::Band,
+                    TokenKind::Bor,
+                    TokenKind::Xor,
+                    TokenKind::Mod,
+                    TokenKind::Shr,
+                ];
+                let op = ops[self.rand.range(0, ops.len() as u64) as usize];
+                write!(self.buf, " {op} ")?;
+                self.expr()
+            }
+            _ => unreachable!(),
+        }
+    }
+}
+
+pub fn fuzz(seed_range: Range<u64>) {
+    let mut gen = FuncGen::default();
+    let mut ctx = CodegenCtx::default();
+    for i in seed_range {
+        ctx.clear();
+        let src = gen.gen(i);
+        let parsed = parser::Ast::new("fuzz", src, &mut ctx.parser, &mut parser::no_loader);
+
+        assert!(ctx.parser.errors.get_mut().is_empty());
+
+        let mut backend = HbvmBackend::default();
+        let mut cdg = Codegen::new(&mut backend, core::slice::from_ref(&parsed), &mut ctx);
+        cdg.generate(Module::MAIN);
+    }
+}

lang/src/fuzz_main.rs

@@ -0,0 +1,3 @@
+fn main() {
+    hblang::fuzz::fuzz(0..1000000);
+}

lang/src/lexer.rs

@@ -0,0 +1,567 @@
+const fn ascii_mask(chars: &[u8]) -> u128 {
+    let mut eq = 0;
+    let mut i = 0;
+    while i < chars.len() {
+        let b = chars[i];
+        eq |= 1 << b;
+        i += 1;
+    }
+    eq
+}
+
+#[derive(Debug, PartialEq, Eq, Clone, Copy)]
+pub struct Token {
+    pub kind: TokenKind,
+    pub start: u32,
+    pub end: u32,
+}
+
+impl Token {
+    pub fn range(&self) -> core::ops::Range<usize> {
+        self.start as usize..self.end as usize
+    }
+}
+
+macro_rules! gen_token_kind {
+    ($(
+        #[$atts:meta])*
+        $vis:vis enum $name:ident {
+            #[patterns] $(
+                $pattern:ident,
+            )*
+            #[keywords] $(
+                $keyword:ident = $keyword_lit:literal,
+            )*
+            #[punkt] $(
+                $punkt:ident = $punkt_lit:literal,
+            )*
+            #[ops] $(
+                #[$prec:ident] $(
+                    $op:ident = $op_lit:literal $(=> $assign:ident)?,
+                )*
+            )*
+        }
+    ) => {
+        impl core::fmt::Display for $name {
+            fn fmt(&self, f: &mut core::fmt::Formatter) -> core::fmt::Result {
+                f.write_str(self.name())
+            }
+        }
+
+        impl $name {
+            pub const OPS: &[Self] = &[$($(Self::$op),*),*];
+
+            pub fn name(&self) -> &str {
+                let sf = unsafe { &*(self as *const _ as *const u8) } ;
+                match *self {
+                    $( Self::$pattern => concat!('<', stringify!($pattern), '>'), )*
+                    $( Self::$keyword => stringify!($keyword_lit), )*
+                    $( Self::$punkt   => stringify!($punkt_lit),   )*
+                    $($( Self::$op    => $op_lit,
+                      $(Self::$assign => concat!($op_lit, "="),)?)*)*
+                    _ => unsafe { core::str::from_utf8_unchecked(core::slice::from_ref(&sf)) },
+                }
+            }
+
+            #[inline(always)]
+            pub fn precedence(&self) -> Option<u8> {
+                Some(match self {
+                    $($(Self::$op => ${ignore($prec)} ${index(1)},
+                      $(Self::$assign => 0,)?)*)*
+                    _ => return None,
+                } + 1)
+            }
+
+            fn from_ident(ident: &[u8]) -> Self {
+                match ident {
+                    $($keyword_lit => Self::$keyword,)*
+                    _ => Self::Ident,
+                }
+            }
+        }
+    };
+}
+
+#[derive(PartialEq, Eq, Clone, Copy, Hash, PartialOrd, Ord)]
+#[repr(u8)]
+pub enum TokenKind {
+    Not = b'!',
+    DQuote = b'"',
+    Pound = b'#',
+    CtIdent = b'$',
+    Mod = b'%',
+    Band = b'&',
+    Quote = b'\'',
+    LParen = b'(',
+    RParen = b')',
+    Mul = b'*',
+    Add = b'+',
+    Comma = b',',
+    Sub = b'-',
+    Dot = b'.',
+    Div = b'/',
+    // Unused = 2-6
+    Shl = b'<' - 5,
+    // Unused = 8
+    Shr = b'>' - 5,
+    Colon = b':',
+    Semi = b';',
+    Lt = b'<',
+    Assign = b'=',
+    Gt = b'>',
+    Que = b'?',
+    Directive = b'@',
+
+    Comment,
+
+    Ident,
+    Number,
+    Float,
+    Eof,
+
+    Ct,
+
+    Ctor,
+    Tupl,
+    TArrow,
+
+    Or,
+    And,
+
+    // Unused = R-Z
+    LBrack = b'[',
+    BSlash = b'\\',
+    RBrack = b']',
+    Xor = b'^',
+    Under = b'_',
+    Tick = b'`',
+
+    Return,
+    If,
+    Match,
+    Else,
+    Loop,
+    Break,
+    Continue,
+    Fn,
+    Struct,
+    Packed,
+    Enum,
+    True,
+    False,
+    Null,
+    Idk,
+    Die,
+
+    // Unused = a-z
+    LBrace = b'{',
+    Bor = b'|',
+    RBrace = b'}',
+    Tilde = b'~',
+
+    Decl = b':' + 128,
+    Eq = b'=' + 128,
+    Ne = b'!' + 128,
+    Le = b'<' + 128,
+    Ge = b'>' + 128,
+
+    BorAss = b'|' + 128,
+    AddAss = b'+' + 128,
+    SubAss = b'-' + 128,
+    MulAss = b'*' + 128,
+    DivAss = b'/' + 128,
+    ModAss = b'%' + 128,
+    XorAss = b'^' + 128,
+    BandAss = b'&' + 128,
+    ShrAss = b'>' - 5 + 128,
+    ShlAss = b'<' - 5 + 128,
+}
+
+impl core::fmt::Debug for TokenKind {
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+        core::fmt::Display::fmt(self, f)
+    }
+}
+
+impl TokenKind {
+    pub fn ass_op(self) -> Option<Self> {
+        let id = (self as u8).saturating_sub(128);
+        if ascii_mask(b"|+-*/%^&79") & (1u128 << id) == 0 {
+            return None;
+        }
+        Some(unsafe { core::mem::transmute::<u8, Self>(id) })
+    }
+
+    pub fn is_comutative(self) -> bool {
+        use TokenKind as S;
+        matches!(self, S::Eq | S::Ne | S::Bor | S::Xor | S::Band | S::Add | S::Mul)
+    }
+
+    pub fn is_compatison(self) -> bool {
+        matches!(self, Self::Lt | Self::Gt | Self::Ge | Self::Le | Self::Ne | Self::Eq)
+    }
+
+    pub fn is_supported_float_op(self) -> bool {
+        matches!(
+            self,
+            Self::Add
+                | Self::Sub
+                | Self::Mul
+                | Self::Div
+                | Self::Eq
+                | Self::Ne
+                | Self::Le
+                | Self::Ge
+                | Self::Lt
+                | Self::Gt
+        )
+    }
+
+    pub fn apply_binop(self, a: i64, b: i64, float: bool) -> i64 {
+        if float {
+            debug_assert!(self.is_supported_float_op());
+            let [a, b] = [f64::from_bits(a as _), f64::from_bits(b as _)];
+            let res = match self {
+                Self::Add => a + b,
+                Self::Sub => a - b,
+                Self::Mul => a * b,
+                Self::Div => a / b,
+                Self::Eq => return (a == b) as i64,
+                Self::Ne => return (a != b) as i64,
+                Self::Lt => return (a < b) as i64,
+                Self::Gt => return (a > b) as i64,
+                Self::Le => return (a >= b) as i64,
+                Self::Ge => return (a <= b) as i64,
+                _ => todo!("floating point op: {self}"),
+            };
+
+            return res.to_bits() as _;
+        }
+
+        match self {
+            Self::Add => a.wrapping_add(b),
+            Self::Sub => a.wrapping_sub(b),
+            Self::Mul => a.wrapping_mul(b),
+            Self::Div if b == 0 => 0,
+            Self::Div => a.wrapping_div(b),
+            Self::Shl => a.wrapping_shl(b as _),
+            Self::Eq => (a == b) as i64,
+            Self::Ne => (a != b) as i64,
+            Self::Lt => (a < b) as i64,
+            Self::Gt => (a > b) as i64,
+            Self::Le => (a >= b) as i64,
+            Self::Ge => (a <= b) as i64,
+            Self::Band => a & b,
+            Self::Bor => a | b,
+            Self::Xor => a ^ b,
+            Self::Mod if b == 0 => 0,
+            Self::Mod => a.wrapping_rem(b),
+            Self::Shr => a.wrapping_shr(b as _),
+            s => todo!("{s}"),
+        }
+    }
+
+    pub fn is_homogenous(&self) -> bool {
+        self.precedence() != Self::Eq.precedence()
+            && self.precedence() != Self::Gt.precedence()
+            && self.precedence() != Self::Eof.precedence()
+    }
+
+    pub fn apply_unop(&self, value: i64, float: bool) -> i64 {
+        match self {
+            Self::Sub if float => (-f64::from_bits(value as _)).to_bits() as _,
+            Self::Sub => value.wrapping_neg(),
+            Self::Not => (value == 0) as _,
+            Self::Float if float => value,
+            Self::Float => (value as f64).to_bits() as _,
+            Self::Number if float => f64::from_bits(value as _) as _,
+            Self::Number => value,
+            s => todo!("{s}"),
+        }
+    }
+
+    pub fn closing(&self) -> Option<TokenKind> {
+        Some(match self {
+            Self::Ctor => Self::RBrace,
+            Self::Tupl => Self::RParen,
+            Self::LParen => Self::RParen,
+            Self::LBrack => Self::RBrack,
+            Self::LBrace => Self::RBrace,
+            _ => return None,
+        })
+    }
+}
+
+gen_token_kind! {
+    pub enum TokenKind {
+        #[patterns]
+        CtIdent,
+        Ident,
+        Number,
+        Float,
+        Eof,
+        Directive,
+        #[keywords]
+        Return    = b"return",
+        If        = b"if",
+        Match     = b"match",
+        Else      = b"else",
+        Loop      = b"loop",
+        Break     = b"break",
+        Continue  = b"continue",
+        Fn        = b"fn",
+        Struct    = b"struct",
+        Packed    = b"packed",
+        Enum      = b"enum",
+        True      = b"true",
+        False     = b"false",
+        Null      = b"null",
+        Idk       = b"idk",
+        Die       = b"die",
+        Under     = b"_",
+        #[punkt]
+        Ctor   = ".{",
+        Tupl   = ".(",
+        TArrow = "=>",
+        // #define OP: each `#[prec]` delimeters a level of precedence from lowest to highest
+        #[ops]
+        #[prec]
+        // this also includess all `<op>=` tokens
+        Decl   = ":=",
+        Assign = "=",
+        #[prec]
+        Or = "||",
+        #[prec]
+        And = "&&",
+        #[prec]
+        Bor = "|" => BorAss,
+        #[prec]
+        Xor = "^" => XorAss,
+        #[prec]
+        Band = "&" => BandAss,
+        #[prec]
+        Eq = "==",
+        Ne = "!=",
+        #[prec]
+        Le = "<=",
+        Ge = ">=",
+        Lt = "<",
+        Gt = ">",
+        #[prec]
+        Shl = "<<" => ShlAss,
+        Shr = ">>" => ShrAss,
+        #[prec]
+        Add = "+" => AddAss,
+        Sub = "-" => SubAss,
+        #[prec]
+        Mul = "*" => MulAss,
+        Div = "/" => DivAss,
+        Mod = "%" => ModAss,
+    }
+}
+
+pub struct Lexer<'a> {
+    pos: u32,
+    source: &'a [u8],
+}
+
+impl<'a> Lexer<'a> {
+    pub fn new(input: &'a str) -> Self {
+        Self::restore(input, 0)
+    }
+
+    pub fn uses(input: &'a str) -> impl Iterator<Item = &'a str> {
+        let mut s = Self::new(input);
+        core::iter::from_fn(move || loop {
+            let t = s.eat();
+            if t.kind == TokenKind::Eof {
+                return None;
+            }
+            if t.kind == TokenKind::Directive
+                && s.slice(t.range()) == "use"
+                && s.eat().kind == TokenKind::LParen
+            {
+                let t = s.eat();
+                if t.kind == TokenKind::DQuote {
+                    return Some(&s.slice(t.range())[1..t.range().len() - 1]);
+                }
+            }
+        })
+    }
+
+    pub fn restore(input: &'a str, pos: u32) -> Self {
+        Self { pos, source: input.as_bytes() }
+    }
+
+    pub fn source(&self) -> &'a str {
+        unsafe { core::str::from_utf8_unchecked(self.source) }
+    }
+
+    pub fn slice(&self, tok: core::ops::Range<usize>) -> &'a str {
+        unsafe { core::str::from_utf8_unchecked(&self.source[tok]) }
+    }
+
+    fn peek(&self) -> Option<u8> {
+        if core::intrinsics::unlikely(self.pos >= self.source.len() as u32) {
+            None
+        } else {
+            Some(unsafe { *self.source.get_unchecked(self.pos as usize) })
+        }
+    }
+
+    fn advance(&mut self) -> Option<u8> {
+        let c = self.peek()?;
+        self.pos += 1;
+        Some(c)
+    }
+
+    pub fn last(&mut self) -> Token {
+        let mut token = self.eat();
+        loop {
+            let next = self.eat();
+            if next.kind == TokenKind::Eof {
+                break;
+            }
+            token = next;
+        }
+        token
+    }
+
+    pub fn eat(&mut self) -> Token {
+        use TokenKind as T;
+        loop {
+            let mut start = self.pos;
+
+            let Some(c) = self.advance() else {
+                return Token { kind: T::Eof, start, end: self.pos };
+            };
+
+            let advance_ident = |s: &mut Self| {
+                while let Some(b'a'..=b'z' | b'A'..=b'Z' | b'0'..=b'9' | b'_' | 127..) = s.peek() {
+                    s.advance();
+                }
+            };
+
+            let identity = |s: u8| unsafe { core::mem::transmute::<u8, T>(s) };
+
+            let kind = match c {
+                ..=b' ' => continue,
+                b'0' if self.advance_if(b'x') => {
+                    while let Some(b'0'..=b'9' | b'A'..=b'F' | b'a'..=b'f') = self.peek() {
+                        self.advance();
+                    }
+                    T::Number
+                }
+                b'0' if self.advance_if(b'b') => {
+                    while let Some(b'0' | b'1') = self.peek() {
+                        self.advance();
+                    }
+                    T::Number
+                }
+                b'0' if self.advance_if(b'o') => {
+                    while let Some(b'0'..=b'7') = self.peek() {
+                        self.advance();
+                    }
+                    T::Number
+                }
+                b'0'..=b'9' => {
+                    while let Some(b'0'..=b'9') = self.peek() {
+                        self.advance();
+                    }
+
+                    if self.advance_if(b'.') {
+                        while let Some(b'0'..=b'9') = self.peek() {
+                            self.advance();
+                        }
+                        T::Float
+                    } else {
+                        T::Number
+                    }
+                }
+                b'a'..=b'z' | b'A'..=b'Z' | b'_' | 127.. => {
+                    advance_ident(self);
+                    let ident = &self.source[start as usize..self.pos as usize];
+                    T::from_ident(ident)
+                }
+                b'"' | b'\'' => loop {
+                    match self.advance() {
+                        Some(b'\\') => _ = self.advance(),
+                        Some(nc) if nc == c => break identity(c),
+                        Some(_) => {}
+                        None => break T::Eof,
+                    }
+                },
+                b'/' if self.advance_if(b'/') => {
+                    while let Some(l) = self.peek()
+                        && l != b'\n'
+                    {
+                        self.pos += 1;
+                    }
+
+                    let end = self.source[..self.pos as usize]
+                        .iter()
+                        .rposition(|&b| !b.is_ascii_whitespace())
+                        .map_or(self.pos, |i| i as u32 + 1);
+
+                    return Token { kind: T::Comment, start, end };
+                }
+                b'/' if self.advance_if(b'*') => {
+                    let mut depth = 1;
+                    while let Some(l) = self.advance() {
+                        match l {
+                            b'/' if self.advance_if(b'*') => depth += 1,
+                            b'*' if self.advance_if(b'/') => match depth {
+                                1 => break,
+                                _ => depth -= 1,
+                            },
+                            _ => {}
+                        }
+                    }
+                    T::Comment
+                }
+                b'.' if self.advance_if(b'{') => T::Ctor,
+                b'.' if self.advance_if(b'(') => T::Tupl,
+                b'=' if self.advance_if(b'>') => T::TArrow,
+                b'&' if self.advance_if(b'&') => T::And,
+                b'|' if self.advance_if(b'|') => T::Or,
+                b'$' if self.advance_if(b':') => T::Ct,
+                b'@' | b'$' => {
+                    start += 1;
+                    advance_ident(self);
+                    identity(c)
+                }
+                b'<' | b'>' if self.advance_if(c) => {
+                    identity(c - 5 + 128 * self.advance_if(b'=') as u8)
+                }
+                b':' | b'=' | b'!' | b'<' | b'>' | b'|' | b'+' | b'-' | b'*' | b'/' | b'%'
+                | b'^' | b'&'
+                    if self.advance_if(b'=') =>
+                {
+                    identity(c + 128)
+                }
+                _ => identity(c),
+            };
+
+            return Token { kind, start, end: self.pos };
+        }
+    }
+
+    fn advance_if(&mut self, arg: u8) -> bool {
+        if self.peek() == Some(arg) {
+            self.advance();
+            true
+        } else {
+            false
+        }
+    }
+}
+
+pub fn line_col(bytes: &[u8], pos: u32) -> (usize, usize) {
+    bytes[..pos as usize]
+        .split(|&b| b == b'\n')
+        .map(<[u8]>::len)
+        .enumerate()
+        .last()
+        .map(|(line, col)| (line + 1, col + 1))
+        .unwrap_or((1, 1))
+}

lang/src/main.rs

@@ -0,0 +1,31 @@
+#[cfg(feature = "std")]
+fn main() {
+    use std::io::Write;
+
+    fn run(out: &mut Vec<u8>, warnings: &mut String) -> std::io::Result<()> {
+        let args = std::env::args().collect::<Vec<_>>();
+        let args = args.iter().map(String::as_str).collect::<Vec<_>>();
+
+        let opts = hblang::Options::from_args(&args, out)?;
+        let file = args.iter().filter(|a| !a.starts_with('-')).nth(1).copied().unwrap_or("main.hb");
+
+        hblang::run_compiler(file, opts, out, warnings)
+    }
+
+    log::set_logger(&hblang::fs::Logger).unwrap();
+    log::set_max_level(log::LevelFilter::Error);
+
+    let mut out = Vec::new();
+    let mut warnings = String::new();
+    match run(&mut out, &mut warnings) {
+        Ok(_) => {
+            std::io::stderr().write_all(warnings.as_bytes()).unwrap();
+            std::io::stdout().write_all(&out).unwrap()
+        }
+        Err(_) => {
+            std::io::stderr().write_all(warnings.as_bytes()).unwrap();
+            std::io::stderr().write_all(&out).unwrap();
+            std::process::exit(1);
+        }
+    }
+}

lang/src/son/hbvm/regalloc.rs

@@ -0,0 +1,772 @@
+use {
+    super::{HbvmBackend, Nid, Nodes},
+    crate::{
+        parser,
+        reg::{self, Reg},
+        son::{debug_assert_matches, Kind, ARG_START, MEM, VOID},
+        ty::{self, Arg, Loc},
+        utils::BitSet,
+        PLoc, Sig, Types,
+    },
+    alloc::{borrow::ToOwned, vec::Vec},
+    core::{mem, ops::Range},
+    hbbytecode::{self as instrs},
+};
+
+impl HbvmBackend {
+    pub(super) fn emit_body_code(
+        &mut self,
+        nodes: &Nodes,
+        sig: Sig,
+        tys: &Types,
+        files: &[parser::Ast],
+    ) -> (usize, bool) {
+        let tail = Function::build(nodes, tys, &mut self.ralloc, sig);
+
+        let strip_load = |value| match nodes[value].kind {
+            Kind::Load { .. } if nodes[value].ty.loc(tys) == Loc::Stack => nodes[value].inputs[1],
+            _ => value,
+        };
+
+        let mut res = mem::take(&mut self.ralloc);
+
+        Regalloc::run(nodes, &mut res);
+
+        '_open_function: {
+            self.emit(instrs::addi64(reg::STACK_PTR, reg::STACK_PTR, 0));
+            self.emit(instrs::st(reg::RET_ADDR + tail as u8, reg::STACK_PTR, 0, 0));
+        }
+
+        res.node_to_reg[MEM as usize] = res.bundles.len() as u8 + 1;
+
+        let reg_offset = if tail { reg::RET + 12 } else { reg::RET_ADDR + 1 };
+
+        res.node_to_reg.iter_mut().filter(|r| **r != 0).for_each(|r| {
+            if *r == u8::MAX {
+                *r = 0
+            } else {
+                *r += reg_offset - 1;
+                if tail && *r >= reg::RET_ADDR {
+                    *r += 1;
+                }
+            }
+        });
+
+        let atr = |allc: Nid| {
+            let allc = strip_load(allc);
+            debug_assert_eq!(
+                nodes[allc].lock_rc.get(),
+                0,
+                "{:?} {}",
+                nodes[allc],
+                ty::Display::new(tys, files, nodes[allc].ty)
+            );
+            res.node_to_reg[allc as usize]
+        };
+
+        let (retl, mut parama) = tys.parama(sig.ret);
+        let mut typs = sig.args.args();
+        let mut args = nodes[VOID].outputs[ARG_START..].iter();
+        while let Some(aty) = typs.next(tys) {
+            let Arg::Value(ty) = aty else { continue };
+            let Some(loc) = parama.next(ty, tys) else { continue };
+            let &arg = args.next().unwrap();
+            let (rg, size) = match loc {
+                PLoc::WideReg(rg, size) => (rg, size),
+                PLoc::Reg(rg, size) if ty.loc(tys) == Loc::Stack => (rg, size),
+                PLoc::Reg(r, ..) | PLoc::Ref(r, ..) => {
+                    self.emit_cp(atr(arg), r);
+                    continue;
+                }
+            };
+            self.emit(instrs::st(rg, reg::STACK_PTR, self.offsets[arg as usize] as _, size));
+            if nodes.is_unlocked(arg) {
+                self.emit(instrs::addi64(rg, reg::STACK_PTR, self.offsets[arg as usize] as _));
+            }
+            self.emit_cp(atr(arg), rg);
+        }
+
+        let mut alloc_buf = vec![];
+        for (i, block) in res.blocks.iter().enumerate() {
+            self.offsets[block.entry as usize] = self.code.len() as _;
+            for &nid in &res.instrs[block.range()] {
+                if nid == VOID {
+                    continue;
+                }
+
+                let node = &nodes[nid];
+                alloc_buf.clear();
+
+                let atr = |allc: Nid| {
+                    let allc = strip_load(allc);
+                    debug_assert_eq!(
+                        nodes[allc].lock_rc.get(),
+                        0,
+                        "{:?} {}",
+                        nodes[allc],
+                        ty::Display::new(tys, files, nodes[allc].ty)
+                    );
+                    #[cfg(debug_assertions)]
+                    debug_assert!(
+                        res.marked.contains(&(allc, nid))
+                            || nid == allc
+                            || nodes.is_hard_zero(allc)
+                            || allc == MEM
+                            || matches!(node.kind, Kind::Loop | Kind::Region),
+                        "{nid} {:?}\n{allc} {:?}",
+                        nodes[nid],
+                        nodes[allc]
+                    );
+                    res.node_to_reg[allc as usize]
+                };
+
+                let mut is_next_block = false;
+                match node.kind {
+                    Kind::If => {
+                        let &[_, cnd] = node.inputs.as_slice() else { unreachable!() };
+                        if nodes.cond_op(cnd).is_some() {
+                            let &[_, lh, rh] = nodes[cnd].inputs.as_slice() else { unreachable!() };
+                            alloc_buf.extend([atr(lh), atr(rh)]);
+                        } else {
+                            alloc_buf.push(atr(cnd));
+                        }
+                    }
+                    Kind::Loop | Kind::Region => {
+                        let index = node
+                            .inputs
+                            .iter()
+                            .position(|&n| block.entry == nodes.idom_of(n))
+                            .unwrap()
+                            + 1;
+
+                        let mut moves = vec![];
+                        for &out in node.outputs.iter() {
+                            if nodes[out].is_data_phi() {
+                                let src = nodes[out].inputs[index];
+                                if atr(out) != atr(src) {
+                                    moves.push([atr(out), atr(src), 0]);
+                                }
+                            }
+                        }
+
+                        debug_assert_eq!(moves.len(), {
+                            moves.sort_unstable();
+                            moves.dedup();
+                            moves.len()
+                        });
+
+                        moves.sort_unstable_by(|[aa, ab, _], [ba, bb, _]| {
+                            if aa == bb && ab == ba {
+                                core::cmp::Ordering::Equal
+                            } else if aa == bb {
+                                core::cmp::Ordering::Greater
+                            } else {
+                                core::cmp::Ordering::Less
+                            }
+                        });
+
+                        moves.dedup_by(|[aa, ab, _], [ba, bb, kind]| {
+                            if aa == bb && ab == ba {
+                                *kind = 1;
+                                true
+                            } else {
+                                false
+                            }
+                        });
+
+                        for [dst, src, kind] in moves {
+                            if kind == 0 {
+                                self.emit(instrs::cp(dst, src));
+                            } else {
+                                self.emit(instrs::swa(dst, src));
+                            }
+                        }
+                        is_next_block = res.backrefs[nid as usize] as usize == i + 1;
+                    }
+                    Kind::Return { .. } => {
+                        let &[_, ret, ..] = node.inputs.as_slice() else { unreachable!() };
+                        match retl {
+                            Some(PLoc::Reg(r, _)) if sig.ret.loc(tys) == Loc::Reg => {
+                                alloc_buf.push(atr(ret));
+                                self.emit(instrs::cp(r, atr(ret)));
+                            }
+                            Some(PLoc::Ref(..)) => alloc_buf.extend([atr(ret), atr(MEM)]),
+                            Some(_) => alloc_buf.push(atr(ret)),
+                            None => {}
+                        }
+                    }
+                    Kind::Die => {}
+                    Kind::CInt { .. } => alloc_buf.push(atr(nid)),
+                    Kind::UnOp { .. } => alloc_buf.extend([atr(nid), atr(node.inputs[1])]),
+                    Kind::BinOp { op } => {
+                        let &[.., lhs, rhs] = node.inputs.as_slice() else { unreachable!() };
+
+                        if let Kind::CInt { .. } = nodes[rhs].kind
+                            && nodes.is_locked(rhs)
+                            && op.imm_binop(node.ty).is_some()
+                        {
+                            alloc_buf.extend([atr(nid), atr(lhs)]);
+                        } else {
+                            alloc_buf.extend([atr(nid), atr(lhs), atr(rhs)]);
+                        }
+                    }
+                    Kind::Call { args, .. } => {
+                        let (ret, mut parama) = tys.parama(node.ty);
+                        if ret.is_some() {
+                            alloc_buf.push(atr(nid));
+                        }
+                        let mut args = args.args();
+                        let mut allocs = node.inputs[1..].iter();
+                        while let Some(arg) = args.next(tys) {
+                            let Arg::Value(ty) = arg else { continue };
+                            let Some(loc) = parama.next(ty, tys) else { continue };
+
+                            let arg = *allocs.next().unwrap();
+                            alloc_buf.push(atr(arg));
+                            match loc {
+                                PLoc::Reg(..) if ty.loc(tys) == Loc::Stack => {}
+                                PLoc::WideReg(..) => alloc_buf.push(0),
+                                PLoc::Reg(r, ..) | PLoc::Ref(r, ..) => {
+                                    self.emit(instrs::cp(r, atr(arg)))
+                                }
+                            };
+                        }
+
+                        if node.ty.loc(tys) == Loc::Stack {
+                            alloc_buf.push(atr(*node.inputs.last().unwrap()));
+                        }
+
+                        if let Some(PLoc::Ref(r, ..)) = ret {
+                            self.emit(instrs::cp(r, *alloc_buf.last().unwrap()))
+                        }
+                    }
+                    Kind::Stck | Kind::Global { .. } => alloc_buf.push(atr(nid)),
+                    Kind::Load => {
+                        let (region, _) = nodes.strip_offset(node.inputs[1], node.ty, tys);
+                        if node.ty.loc(tys) != Loc::Stack {
+                            alloc_buf.push(atr(nid));
+                            match nodes[region].kind {
+                                Kind::Stck => {}
+                                _ => alloc_buf.push(atr(region)),
+                            }
+                        }
+                    }
+                    Kind::Stre if node.inputs[1] == VOID => {}
+                    Kind::Stre => {
+                        let (region, _) = nodes.strip_offset(node.inputs[2], node.ty, tys);
+                        match nodes[region].kind {
+                            Kind::Stck if node.ty.loc(tys) == Loc::Reg => {
+                                alloc_buf.push(atr(node.inputs[1]))
+                            }
+                            _ => alloc_buf.extend([atr(region), atr(node.inputs[1])]),
+                        }
+                    }
+                    Kind::Mem => {
+                        self.emit(instrs::cp(atr(MEM), reg::RET));
+                        continue;
+                    }
+                    Kind::Arg => {
+                        continue;
+                    }
+                    _ => {}
+                }
+
+                self.emit_instr(super::InstrCtx {
+                    nid,
+                    sig,
+                    is_next_block,
+                    is_last_block: i == res.blocks.len() - 1,
+                    retl,
+                    allocs: &alloc_buf,
+                    nodes,
+                    tys,
+                    files,
+                });
+
+                if let Kind::Call { .. } = node.kind {
+                    let (ret, ..) = tys.parama(node.ty);
+
+                    match ret {
+                        Some(PLoc::WideReg(..)) => {}
+                        Some(PLoc::Reg(..)) if node.ty.loc(tys) == Loc::Stack => {}
+                        Some(PLoc::Reg(r, ..)) => self.emit_cp(atr(nid), r),
+                        None | Some(PLoc::Ref(..)) => {}
+                    }
+                }
+            }
+        }
+
+        self.ralloc = res;
+
+        let bundle_count = self.ralloc.bundles.len() + (reg_offset as usize);
+        (
+            if tail {
+                assert!(bundle_count < reg::STACK_PTR as usize, "TODO: spill memory");
+                self.ralloc.bundles.len()
+            } else {
+                bundle_count.saturating_sub((reg::RET_ADDR - 1) as _)
+            },
+            tail,
+        )
+    }
+
+    fn emit_cp(&mut self, dst: Reg, src: Reg) {
+        if dst != 0 {
+            self.emit(instrs::cp(dst, src));
+        }
+    }
+}
+
+struct Function<'a> {
+    sig: Sig,
+    tail: bool,
+    nodes: &'a Nodes,
+    tys: &'a Types,
+    func: &'a mut Res,
+}
+
+impl core::fmt::Debug for Function<'_> {
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+        for block in &self.func.blocks {
+            writeln!(f, "{:?}", self.nodes[block.entry].kind)?;
+            for &instr in &self.func.instrs[block.range()] {
+                writeln!(f, "{:?}", self.nodes[instr].kind)?;
+            }
+        }
+
+        Ok(())
+    }
+}
+
+impl<'a> Function<'a> {
+    fn build(nodes: &'a Nodes, tys: &'a Types, func: &'a mut Res, sig: Sig) -> bool {
+        func.blocks.clear();
+        func.instrs.clear();
+        func.backrefs.resize(nodes.values.len(), u16::MAX);
+        func.visited.clear(nodes.values.len());
+        let mut s = Self { tail: true, nodes, tys, sig, func };
+        s.emit_node(VOID);
+        s.tail
+    }
+
+    fn add_block(&mut self, entry: Nid) {
+        self.func.blocks.push(Block {
+            start: self.func.instrs.len() as _,
+            end: self.func.instrs.len() as _,
+            entry,
+        });
+        self.func.backrefs[entry as usize] = self.func.blocks.len() as u16 - 1;
+    }
+
+    fn close_block(&mut self, exit: Nid) {
+        if !matches!(self.nodes[exit].kind, Kind::Loop | Kind::Region) {
+            self.add_instr(exit);
+        } else {
+            self.func.instrs.push(exit);
+        }
+        let prev = self.func.blocks.last_mut().unwrap();
+        prev.end = self.func.instrs.len() as _;
+    }
+
+    fn add_instr(&mut self, nid: Nid) {
+        debug_assert_ne!(self.nodes[nid].kind, Kind::Loop);
+        self.func.backrefs[nid as usize] = self.func.instrs.len() as u16;
+        self.func.instrs.push(nid);
+    }
+
+    fn emit_node(&mut self, nid: Nid) {
+        if matches!(self.nodes[nid].kind, Kind::Region | Kind::Loop) {
+            match (self.nodes[nid].kind, self.func.visited.set(nid)) {
+                (Kind::Loop, false) | (Kind::Region, true) => {
+                    self.close_block(nid);
+                    return;
+                }
+                _ => {}
+            }
+        } else if !self.func.visited.set(nid) {
+            return;
+        }
+
+        if self.nodes.is_never_used(nid, self.tys) {
+            self.nodes.lock(nid);
+            return;
+        }
+
+        let mut node = self.nodes[nid].clone();
+        match node.kind {
+            Kind::Start => {
+                debug_assert_matches!(self.nodes[node.outputs[0]].kind, Kind::Entry);
+                self.add_block(VOID);
+                self.emit_node(node.outputs[0])
+            }
+            Kind::If => {
+                let &[_, cnd] = node.inputs.as_slice() else { unreachable!() };
+                let &[mut then, mut else_] = node.outputs.as_slice() else { unreachable!() };
+
+                if let Some((_, swapped)) = self.nodes.cond_op(cnd) {
+                    if swapped {
+                        mem::swap(&mut then, &mut else_);
+                    }
+                } else {
+                    mem::swap(&mut then, &mut else_);
+                }
+
+                self.close_block(nid);
+                self.emit_node(then);
+                self.emit_node(else_);
+            }
+            Kind::Region | Kind::Loop => {
+                self.close_block(nid);
+                self.add_block(nid);
+                self.nodes.reschedule_block(nid, &mut node.outputs);
+                for o in node.outputs.into_iter().rev() {
+                    self.emit_node(o);
+                }
+            }
+            Kind::Return { .. } | Kind::Die => {
+                self.close_block(nid);
+                self.emit_node(node.outputs[0]);
+            }
+            Kind::Entry => {
+                let (ret, mut parama) = self.tys.parama(self.sig.ret);
+
+                if let Some(PLoc::Ref(..)) = ret {
+                    self.add_instr(MEM);
+                }
+
+                let mut typs = self.sig.args.args();
+                #[expect(clippy::unnecessary_to_owned)]
+                let mut args = self.nodes[VOID].outputs[ARG_START..].to_owned().into_iter();
+                while let Some(ty) = typs.next_value(self.tys) {
+                    let arg = args.next().unwrap();
+                    debug_assert_eq!(self.nodes[arg].kind, Kind::Arg);
+                    match parama.next(ty, self.tys) {
+                        None => {}
+                        Some(_) => self.add_instr(arg),
+                    }
+                }
+
+                self.nodes.reschedule_block(nid, &mut node.outputs);
+                for o in node.outputs.into_iter().rev() {
+                    self.emit_node(o);
+                }
+            }
+            Kind::Then | Kind::Else => {
+                self.add_block(nid);
+                self.nodes.reschedule_block(nid, &mut node.outputs);
+                for o in node.outputs.into_iter().rev() {
+                    self.emit_node(o);
+                }
+            }
+            Kind::Call { func, .. } => {
+                self.tail &= func == ty::Func::ECA;
+
+                self.add_instr(nid);
+
+                self.nodes.reschedule_block(nid, &mut node.outputs);
+                for o in node.outputs.into_iter().rev() {
+                    if self.nodes[o].inputs[0] == nid
+                        || (matches!(self.nodes[o].kind, Kind::Loop | Kind::Region)
+                            && self.nodes[o].inputs[1] == nid)
+                    {
+                        self.emit_node(o);
+                    }
+                }
+            }
+            Kind::CInt { value: 0 } if self.nodes.is_hard_zero(nid) => {}
+            Kind::CInt { .. }
+            | Kind::BinOp { .. }
+            | Kind::UnOp { .. }
+            | Kind::Global { .. }
+            | Kind::Load { .. }
+            | Kind::Stre
+            | Kind::Stck => self.add_instr(nid),
+            Kind::End | Kind::Phi | Kind::Arg | Kind::Mem | Kind::Loops | Kind::Join => {}
+            Kind::Assert { .. } => unreachable!(),
+        }
+    }
+}
+
+impl Nodes {
+    fn vreg_count(&self) -> usize {
+        self.values.len()
+    }
+
+    fn use_block_of(&self, inst: Nid, uinst: Nid) -> Nid {
+        let mut block = self.use_block(inst, uinst);
+        while !self[block].kind.starts_basic_block() {
+            block = self.idom(block);
+        }
+        block
+    }
+
+    fn phi_inputs_of(&self, nid: Nid) -> impl Iterator<Item = [Nid; 3]> + use<'_> {
+        match self[nid].kind {
+            Kind::Region | Kind::Loop => Some({
+                self[nid]
+                    .outputs
+                    .as_slice()
+                    .iter()
+                    .filter(|&&n| self[n].is_data_phi())
+                    .map(|&n| [n, self[n].inputs[1], self[n].inputs[2]])
+            })
+            .into_iter()
+            .flatten(),
+            _ => None.into_iter().flatten(),
+        }
+    }
+
+    fn idom_of(&self, mut nid: Nid) -> Nid {
+        while !self[nid].kind.starts_basic_block() {
+            nid = self.idom(nid);
+        }
+        nid
+    }
+
+    fn uses_of(&self, nid: Nid) -> impl Iterator<Item = (Nid, Nid)> + use<'_> {
+        if self[nid].kind.is_cfg() && !matches!(self[nid].kind, Kind::Call { .. }) {
+            return None.into_iter().flatten();
+        }
+
+        Some(
+            self[nid]
+                .outputs
+                .iter()
+                .filter(move |&&n| self.is_data_dep(nid, n))
+                .map(move |n| self.this_or_delegates(nid, n))
+                .flat_map(|(p, ls)| ls.iter().map(move |l| (p, l)))
+                .filter(|&(o, &n)| self.is_data_dep(o, n))
+                .map(|(p, &n)| (self.use_block_of(p, n), n))
+                .inspect(|&(_, n)| debug_assert_eq!(self[n].lock_rc.get(), 0)),
+        )
+        .into_iter()
+        .flatten()
+    }
+}
+
+struct Regalloc<'a> {
+    nodes: &'a Nodes,
+    res: &'a mut Res,
+}
+
+impl<'a> Regalloc<'a> {
+    fn instr_of(&self, nid: Nid) -> Option<Nid> {
+        if self.nodes[nid].kind == Kind::Phi || self.nodes.is_locked(nid) {
+            return None;
+        }
+        debug_assert_ne!(self.res.backrefs[nid as usize], Nid::MAX, "{:?}", self.nodes[nid]);
+        Some(self.res.backrefs[nid as usize])
+    }
+
+    fn block_of(&self, nid: Nid) -> Nid {
+        debug_assert!(self.nodes[nid].kind.starts_basic_block());
+        self.res.backrefs[nid as usize]
+    }
+
+    fn run(ctx: &'a Nodes, res: &'a mut Res) {
+        Self { nodes: ctx, res }.run_low();
+    }
+
+    fn run_low(&mut self) {
+        self.res.bundles.clear();
+        self.res.node_to_reg.clear();
+        #[cfg(debug_assertions)]
+        self.res.marked.clear();
+        self.res.node_to_reg.resize(self.nodes.vreg_count(), 0);
+
+        debug_assert!(self.res.dfs_buf.is_empty());
+
+        let mut bundle = Bundle::new(self.res.instrs.len());
+        self.res.visited.clear(self.nodes.values.len());
+
+        for i in (0..self.res.blocks.len()).rev() {
+            for [a, rest @ ..] in self.nodes.phi_inputs_of(self.res.blocks[i].entry) {
+                if self.res.visited.set(a) {
+                    self.append_bundle(a, &mut bundle, None);
+                }
+
+                for r in rest {
+                    if !self.res.visited.set(r) {
+                        continue;
+                    }
+
+                    self.append_bundle(
+                        r,
+                        &mut bundle,
+                        Some(self.res.node_to_reg[a as usize] as usize - 1),
+                    );
+                }
+            }
+        }
+
+        let instrs = mem::take(&mut self.res.instrs);
+        for &inst in &instrs {
+            if self.nodes[inst].has_no_value() || self.res.visited.get(inst) || inst == 0 {
+                continue;
+            }
+            self.append_bundle(inst, &mut bundle, None);
+        }
+        self.res.instrs = instrs;
+    }
+
+    fn collect_bundle(&mut self, inst: Nid, into: &mut Bundle) {
+        let dom = self.nodes.idom_of(inst);
+        self.res.dfs_seem.clear(self.nodes.values.len());
+        for (cursor, uinst) in self.nodes.uses_of(inst) {
+            if !self.res.dfs_seem.set(uinst) {
+                continue;
+            }
+            #[cfg(debug_assertions)]
+            debug_assert!(self.res.marked.insert((inst, uinst)));
+
+            self.reverse_cfg_dfs(cursor, dom, |s, n, b| {
+                let mut range = b.range();
+                debug_assert!(range.start < range.end);
+                range.start = range.start.max(s.instr_of(inst).map_or(0, |n| n + 1) as usize);
+                debug_assert!(range.start < range.end, "{:?}", range);
+                let new = range.end.min(
+                    s.instr_of(uinst)
+                        .filter(|_| {
+                            n == cursor
+                                && self.nodes.loop_depth(dom) == self.nodes.loop_depth(cursor)
+                        })
+                        .map_or(Nid::MAX, |n| n + 1) as usize,
+                );
+
+                range.end = new;
+                debug_assert!(range.start < range.end, "{:?} {inst} {uinst}", range);
+
+                into.add(range);
+            });
+        }
+    }
+
+    fn append_bundle(&mut self, inst: Nid, tmp: &mut Bundle, prefered: Option<usize>) {
+        self.collect_bundle(inst, tmp);
+
+        if tmp.is_empty() {
+            self.res.node_to_reg[inst as usize] = u8::MAX;
+            return;
+        }
+
+        if let Some(prefered) = prefered
+            && !self.res.bundles[prefered].overlaps(tmp)
+        {
+            self.res.bundles[prefered].merge(tmp);
+            tmp.clear();
+            self.res.node_to_reg[inst as usize] = prefered as Reg + 1;
+            return;
+        }
+
+        match self.res.bundles.iter_mut().enumerate().find(|(_, b)| !b.overlaps(tmp)) {
+            Some((i, other)) => {
+                other.merge(tmp);
+                tmp.clear();
+                self.res.node_to_reg[inst as usize] = i as Reg + 1;
+            }
+            None => {
+                self.res.bundles.push(tmp.take());
+                self.res.node_to_reg[inst as usize] = self.res.bundles.len() as Reg;
+            }
+        }
+    }
+
+    fn reverse_cfg_dfs(
+        &mut self,
+        from: Nid,
+        until: Nid,
+        mut each: impl FnMut(&mut Self, Nid, Block),
+    ) {
+        debug_assert!(self.res.dfs_buf.is_empty());
+        self.res.dfs_buf.push(from);
+
+        debug_assert!(self.nodes.dominates(until, from));
+
+        while let Some(nid) = self.res.dfs_buf.pop() {
+            debug_assert!(self.nodes.dominates(until, nid), "{until} {:?}", self.nodes[until]);
+            each(self, nid, self.res.blocks[self.block_of(nid) as usize]);
+            if nid == until {
+                continue;
+            }
+            match self.nodes[nid].kind {
+                Kind::Then | Kind::Else | Kind::Region | Kind::Loop => {
+                    for &n in self.nodes[nid].inputs.iter() {
+                        if self.nodes[n].kind == Kind::Loops {
+                            continue;
+                        }
+                        let d = self.nodes.idom_of(n);
+                        if self.res.dfs_seem.set(d) {
+                            self.res.dfs_buf.push(d);
+                        }
+                    }
+                }
+                Kind::Start => {}
+                _ => unreachable!(),
+            }
+        }
+    }
+}
+
+#[derive(Default)]
+pub(super) struct Res {
+    blocks: Vec<Block>,
+    instrs: Vec<Nid>,
+    backrefs: Vec<u16>,
+
+    bundles: Vec<Bundle>,
+    node_to_reg: Vec<Reg>,
+
+    visited: BitSet,
+    dfs_buf: Vec<Nid>,
+    dfs_seem: BitSet,
+    #[cfg(debug_assertions)]
+    marked: hashbrown::HashSet<(Nid, Nid), crate::FnvBuildHasher>,
+}
+
+struct Bundle {
+    taken: Vec<bool>,
+}
+
+impl Bundle {
+    fn new(size: usize) -> Self {
+        Self { taken: vec![false; size] }
+    }
+
+    fn add(&mut self, range: Range<usize>) {
+        self.taken[range].fill(true);
+    }
+
+    fn overlaps(&self, other: &Self) -> bool {
+        self.taken.iter().zip(other.taken.iter()).any(|(a, b)| a & b)
+    }
+
+    fn merge(&mut self, other: &Self) {
+        debug_assert!(!self.overlaps(other));
+        self.taken.iter_mut().zip(other.taken.iter()).for_each(|(a, b)| *a |= *b);
+    }
+
+    fn clear(&mut self) {
+        self.taken.fill(false);
+    }
+
+    fn is_empty(&self) -> bool {
+        !self.taken.contains(&true)
+    }
+
+    fn take(&mut self) -> Self {
+        mem::replace(self, Self::new(self.taken.len()))
+    }
+}
+
+#[derive(Clone, Copy)]
+struct Block {
+    start: u16,
+    end: u16,
+    entry: Nid,
+}
+
+impl Block {
+    pub fn range(&self) -> Range<usize> {
+        self.start as usize..self.end as usize
+    }
+}

lang/src/utils.rs

@@ -0,0 +1,649 @@
+#![expect(dead_code)]
+use {
+    alloc::alloc,
+    core::{
+        alloc::Layout,
+        fmt::Debug,
+        hint::unreachable_unchecked,
+        marker::PhantomData,
+        mem::MaybeUninit,
+        ops::{Deref, DerefMut, Not},
+        ptr::Unique,
+    },
+};
+
+fn decide(b: bool, name: &'static str) -> Result<(), &'static str> {
+    b.then_some(()).ok_or(name)
+}
+
+pub fn is_snake_case(str: &str) -> Result<(), &'static str> {
+    decide(str.bytes().all(|c| matches!(c, b'a'..=b'z' | b'0'..=b'9' | b'_')), "snake_case")
+}
+
+pub fn is_pascal_case(str: &str) -> Result<(), &'static str> {
+    decide(
+        str.as_bytes()[0].is_ascii_uppercase() && str.bytes().all(|c| c.is_ascii_alphanumeric()),
+        "PascalCase",
+    )
+}
+
+pub fn is_screaming_case(str: &str) -> Result<(), &'static str> {
+    decide(str.bytes().all(|c| matches!(c, b'A'..=b'Z' | b'0'..=b'9' | b'_')), "SCREAMING_CASE")
+}
+
+type Nid = u16;
+
+pub union BitSet {
+    inline: usize,
+    alloced: Unique<AllocedBitSet>,
+}
+
+impl Debug for BitSet {
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+        f.debug_list().entries(self.iter()).finish()
+    }
+}
+
+impl Clone for BitSet {
+    fn clone(&self) -> Self {
+        if self.is_inline() {
+            Self { inline: unsafe { self.inline } }
+        } else {
+            let (data, _) = self.data_and_len();
+            let (layout, _) = Self::layout(data.len());
+            unsafe {
+                let ptr = alloc::alloc(layout);
+                ptr.copy_from_nonoverlapping(self.alloced.as_ptr() as _, layout.size());
+                Self { alloced: Unique::new_unchecked(ptr as _) }
+            }
+        }
+    }
+}
+
+impl Drop for BitSet {
+    fn drop(&mut self) {
+        if !self.is_inline() {
+            unsafe {
+                let cap = self.alloced.as_ref().cap;
+                alloc::dealloc(self.alloced.as_ptr() as _, Self::layout(cap).0);
+            }
+        }
+    }
+}
+
+impl Default for BitSet {
+    fn default() -> Self {
+        Self { inline: Self::FLAG }
+    }
+}
+
+impl BitSet {
+    const FLAG: usize = 1 << (Self::UNIT - 1);
+    const INLINE_ELEMS: usize = Self::UNIT - 1;
+    const UNIT: usize = core::mem::size_of::<usize>() * 8;
+
+    pub fn with_capacity(len: usize) -> Self {
+        let mut s = Self::default();
+        s.reserve(len);
+        s
+    }
+
+    fn is_inline(&self) -> bool {
+        unsafe { self.inline & Self::FLAG != 0 }
+    }
+
+    fn data_and_len(&self) -> (&[usize], usize) {
+        unsafe {
+            if self.is_inline() {
+                (core::slice::from_ref(&self.inline), Self::INLINE_ELEMS)
+            } else {
+                let small_vec = self.alloced.as_ref();
+                (
+                    core::slice::from_raw_parts(
+                        &small_vec.data as *const _ as *const usize,
+                        small_vec.cap,
+                    ),
+                    small_vec.cap * core::mem::size_of::<usize>() * 8,
+                )
+            }
+        }
+    }
+
+    fn data_mut_and_len(&mut self) -> (&mut [usize], usize) {
+        unsafe {
+            if self.is_inline() {
+                (core::slice::from_mut(&mut self.inline), INLINE_ELEMS)
+            } else {
+                let small_vec = self.alloced.as_mut();
+                (
+                    core::slice::from_raw_parts_mut(
+                        &mut small_vec.data as *mut _ as *mut usize,
+                        small_vec.cap,
+                    ),
+                    small_vec.cap * Self::UNIT,
+                )
+            }
+        }
+    }
+
+    fn indexes(index: usize) -> (usize, usize) {
+        (index / Self::UNIT, index % Self::UNIT)
+    }
+
+    pub fn get(&self, index: Nid) -> bool {
+        let index = index as usize;
+        let (data, len) = self.data_and_len();
+        if index >= len {
+            return false;
+        }
+        let (elem, bit) = Self::indexes(index);
+        (unsafe { *data.get_unchecked(elem) }) & (1 << bit) != 0
+    }
+
+    pub fn set(&mut self, index: Nid) -> bool {
+        let index = index as usize;
+        let (mut data, len) = self.data_mut_and_len();
+        if core::intrinsics::unlikely(index >= len) {
+            self.grow((index + 1).next_power_of_two().max(4 * Self::UNIT));
+            (data, _) = self.data_mut_and_len();
+        }
+
+        let (elem, bit) = Self::indexes(index);
+        debug_assert!(elem < data.len(), "{} < {}", elem, data.len());
+        let elem = unsafe { data.get_unchecked_mut(elem) };
+        let prev = *elem;
+        *elem |= 1 << bit;
+        *elem != prev
+    }
+
+    fn grow(&mut self, size: usize) {
+        debug_assert!(size.is_power_of_two());
+        let slot_count = size / Self::UNIT;
+        let (layout, off) = Self::layout(slot_count);
+        let (ptr, prev_len) = unsafe {
+            if self.is_inline() {
+                let ptr = alloc::alloc(layout);
+                *ptr.add(off).cast::<usize>() = self.inline & !Self::FLAG;
+                (ptr, 1)
+            } else {
+                let prev_len = self.alloced.as_ref().cap;
+                let (prev_layout, _) = Self::layout(prev_len);
+                (alloc::realloc(self.alloced.as_ptr() as _, prev_layout, layout.size()), prev_len)
+            }
+        };
+        unsafe {
+            MaybeUninit::fill(
+                core::slice::from_raw_parts_mut(
+                    ptr.add(off).cast::<MaybeUninit<usize>>().add(prev_len),
+                    slot_count - prev_len,
+                ),
+                0,
+            );
+            *ptr.cast::<usize>() = slot_count;
+            core::ptr::write(self, Self { alloced: Unique::new_unchecked(ptr as _) });
+        }
+    }
+
+    fn layout(slot_count: usize) -> (core::alloc::Layout, usize) {
+        unsafe {
+            core::alloc::Layout::new::<AllocedBitSet>()
+                .extend(Layout::array::<usize>(slot_count).unwrap_unchecked())
+                .unwrap_unchecked()
+        }
+    }
+
+    pub fn iter(&self) -> BitSetIter {
+        if self.is_inline() {
+            BitSetIter { index: 0, current: unsafe { self.inline & !Self::FLAG }, remining: &[] }
+        } else {
+            let &[current, ref remining @ ..] = self.data_and_len().0 else {
+                unsafe { unreachable_unchecked() }
+            };
+            BitSetIter { index: 0, current, remining }
+        }
+    }
+
+    pub fn clear(&mut self, len: usize) {
+        self.reserve(len);
+        if self.is_inline() {
+            unsafe { self.inline &= Self::FLAG };
+        } else {
+            self.data_mut_and_len().0.fill(0);
+        }
+    }
+
+    pub fn units<'a>(&'a self, slot: &'a mut usize) -> &'a [usize] {
+        if self.is_inline() {
+            *slot = unsafe { self.inline } & !Self::FLAG;
+            core::slice::from_ref(slot)
+        } else {
+            self.data_and_len().0
+        }
+    }
+
+    pub fn reserve(&mut self, len: usize) {
+        if len > self.data_and_len().1 {
+            self.grow(len.next_power_of_two().max(4 * Self::UNIT));
+        }
+    }
+
+    pub fn units_mut(&mut self) -> Result<&mut [usize], &mut InlineBitSetView> {
+        if self.is_inline() {
+            Err(unsafe {
+                core::mem::transmute::<&mut usize, &mut InlineBitSetView>(&mut self.inline)
+            })
+        } else {
+            Ok(self.data_mut_and_len().0)
+        }
+    }
+}
+
+pub struct InlineBitSetView(usize);
+
+impl InlineBitSetView {
+    pub(crate) fn add_mask(&mut self, tmp: usize) {
+        debug_assert!(tmp & BitSet::FLAG == 0);
+        self.0 |= tmp;
+    }
+}
+
+pub struct BitSetIter<'a> {
+    index: usize,
+    current: usize,
+    remining: &'a [usize],
+}
+
+impl Iterator for BitSetIter<'_> {
+    type Item = usize;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        while self.current == 0 {
+            self.current = *self.remining.take_first()?;
+            self.index += 1;
+        }
+
+        let sub_idx = self.current.trailing_zeros() as usize;
+        self.current &= self.current - 1;
+        Some(self.index * BitSet::UNIT + sub_idx)
+    }
+}
+
+struct AllocedBitSet {
+    cap: usize,
+    data: [usize; 0],
+}
+
+#[cfg(test)]
+#[test]
+fn test_small_bit_set() {
+    use std::vec::Vec;
+
+    let mut sv = BitSet::default();
+
+    sv.set(10);
+    debug_assert!(sv.get(10));
+    sv.set(100);
+    debug_assert!(sv.get(100));
+    sv.set(10000);
+    debug_assert!(sv.get(10000));
+    debug_assert_eq!(sv.iter().collect::<Vec<_>>(), &[10, 100, 10000]);
+    sv.clear(10000);
+    debug_assert_eq!(sv.iter().collect::<Vec<_>>(), &[]);
+}
+
+pub union Vc {
+    inline: InlineVc,
+    alloced: AllocedVc,
+}
+
+impl Default for Vc {
+    fn default() -> Self {
+        Vc { inline: InlineVc { elems: MaybeUninit::uninit(), cap: Default::default() } }
+    }
+}
+
+impl Debug for Vc {
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+        self.as_slice().fmt(f)
+    }
+}
+
+impl FromIterator<Nid> for Vc {
+    fn from_iter<T: IntoIterator<Item = Nid>>(iter: T) -> Self {
+        let mut slf = Self::default();
+        for i in iter {
+            slf.push(i);
+        }
+        slf
+    }
+}
+
+const INLINE_ELEMS: usize = VC_SIZE / 2 - 1;
+const VC_SIZE: usize = 16;
+
+impl Vc {
+    fn is_inline(&self) -> bool {
+        unsafe { self.inline.cap <= INLINE_ELEMS as Nid }
+    }
+
+    fn layout(&self) -> Option<core::alloc::Layout> {
+        unsafe {
+            self.is_inline().not().then(|| {
+                core::alloc::Layout::array::<Nid>(self.alloced.cap as _).unwrap_unchecked()
+            })
+        }
+    }
+
+    pub fn len(&self) -> usize {
+        unsafe {
+            if self.is_inline() {
+                self.inline.cap as _
+            } else {
+                self.alloced.len as _
+            }
+        }
+    }
+
+    fn len_mut(&mut self) -> &mut Nid {
+        unsafe {
+            if self.is_inline() {
+                &mut self.inline.cap
+            } else {
+                &mut self.alloced.len
+            }
+        }
+    }
+
+    fn as_ptr(&self) -> *const Nid {
+        unsafe {
+            match self.is_inline() {
+                true => self.inline.elems.as_ptr().cast(),
+                false => self.alloced.base.as_ptr(),
+            }
+        }
+    }
+
+    fn as_mut_ptr(&mut self) -> *mut Nid {
+        unsafe {
+            match self.is_inline() {
+                true => self.inline.elems.as_mut_ptr().cast(),
+                false => self.alloced.base.as_ptr(),
+            }
+        }
+    }
+
+    pub fn as_slice(&self) -> &[Nid] {
+        unsafe { core::slice::from_raw_parts(self.as_ptr(), self.len()) }
+    }
+
+    fn as_slice_mut(&mut self) -> &mut [Nid] {
+        unsafe { core::slice::from_raw_parts_mut(self.as_mut_ptr(), self.len()) }
+    }
+
+    pub fn push(&mut self, value: Nid) {
+        if let Some(layout) = self.layout()
+            && unsafe { self.alloced.len == self.alloced.cap }
+        {
+            unsafe {
+                self.alloced.cap *= 2;
+                self.alloced.base = Unique::new_unchecked(
+                    alloc::realloc(
+                        self.alloced.base.as_ptr().cast(),
+                        layout,
+                        self.alloced.cap as usize * core::mem::size_of::<Nid>(),
+                    )
+                    .cast(),
+                );
+            }
+        } else if self.len() == INLINE_ELEMS {
+            unsafe {
+                let mut allcd =
+                    Self::alloc((self.inline.cap + 1).next_power_of_two() as _, self.len());
+                core::ptr::copy_nonoverlapping(self.as_ptr(), allcd.as_mut_ptr(), self.len());
+                *self = allcd;
+            }
+        }
+
+        unsafe {
+            *self.len_mut() += 1;
+            self.as_mut_ptr().add(self.len() - 1).write(value);
+        }
+    }
+
+    unsafe fn alloc(cap: usize, len: usize) -> Self {
+        debug_assert!(cap > INLINE_ELEMS);
+        let layout = unsafe { core::alloc::Layout::array::<Nid>(cap).unwrap_unchecked() };
+        let alloc = unsafe { alloc::alloc(layout) };
+        unsafe {
+            Vc {
+                alloced: AllocedVc {
+                    base: Unique::new_unchecked(alloc.cast()),
+                    len: len as _,
+                    cap: cap as _,
+                },
+            }
+        }
+    }
+
+    pub fn swap_remove(&mut self, index: usize) {
+        let len = self.len() - 1;
+        self.as_slice_mut().swap(index, len);
+        *self.len_mut() -= 1;
+    }
+
+    pub fn remove(&mut self, index: usize) {
+        self.as_slice_mut().copy_within(index + 1.., index);
+        *self.len_mut() -= 1;
+    }
+}
+
+impl Drop for Vc {
+    fn drop(&mut self) {
+        if let Some(layout) = self.layout() {
+            unsafe {
+                alloc::dealloc(self.alloced.base.as_ptr().cast(), layout);
+            }
+        }
+    }
+}
+
+impl Clone for Vc {
+    fn clone(&self) -> Self {
+        self.as_slice().into()
+    }
+}
+
+impl IntoIterator for Vc {
+    type IntoIter = VcIntoIter;
+    type Item = Nid;
+
+    fn into_iter(self) -> Self::IntoIter {
+        VcIntoIter { start: 0, end: self.len(), vc: self }
+    }
+}
+
+pub struct VcIntoIter {
+    start: usize,
+    end: usize,
+    vc: Vc,
+}
+
+impl Iterator for VcIntoIter {
+    type Item = Nid;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        if self.start == self.end {
+            return None;
+        }
+
+        let ret = unsafe { core::ptr::read(self.vc.as_slice().get_unchecked(self.start)) };
+        self.start += 1;
+        Some(ret)
+    }
+
+    fn size_hint(&self) -> (usize, Option<usize>) {
+        let len = self.end - self.start;
+        (len, Some(len))
+    }
+}
+
+impl DoubleEndedIterator for VcIntoIter {
+    fn next_back(&mut self) -> Option<Self::Item> {
+        if self.start == self.end {
+            return None;
+        }
+
+        self.end -= 1;
+        Some(unsafe { core::ptr::read(self.vc.as_slice().get_unchecked(self.end)) })
+    }
+}
+
+impl ExactSizeIterator for VcIntoIter {}
+
+impl<const SIZE: usize> From<[Nid; SIZE]> for Vc {
+    fn from(value: [Nid; SIZE]) -> Self {
+        value.as_slice().into()
+    }
+}
+
+impl<'a> From<&'a [Nid]> for Vc {
+    fn from(value: &'a [Nid]) -> Self {
+        if value.len() <= INLINE_ELEMS {
+            let mut dflt = Self::default();
+            unsafe {
+                core::ptr::copy_nonoverlapping(value.as_ptr(), dflt.as_mut_ptr(), value.len())
+            };
+            dflt.inline.cap = value.len() as _;
+            dflt
+        } else {
+            let mut allcd = unsafe { Self::alloc(value.len(), value.len()) };
+            unsafe {
+                core::ptr::copy_nonoverlapping(value.as_ptr(), allcd.as_mut_ptr(), value.len())
+            };
+            allcd
+        }
+    }
+}
+
+impl Deref for Vc {
+    type Target = [Nid];
+
+    fn deref(&self) -> &Self::Target {
+        self.as_slice()
+    }
+}
+
+impl DerefMut for Vc {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        self.as_slice_mut()
+    }
+}
+
+#[derive(Clone, Copy)]
+#[repr(C)]
+struct InlineVc {
+    cap: Nid,
+    elems: MaybeUninit<[Nid; INLINE_ELEMS]>,
+}
+
+#[derive(Clone, Copy)]
+#[repr(C)]
+struct AllocedVc {
+    cap: Nid,
+    len: Nid,
+    base: Unique<Nid>,
+}
+
+pub trait Ent: Copy {
+    fn new(index: usize) -> Self;
+    fn index(self) -> usize;
+}
+
+pub struct EntVec<K: Ent, T> {
+    data: ::alloc::vec::Vec<T>,
+    k: PhantomData<fn(K)>,
+}
+
+impl<K: Ent, T> Default for EntVec<K, T> {
+    fn default() -> Self {
+        Self { data: Default::default(), k: PhantomData }
+    }
+}
+
+impl<K: Ent, T> EntVec<K, T> {
+    pub fn clear(&mut self) {
+        self.data.clear();
+    }
+
+    pub fn is_empty(&self) -> bool {
+        self.data.is_empty()
+    }
+
+    pub fn len(&self) -> usize {
+        self.data.len()
+    }
+
+    pub fn push(&mut self, value: T) -> K {
+        let k = K::new(self.data.len());
+        self.data.push(value);
+        k
+    }
+
+    pub fn next(&self, index: K) -> Option<&T> {
+        self.data.get(index.index() + 1)
+    }
+
+    pub fn shadow(&mut self, len: usize)
+    where
+        T: Default,
+    {
+        if self.data.len() < len {
+            self.data.resize_with(len, Default::default);
+        }
+    }
+
+    pub fn iter(&self) -> core::slice::Iter<T> {
+        self.data.iter()
+    }
+}
+
+impl<K: Ent, T> core::ops::Index<K> for EntVec<K, T> {
+    type Output = T;
+
+    fn index(&self, index: K) -> &Self::Output {
+        &self.data[index.index()]
+    }
+}
+
+impl<K: Ent, T> core::ops::IndexMut<K> for EntVec<K, T> {
+    fn index_mut(&mut self, index: K) -> &mut Self::Output {
+        &mut self.data[index.index()]
+    }
+}
+
+macro_rules! decl_ent {
+    ($(
+        $vis:vis struct $name:ident($index:ty);
+    )*) => {$(
+        #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, Debug)]
+        $vis struct $name($index);
+
+        impl crate::utils::Ent for $name {
+            fn new(index: usize) -> Self {
+                Self(index as $index)
+            }
+
+            fn index(self) -> usize {
+                self.0 as _
+            }
+        }
+
+
+        impl core::fmt::Display for $name {
+            fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+                write!(f, concat!(stringify!($name), "{}"), self.0)
+            }
+        }
+    )*};
+}
+pub(crate) use decl_ent;

lang/tests/son_tests_advanced_floating_point_arithmetic.txt

@@ -0,0 +1,48 @@
+main:
+    ADDI64 r254, r254, -24d
+    ST r31, r254, 0a, 24h
+    LI32 r32, 1148846080w
+    CP r2, r32
+    JAL r31, r0, :sin
+    CP r33, r1
+    FMUL32 r32, r33, r32
+    FTI32 r32, r32, 1b
+    CP r1, r32
+    LD r31, r254, 0a, 24h
+    ADDI64 r254, r254, 24d
+    JALA r0, r31, 0a
+sin:
+    CP r13, r2
+    LI32 r14, 1124073472w
+    LI32 r15, 1078530011w
+    FMUL32 r14, r13, r14
+    FDIV32 r14, r14, r15
+    FTI32 r14, r14, 1b
+    ANDI r15, r14, 255d
+    ITF64 r16, r14
+    MULI64 r15, r15, 4d
+    LRA r17, r0, :sin_table
+    LI32 r18, 1086918619w
+    FC64T32 r16, r16, 1b
+    ADDI64 r14, r14, 64d
+    ADD64 r15, r17, r15
+    LI32 r19, 1132462080w
+    FMUL32 r16, r16, r18
+    ANDI r14, r14, 255d
+    LI32 r18, 1056964608w
+    LD r15, r15, 0a, 4h
+    FDIV32 r16, r16, r19
+    MULI64 r14, r14, 4d
+    FMUL32 r18, r15, r18
+    FSUB32 r13, r13, r16
+    ADD64 r14, r17, r14
+    FMUL32 r16, r13, r18
+    LD r14, r14, 0a, 4h
+    FSUB32 r14, r14, r16
+    FMUL32 r13, r14, r13
+    FADD32 r13, r15, r13
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 1315
+ret: 826
+status: Ok(())

lang/tests/son_tests_aliasing_overoptimization.txt

@@ -0,0 +1,6 @@
+main:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 22
+ret: 0
+status: Ok(())

lang/tests/son_tests_arithmetic.txt

@@ -0,0 +1,6 @@
+main:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 22
+ret: 0
+status: Ok(())

lang/tests/son_tests_arrays.txt

@@ -0,0 +1,32 @@
+main:
+    ADDI64 r254, r254, -56d
+    ST r31, r254, 24a, 32h
+    LI64 r32, 1d
+    ADDI64 r33, r254, 0d
+    ST r32, r254, 0a, 8h
+    LI64 r34, 2d
+    ST r34, r254, 8a, 8h
+    LI64 r34, 4d
+    ST r34, r254, 16a, 8h
+    CP r2, r33
+    JAL r31, r0, :pass
+    CP r33, r1
+    ADD64 r32, r33, r32
+    CP r1, r32
+    LD r31, r254, 24a, 32h
+    ADDI64 r254, r254, 56d
+    JALA r0, r31, 0a
+pass:
+    CP r13, r2
+    LD r14, r13, 8a, 8h
+    MULI64 r15, r14, 8d
+    LD r16, r13, 0a, 8h
+    ADD64 r13, r15, r13
+    ADD64 r14, r14, r16
+    LD r13, r13, 0a, 8h
+    ADD64 r13, r13, r14
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 246
+ret: 8
+status: Ok(())

lang/tests/son_tests_big_array_crash.txt

@@ -0,0 +1,8 @@
+main:
+    LRA r13, r0, :sin_table
+    LD r13, r13, 80a, 8h
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 770
+ret: 1736
+status: Ok(())

lang/tests/son_tests_branch_assignments.txt

@@ -0,0 +1,14 @@
+main:
+    CP r14, r2
+    LI64 r13, 1d
+    JNE r14, r13, :0
+    JMP :1
+ 0: JNE r14, r0, :2
+    LI64 r13, 2d
+    JMP :1
+ 2: LI64 r13, 3d
+ 1: CP r1, r13
+    JALA r0, r31, 0a
+code size: 75
+ret: 2
+status: Ok(())

lang/tests/son_tests_c_strings.txt

@@ -0,0 +1,32 @@
+main:
+    ADDI64 r254, r254, -24d
+    ST r31, r254, 0a, 24h
+    LRA r32, r0, :"abඞ\n\r\t56789\0"
+    CP r2, r32
+    JAL r31, r0, :str_len
+    CP r32, r1
+    LRA r33, r0, :"fff\0"
+    CP r2, r33
+    JAL r31, r0, :str_len
+    CP r33, r1
+    ADD64 r32, r33, r32
+    CP r1, r32
+    LD r31, r254, 0a, 24h
+    ADDI64 r254, r254, 24d
+    JALA r0, r31, 0a
+str_len:
+    CP r15, r2
+    CP r14, r0
+    CP r13, r14
+ 2: LD r16, r15, 0a, 1h
+    ANDI r16, r16, 255d
+    JNE r16, r14, :0
+    CP r1, r13
+    JMP :1
+ 0: ADDI64 r15, r15, 1d
+    ADDI64 r13, r13, 1d
+    JMP :2
+ 1: JALA r0, r31, 0a
+code size: 216
+ret: 16
+status: Ok(())

lang/tests/son_tests_comments.txt

@@ -0,0 +1,13 @@
+foo:
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -8d
+    ST r31, r254, 0a, 8h
+    JAL r31, r0, :foo
+    CP r1, r0
+    LD r31, r254, 0a, 8h
+    ADDI64 r254, r254, 8d
+    JALA r0, r31, 0a
+code size: 88
+ret: 0
+status: Ok(())

lang/tests/son_tests_comptime_function_from_another_file.txt

@@ -0,0 +1,8 @@
+main:
+    LRA r13, r0, :a
+    LD r13, r13, 0a, 8h
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 50
+ret: 50
+status: Ok(())

lang/tests/son_tests_comptime_min_reg_leak.txt

@@ -0,0 +1,8 @@
+main:
+    LRA r13, r0, :a
+    LD r13, r13, 0a, 8h
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 50
+ret: 50
+status: Ok(())

lang/tests/son_tests_conditional_stores.txt

@@ -0,0 +1,19 @@
+cond:
+    CP r1, r0
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -24d
+    ST r31, r254, 0a, 24h
+    JAL r31, r0, :cond
+    CP r33, r1
+    CP r32, r0
+    JNE r33, r32, :0
+    JMP :1
+ 0: LI64 r32, 2d
+ 1: CP r1, r32
+    LD r31, r254, 0a, 24h
+    ADDI64 r254, r254, 24d
+    JALA r0, r31, 0a
+code size: 117
+ret: 0
+status: Ok(())

lang/tests/son_tests_const_folding_with_arg.txt

@@ -0,0 +1,6 @@
+main:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 22
+ret: 0
+status: Ok(())

lang/tests/son_tests_constants.txt

@@ -0,0 +1,7 @@
+main:
+    LI32 r13, 69w
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 28
+ret: 69
+status: Ok(())

lang/tests/son_tests_dead_code_in_loop.txt

@@ -0,0 +1,6 @@
+main:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 22
+ret: 0
+status: Ok(())

lang/tests/son_tests_die.txt

@@ -0,0 +1,5 @@
+main:
+    UN 
+code size: 9
+ret: 0
+status: Err(Unreachable)

lang/tests/son_tests_different_function_destinations.txt

@@ -0,0 +1,84 @@
+main:
+    ADDI64 r254, r254, -144d
+    ST r31, r254, 80a, 64h
+    LRA r32, r0, :glob_stru
+    JAL r31, r0, :new_stru
+    ST r1, r32, 0a, 16h
+    CP r33, r0
+    LD r34, r32, 0a, 8h
+    JEQ r34, r33, :0
+    LI64 r32, 300d
+    CP r1, r32
+    JMP :1
+ 0: LI64 r35, 1d
+    ST r35, r32, 0a, 8h
+    ST r35, r32, 8a, 8h
+    ST r33, r32, 0a, 8h
+    LD r34, r32, 0a, 8h
+    JEQ r34, r33, :2
+    LI64 r32, 200d
+    CP r1, r32
+    JMP :1
+ 2: LI64 r36, 3d
+    ST r35, r32, 0a, 8h
+    ST r35, r32, 8a, 8h
+    ADDI64 r37, r254, 16d
+    ST r35, r254, 16a, 8h
+    ST r35, r254, 24a, 8h
+    ST r35, r254, 32a, 8h
+    ST r35, r254, 40a, 8h
+    ST r35, r254, 48a, 8h
+    ST r35, r254, 56a, 8h
+    CP r32, r33
+ 8: JNE r32, r36, :3
+    LD r32, r254, 48a, 8h
+    JEQ r32, r33, :4
+    LI64 r32, 100d
+    CP r1, r32
+    JMP :1
+ 4: ST r33, r254, 0a, 8h
+    ST r33, r254, 8a, 8h
+    ST r33, r254, 64a, 8h
+    ST r33, r254, 72a, 8h
+    ST r35, r254, 16a, 8h
+    ST r35, r254, 24a, 8h
+    ST r35, r254, 32a, 8h
+    ST r35, r254, 40a, 8h
+    ST r35, r254, 48a, 8h
+    ST r35, r254, 56a, 8h
+    CP r32, r33
+ 7: LD r38, r254, 48a, 8h
+    JNE r32, r36, :5
+    JEQ r38, r33, :6
+    LI64 r32, 10d
+    CP r1, r32
+    JMP :1
+ 6: CP r1, r33
+    JMP :1
+ 5: ADD64 r34, r32, r35
+    MULI64 r32, r32, 16d
+    ADD64 r32, r37, r32
+    ST r33, r32, 0a, 8h
+    ST r33, r32, 8a, 8h
+    CP r32, r34
+    JMP :7
+ 3: MULI64 r34, r32, 16d
+    ADD64 r34, r37, r34
+    JAL r31, r0, :new_stru
+    ST r1, r34, 0a, 16h
+    ADD64 r32, r32, r35
+    JMP :8
+ 1: LD r31, r254, 80a, 64h
+    ADDI64 r254, r254, 144d
+    JALA r0, r31, 0a
+new_stru:
+    ADDI64 r254, r254, -16d
+    ADDI64 r13, r254, 0d
+    ST r0, r254, 0a, 8h
+    ST r0, r254, 8a, 8h
+    LD r1, r13, 0a, 16h
+    ADDI64 r254, r254, 16d
+    JALA r0, r31, 0a
+code size: 765
+ret: 0
+status: Ok(())

lang/tests/son_tests_different_types.txt

@@ -0,0 +1,29 @@
+main:
+    ADDI64 r254, r254, -12d
+    LI8 r13, 255b
+    ST r13, r254, 0a, 1h
+    ST r0, r254, 1a, 1h
+    ST r0, r254, 2a, 1h
+    ST r13, r254, 3a, 1h
+    ST r0, r254, 4a, 4h
+    LD r13, r254, 4a, 4h
+    LI32 r14, 2w
+    ST r14, r254, 8a, 4h
+    LD r14, r254, 8a, 4h
+    LI64 r15, 2d
+    ANDI r14, r14, 4294967295d
+    JEQ r14, r15, :0
+    CP r1, r0
+    JMP :1
+ 0: ANDI r13, r13, 4294967295d
+    JEQ r13, r0, :2
+    LI64 r13, 64d
+    CP r1, r13
+    JMP :1
+ 2: LI64 r13, 512d
+    CP r1, r13
+ 1: ADDI64 r254, r254, 12d
+    JALA r0, r31, 0a
+code size: 235
+ret: 512
+status: Ok(())

lang/tests/son_tests_directives.txt

@@ -0,0 +1,22 @@
+main:
+    ADDI64 r254, r254, -16d
+    LI64 r13, 10d
+    ADDI64 r14, r254, 0d
+    ST r13, r254, 0a, 8h
+    LI64 r13, 20d
+    ST r13, r254, 8a, 8h
+    LI64 r13, 6d
+    LI64 r15, 5d
+    LI64 r16, 1d
+    CP r2, r16
+    CP r5, r15
+    CP r6, r13
+    LD r3, r14, 0a, 16h
+    ECA 
+    CP r1, r0
+    ADDI64 r254, r254, 16d
+    JALA r0, r31, 0a
+ev: Ecall
+code size: 154
+ret: 0
+status: Ok(())

lang/tests/son_tests_enums.txt

@@ -0,0 +1,20 @@
+main:
+    ADDI64 r254, r254, -16d
+    ST r31, r254, 0a, 16h
+    JAL r31, r0, :some_enum
+    CP r32, r1
+    ANDI r32, r32, 255d
+    JNE r32, r0, :0
+    CP r1, r0
+    JMP :1
+ 0: LI64 r32, 100d
+    CP r1, r32
+ 1: LD r31, r254, 0a, 16h
+    ADDI64 r254, r254, 16d
+    JALA r0, r31, 0a
+some_enum:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 128
+ret: 0
+status: Ok(())

lang/tests/son_tests_exhaustive_loop_testing.txt

@@ -0,0 +1,114 @@
+continue_and_state_change:
+    CP r13, r2
+    CP r15, r0
+    LI64 r16, 3d
+    LI64 r14, 4d
+    LI64 r17, 2d
+    LI64 r18, 10d
+ 6: JLTU r13, r18, :0
+    JMP :1
+ 0: JNE r13, r17, :2
+    CP r13, r14
+    JMP :3
+ 2: JNE r13, r16, :4
+    CP r13, r15
+ 1: CP r1, r13
+    JMP :5
+ 4: ADDI64 r13, r13, 1d
+ 3: JMP :6
+ 5: JALA r0, r31, 0a
+infinite_loop:
+    ADDI64 r254, r254, -32d
+    ST r31, r254, 0a, 32h
+    LI64 r34, 1d
+    CP r33, r0
+    CP r32, r33
+ 1: JNE r32, r34, :0
+    JMP :0
+ 0: CP r2, r33
+    JAL r31, r0, :continue_and_state_change
+    CP r32, r1
+    JMP :1
+    LD r31, r254, 0a, 32h
+    ADDI64 r254, r254, 32d
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -40d
+    ST r31, r254, 0a, 40h
+    CP r2, r0
+    JAL r31, r0, :multiple_breaks
+    CP r32, r1
+    LI64 r33, 3d
+    JEQ r32, r33, :0
+    LI64 r32, 1d
+    CP r1, r32
+    JMP :1
+ 0: LI64 r32, 4d
+    CP r2, r32
+    JAL r31, r0, :multiple_breaks
+    CP r34, r1
+    LI64 r35, 10d
+    JEQ r34, r35, :2
+    LI64 r32, 2d
+    CP r1, r32
+    JMP :1
+ 2: CP r2, r0
+    JAL r31, r0, :state_change_in_break
+    CP r34, r1
+    JEQ r34, r0, :3
+    CP r1, r33
+    JMP :1
+ 3: CP r2, r32
+    JAL r31, r0, :state_change_in_break
+    CP r34, r1
+    JEQ r34, r35, :4
+    CP r1, r32
+    JMP :1
+ 4: CP r2, r35
+    JAL r31, r0, :continue_and_state_change
+    CP r32, r1
+    JEQ r32, r35, :5
+    LI64 r32, 5d
+    CP r1, r32
+    JMP :1
+ 5: CP r2, r33
+    JAL r31, r0, :continue_and_state_change
+    CP r32, r1
+    JEQ r32, r0, :6
+    LI64 r32, 6d
+    CP r1, r32
+    JMP :1
+ 6: JAL r31, r0, :infinite_loop
+    CP r1, r0
+ 1: LD r31, r254, 0a, 40h
+    ADDI64 r254, r254, 40d
+    JALA r0, r31, 0a
+multiple_breaks:
+    CP r13, r2
+    LI64 r14, 3d
+    LI64 r15, 10d
+ 4: JLTU r13, r15, :0
+    JMP :1
+ 0: ADDI64 r13, r13, 1d
+    JNE r13, r14, :2
+ 1: CP r1, r13
+    JMP :3
+ 2: JMP :4
+ 3: JALA r0, r31, 0a
+state_change_in_break:
+    CP r13, r2
+    LI64 r14, 3d
+    LI64 r15, 10d
+ 4: JLTU r13, r15, :0
+    JMP :1
+ 0: JNE r13, r14, :2
+    CP r13, r0
+ 1: CP r1, r13
+    JMP :3
+ 2: ADDI64 r13, r13, 1d
+    JMP :4
+ 3: JALA r0, r31, 0a
+timed out
+code size: 667
+ret: 10
+status: Ok(())

lang/tests/son_tests_fb_driver.txt

@@ -0,0 +1,55 @@
+check_platform:
+    ADDI64 r254, r254, -16d
+    ST r31, r254, 0a, 16h
+    JAL r31, r0, :x86_fb_ptr
+    CP r32, r1
+    CP r1, r32
+    LD r31, r254, 0a, 16h
+    ADDI64 r254, r254, 16d
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -56d
+    ST r31, r254, 0a, 56h
+    JAL r31, r0, :check_platform
+    CP r33, r0
+    LI64 r36, 30d
+    LI64 r37, 100d
+    CP r35, r33
+    CP r34, r33
+    CP r32, r33
+ 5: JLTU r32, r36, :0
+    ADDI64 r34, r34, 1d
+    CP r2, r33
+    CP r3, r34
+    CP r4, r36
+    JAL r31, r0, :set_pixel
+    CP r32, r1
+    JEQ r32, r35, :1
+    CP r1, r33
+    JMP :2
+ 1: JNE r34, r37, :3
+    CP r1, r35
+    JMP :2
+ 3: CP r32, r33
+    JMP :4
+ 0: ADDI64 r35, r35, 1d
+    ADDI64 r32, r32, 1d
+ 4: JMP :5
+ 2: LD r31, r254, 0a, 56h
+    ADDI64 r254, r254, 56d
+    JALA r0, r31, 0a
+set_pixel:
+    CP r13, r2
+    CP r14, r3
+    CP r15, r4
+    MUL64 r14, r14, r15
+    ADD64 r13, r14, r13
+    CP r1, r13
+    JALA r0, r31, 0a
+x86_fb_ptr:
+    LI64 r13, 100d
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 329
+ret: 3000
+status: Ok(())

lang/tests/son_tests_floating_point_arithmetic.txt

@@ -0,0 +1,7 @@
+main:
+    LI32 r13, 3212836864w
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 28
+ret: 3212836864
+status: Ok(())

lang/tests/son_tests_functions.txt

@@ -0,0 +1,29 @@
+add_one:
+    CP r13, r2
+    ADDI64 r13, r13, 1d
+    CP r1, r13
+    JALA r0, r31, 0a
+add_two:
+    CP r13, r2
+    ADDI64 r13, r13, 2d
+    CP r1, r13
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -24d
+    ST r31, r254, 0a, 24h
+    LI64 r32, 10d
+    CP r2, r32
+    JAL r31, r0, :add_one
+    CP r32, r1
+    LI64 r33, 20d
+    CP r2, r33
+    JAL r31, r0, :add_two
+    CP r33, r1
+    ADD64 r32, r33, r32
+    CP r1, r32
+    LD r31, r254, 0a, 24h
+    ADDI64 r254, r254, 24d
+    JALA r0, r31, 0a
+code size: 176
+ret: 33
+status: Ok(())

lang/tests/son_tests_generic_functions.txt

@@ -0,0 +1,35 @@
+add:
+    CP r13, r2
+    CP r14, r3
+    ADD64 r13, r13, r14
+    CP r1, r13
+    JALA r0, r31, 0a
+add:
+    CP r13, r2
+    CP r14, r3
+    ADD32 r13, r13, r14
+    CP r1, r13
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -32d
+    ST r31, r254, 0a, 32h
+    LI32 r32, 2w
+    CP r2, r32
+    CP r3, r32
+    JAL r31, r0, :add
+    CP r32, r1
+    LI64 r33, 3d
+    LI64 r34, 1d
+    CP r2, r34
+    CP r3, r33
+    JAL r31, r0, :add
+    CP r33, r1
+    ANDI r32, r32, 4294967295d
+    SUB64 r32, r32, r33
+    CP r1, r32
+    LD r31, r254, 0a, 32h
+    ADDI64 r254, r254, 32d
+    JALA r0, r31, 0a
+code size: 191
+ret: 0
+status: Ok(())

lang/tests/son_tests_generic_type_mishap.txt

@@ -0,0 +1,32 @@
+main:
+    ADDI64 r254, r254, -8d
+    ST r31, r254, 0a, 8h
+    JAL r31, r0, :process
+    LD r31, r254, 0a, 8h
+    ADDI64 r254, r254, 8d
+    JALA r0, r31, 0a
+opaque:
+    JALA r0, r31, 0a
+process:
+    ADDI64 r254, r254, -48d
+    ST r31, r254, 16a, 32h
+    LI64 r32, 1000d
+    ADDI64 r33, r254, 0d
+    ST r0, r254, 0a, 1h
+ 4: JGTU r32, r0, :0
+    JMP :1
+ 0: CP r2, r33
+    JAL r31, r0, :opaque
+    LD r34, r254, 0a, 1h
+    ANDI r34, r34, 255d
+    JEQ r34, r0, :2
+    JMP :3
+ 2: ADDI64 r32, r32, -1d
+ 1: JMP :4
+ 3: LD r31, r254, 16a, 32h
+    ADDI64 r254, r254, 48d
+    JALA r0, r31, 0a
+timed out
+code size: 248
+ret: 0
+status: Ok(())

lang/tests/son_tests_generic_types.txt

@@ -0,0 +1,130 @@
+deinit:
+    ADDI64 r254, r254, -40d
+    ST r31, r254, 0a, 40h
+    CP r32, r2
+    LD r33, r32, 16a, 8h
+    LI64 r34, 8d
+    MUL64 r33, r33, r34
+    LD r35, r32, 0a, 8h
+    CP r2, r35
+    CP r3, r33
+    CP r4, r34
+    JAL r31, r0, :free
+    CP r1, r32
+    JAL r31, r0, :new
+    LD r31, r254, 0a, 40h
+    ADDI64 r254, r254, 40d
+    JALA r0, r31, 0a
+free:
+    CP r13, r2
+    CP r14, r3
+    CP r15, r4
+    LRA r16, r0, :free_sys_call
+    LD r16, r16, 0a, 8h
+    CP r2, r16
+    CP r3, r13
+    CP r4, r14
+    CP r5, r15
+    ECA 
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -56d
+    ST r31, r254, 24a, 32h
+    ADDI64 r32, r254, 0d
+    CP r1, r32
+    JAL r31, r0, :new
+    LI64 r33, 69d
+    CP r2, r32
+    CP r3, r33
+    JAL r31, r0, :push
+    CP r33, r1
+    LD r34, r254, 0a, 8h
+    LD r33, r34, 0a, 8h
+    CP r2, r32
+    JAL r31, r0, :deinit
+    CP r1, r33
+    LD r31, r254, 24a, 32h
+    ADDI64 r254, r254, 56d
+    JALA r0, r31, 0a
+malloc:
+    CP r13, r2
+    CP r14, r3
+    LRA r15, r0, :malloc_sys_call
+    LD r15, r15, 0a, 8h
+    CP r2, r15
+    CP r3, r13
+    CP r4, r14
+    ECA 
+    CP r13, r1
+    CP r1, r13
+    JALA r0, r31, 0a
+new:
+    ADDI64 r254, r254, -24d
+    CP r14, r1
+    ADDI64 r13, r254, 0d
+    ST r0, r254, 0a, 8h
+    ST r0, r254, 8a, 8h
+    ST r0, r254, 16a, 8h
+    BMC r13, r14, 24h
+    ADDI64 r254, r254, 24d
+    JALA r0, r31, 0a
+push:
+    ADDI64 r254, r254, -104d
+    ST r31, r254, 0a, 104h
+    CP r38, r2
+    CP r39, r3
+    LI64 r37, 1d
+    LD r33, r38, 8a, 8h
+    LD r32, r38, 16a, 8h
+    JNE r32, r33, :0
+    JNE r32, r0, :1
+    CP r32, r37
+    JMP :2
+ 1: MULI64 r32, r32, 2d
+ 2: LI64 r40, 8d
+    MUL64 r34, r32, r40
+    CP r2, r34
+    CP r3, r40
+    JAL r31, r0, :malloc
+    CP r35, r1
+    ST r32, r38, 16a, 8h
+    JNE r35, r0, :3
+    CP r1, r0
+    JMP :4
+ 3: MULI64 r33, r33, 8d
+    LD r32, r38, 0a, 8h
+    ADD64 r41, r32, r33
+    CP r34, r35
+ 7: LD r42, r38, 0a, 8h
+    LD r43, r38, 8a, 8h
+    JNE r41, r32, :5
+    JEQ r43, r0, :6
+    MUL64 r32, r43, r40
+    CP r2, r42
+    CP r3, r32
+    CP r4, r40
+    JAL r31, r0, :free
+    JMP :6
+ 6: ST r35, r38, 0a, 8h
+    JMP :0
+ 5: ADDI64 r36, r34, 8d
+    ADDI64 r33, r32, 8d
+    LD r32, r32, 0a, 8h
+    ST r32, r34, 0a, 8h
+    CP r34, r36
+    CP r32, r33
+    JMP :7
+ 0: LD r32, r38, 8a, 8h
+    MULI64 r33, r32, 8d
+    LD r34, r38, 0a, 8h
+    ADD64 r33, r34, r33
+    ST r39, r33, 0a, 8h
+    ADD64 r32, r32, r37
+    ST r32, r38, 8a, 8h
+    CP r1, r33
+ 4: LD r31, r254, 0a, 104h
+    ADDI64 r254, r254, 104d
+    JALA r0, r31, 0a
+code size: 923
+ret: 69
+status: Ok(())

lang/tests/son_tests_global_aliasing_overptimization.txt

@@ -0,0 +1,19 @@
+clobber:
+    LRA r13, r0, :var
+    ST r0, r13, 0a, 8h
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -24d
+    ST r31, r254, 0a, 24h
+    LRA r32, r0, :var
+    LI64 r33, 2d
+    ST r33, r32, 0a, 8h
+    JAL r31, r0, :clobber
+    LD r32, r32, 0a, 8h
+    CP r1, r32
+    LD r31, r254, 0a, 24h
+    ADDI64 r254, r254, 24d
+    JALA r0, r31, 0a
+code size: 159
+ret: 0
+status: Ok(())

lang/tests/son_tests_global_variable_wiredness.txt

@@ -0,0 +1,23 @@
+inb:
+    CP r1, r0
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -32d
+    ST r31, r254, 0a, 32h
+    LRA r32, r0, :ports
+    LD r33, r32, 0a, 1h
+    ANDI r33, r33, 255d
+    JNE r33, r0, :0
+    JMP :1
+ 0: JAL r31, r0, :inb
+    CP r33, r1
+    CMPU r34, r33, r0
+    CMPUI r34, r34, 0d
+    NOT r34, r34
+    ST r34, r32, 0a, 1h
+ 1: LD r31, r254, 0a, 32h
+    ADDI64 r254, r254, 32d
+    JALA r0, r31, 0a
+code size: 164
+ret: 0
+status: Ok(())

lang/tests/son_tests_global_variables.txt

@@ -0,0 +1,10 @@
+main:
+    LRA r13, r0, :complex_global_var
+    LD r14, r13, 0a, 8h
+    ADDI64 r14, r14, 5d
+    ST r14, r13, 0a, 8h
+    CP r1, r14
+    JALA r0, r31, 0a
+code size: 74
+ret: 55
+status: Ok(())

lang/tests/son_tests_hex_octal_binary_literals.txt

@@ -0,0 +1,6 @@
+main:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 22
+ret: 0
+status: Ok(())

lang/tests/son_tests_idk.txt

@@ -0,0 +1,21 @@
+main:
+    ADDI64 r254, r254, -128d
+    LI8 r15, 69b
+    LI64 r16, 128d
+    CP r13, r0
+    ADDI64 r17, r254, 0d
+ 2: LD r18, r254, 42a, 1h
+    JLTU r13, r16, :0
+    ANDI r13, r18, 255d
+    CP r1, r13
+    JMP :1
+ 0: ADDI64 r14, r13, 1d
+    ADD64 r13, r17, r13
+    ST r15, r13, 0a, 1h
+    CP r13, r14
+    JMP :2
+ 1: ADDI64 r254, r254, 128d
+    JALA r0, r31, 0a
+code size: 141
+ret: 69
+status: Ok(())

lang/tests/son_tests_if_statements.txt

@@ -0,0 +1,36 @@
+fib:
+    ADDI64 r254, r254, -32d
+    ST r31, r254, 0a, 32h
+    CP r32, r2
+    LI64 r33, 1d
+    LI64 r34, 2d
+    JGTU r32, r34, :0
+    CP r1, r33
+    JMP :1
+ 0: SUB64 r33, r32, r33
+    CP r2, r33
+    JAL r31, r0, :fib
+    CP r33, r1
+    SUB64 r32, r32, r34
+    CP r2, r32
+    JAL r31, r0, :fib
+    CP r32, r1
+    ADD64 r32, r32, r33
+    CP r1, r32
+ 1: LD r31, r254, 0a, 32h
+    ADDI64 r254, r254, 32d
+    JALA r0, r31, 0a
+main:
+    ADDI64 r254, r254, -16d
+    ST r31, r254, 0a, 16h
+    LI64 r32, 10d
+    CP r2, r32
+    JAL r31, r0, :fib
+    CP r32, r1
+    CP r1, r32
+    LD r31, r254, 0a, 16h
+    ADDI64 r254, r254, 16d
+    JALA r0, r31, 0a
+code size: 229
+ret: 55
+status: Ok(())

lang/tests/son_tests_infinite_loop_after_peephole.txt

@@ -0,0 +1,9 @@
+main:
+    CP r13, r0
+ 0: ADDI64 r13, r13, 1d
+    JMP :0
+    JALA r0, r31, 0a
+timed out
+code size: 38
+ret: 0
+status: Ok(())

lang/tests/son_tests_inline.txt

@@ -0,0 +1,21 @@
+main:
+    LI64 r13, 8d
+    CP r2, r13
+    ECA 
+    LI64 r14, 6d
+    LRA r13, r0, :gb
+    LD r13, r13, 0a, 8h
+    CMPU r13, r13, r0
+    CMPUI r13, r13, 0d
+    OR r13, r13, r0
+    ANDI r13, r13, 255d
+    JNE r13, r0, :0
+    CP r13, r14
+    JMP :1
+ 0: LI64 r13, 1d
+ 1: SUB64 r13, r13, r14
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 131
+ret: 0
+status: Ok(())

lang/tests/son_tests_inline_return_stack.txt

@@ -0,0 +1,6 @@
+main:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 22
+ret: 0
+status: Ok(())

lang/tests/son_tests_inline_test.txt

@@ -0,0 +1,29 @@
+main:
+    ADDI64 r254, r254, -16d
+    ST r31, r254, 0a, 16h
+    JAL r31, r0, :scalar_values
+    CP r32, r1
+    JEQ r32, r0, :0
+    LI64 r32, 1d
+    CP r1, r32
+    JMP :1
+ 0: JAL r31, r0, :structs
+    CP r32, r1
+    JEQ r32, r0, :2
+    JAL r31, r0, :structs
+    CP r32, r1
+    CP r1, r32
+    JMP :1
+ 2: CP r1, r0
+ 1: LD r31, r254, 0a, 16h
+    ADDI64 r254, r254, 16d
+    JALA r0, r31, 0a
+scalar_values:
+    CP r1, r0
+    JALA r0, r31, 0a
+structs:
+    CP r1, r0
+    JALA r0, r31, 0a
+code size: 164
+ret: 0
+status: Ok(())

lang/tests/son_tests_inlined_generic_functions.txt

@@ -0,0 +1,7 @@
+main:
+    LI64 r13, 10d
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 32
+ret: 10
+status: Ok(())

lang/tests/son_tests_inlining_issues.txt

@@ -0,0 +1,104 @@
+main:
+    ADDI64 r254, r254, -122d
+    ST r31, r254, 58a, 64h
+    ADDI64 r32, r254, 33d
+    ADDI64 r33, r254, 34d
+    ADDI64 r34, r254, 1d
+    ADDI64 r35, r254, 17d
+    ST r32, r254, 34a, 8h
+    LI64 r36, 100d
+    ADDI64 r37, r254, 0d
+    LI8 r38, 1b
+    ST r0, r254, 1a, 8h
+    ST r0, r254, 17a, 8h
+    ST r36, r254, 42a, 8h
+    ST r38, r254, 0a, 1h
+    ST r0, r254, 9a, 8h
+    ST r0, r254, 25a, 8h
+    ST r36, r254, 50a, 8h
+    ST r0, r254, 33a, 1h
+    CP r2, r33
+    LD r3, r35, 0a, 16h
+    LD r5, r34, 0a, 16h
+    LD r7, r37, 0a, 1h
+    JAL r31, r0, :put_filled_rect
+    LD r31, r254, 58a, 64h
+    ADDI64 r254, r254, 122d
+    JALA r0, r31, 0a
+put_filled_rect:
+    ADDI64 r254, r254, -108d
+    CP r14, r2
+    ST r3, r254, 92a, 16h
+    ADDI64 r3, r254, 92d
+    CP r15, r3
+    ST r5, r254, 76a, 16h
+    ADDI64 r5, r254, 76d
+    CP r13, r5
+    ST r7, r254, 75a, 1h
+    ADDI64 r7, r254, 75d
+    CP r16, r7
+    LI64 r17, 25d
+    LI64 r18, 2d
+    LI64 r19, 8d
+    ADDI64 r20, r254, 25d
+    ADDI64 r21, r254, 50d
+    LI8 r22, 5b
+    ST r22, r254, 25a, 1h
+    LD r23, r13, 0a, 8h
+    ST r23, r254, 26a, 4h
+    LI64 r24, 1d
+    ST r24, r254, 30a, 4h
+    ST r16, r254, 34a, 8h
+    ST r22, r254, 50a, 1h
+    ST r23, r254, 51a, 4h
+    ST r24, r254, 55a, 4h
+    ST r16, r254, 59a, 8h
+    LD r25, r15, 8a, 8h
+    LD r13, r13, 8a, 8h
+    ADD64 r26, r13, r25
+    SUB64 r26, r26, r24
+    LD r27, r14, 8a, 8h
+    MUL64 r26, r27, r26
+    LD r14, r14, 0a, 8h
+    ADD64 r26, r14, r26
+    LD r28, r15, 0a, 8h
+    ADD64 r15, r28, r26
+    MUL64 r25, r27, r25
+    ADD64 r14, r14, r25
+    ADD64 r14, r28, r14
+ 3: JGTU r13, r24, :0
+    JNE r13, r24, :1
+    ADDI64 r13, r254, 0d
+    ST r22, r254, 0a, 1h
+    ST r23, r254, 1a, 4h
+    ST r24, r254, 5a, 4h
+    ST r16, r254, 9a, 8h
+    ST r14, r254, 17a, 8h
+    CP r2, r19
+    CP r3, r18
+    CP r4, r13
+    CP r5, r17
+    ECA 
+    JMP :1
+ 1: JMP :2
+ 0: ST r14, r254, 67a, 8h
+    CP r2, r19
+    CP r3, r18
+    CP r4, r21
+    CP r5, r17
+    ECA 
+    ST r15, r254, 42a, 8h
+    CP r2, r19
+    CP r3, r18
+    CP r4, r20
+    CP r5, r17
+    ECA 
+    SUB64 r15, r15, r27
+    ADD64 r14, r27, r14
+    SUB64 r13, r13, r18
+    JMP :3
+ 2: ADDI64 r254, r254, 108d
+    JALA r0, r31, 0a
+code size: 875
+ret: 0
+status: Ok(())

lang/tests/son_tests_intcast_store.txt

@@ -0,0 +1,25 @@
+main:
+    ADDI64 r254, r254, -48d
+    ST r31, r254, 16a, 32h
+    ADDI64 r32, r254, 0d
+    ADDI64 r33, r254, 8d
+    ST r0, r254, 0a, 8h
+    ST r0, r254, 8a, 8h
+    LI64 r34, 1024d
+    CP r2, r33
+    CP r3, r32
+    CP r4, r34
+    JAL r31, r0, :set
+    CP r32, r1
+    ANDI r32, r32, 4294967295d
+    CP r1, r32
+    LD r31, r254, 16a, 32h
+    ADDI64 r254, r254, 48d
+    JALA r0, r31, 0a
+set:
+    CP r13, r4
+    CP r1, r13
+    JALA r0, r31, 0a
+code size: 175
+ret: 1024
+status: Ok(())