From 87919ee553483923f4d8e42c259fdb9c79b00df4 Mon Sep 17 00:00:00 2001 From: koniifer Date: Sun, 1 Sep 2024 21:19:29 +0100 Subject: [PATCH] merge master/trunk --- hblang/README.md | 38 +++++++- hblang/src/codegen.rs | 27 ++++-- hblang/src/lexer.rs | 111 ---------------------- hblang/src/lib.rs | 2 +- hblang/src/parser.rs | 16 ++-- hblang/tests/codegen_tests_arithmetic.txt | 2 +- 6 files changed, 64 insertions(+), 132 deletions(-) diff --git a/hblang/README.md b/hblang/README.md index 4990af1..2a29379 100644 --- a/hblang/README.md +++ b/hblang/README.md @@ -47,7 +47,7 @@ main := fn(): int { #### arithmetic ```hb main := fn(): int { - return 10 - 20 / 2 + 4 * (2 + 2) - 4 * 4 + 1 + return 10 - 20 / 2 + 4 * (2 + 2) - 4 * 4 + 1 << 0 } ``` @@ -71,6 +71,7 @@ add_one := fn(x: int): int { // commant is an item main := fn(): int { // comment is a statement + foo(/* comment is an exprression /* if you are crazy */ */) return 0 } @@ -452,7 +453,6 @@ main := fn(): int { ``` ### Purely Testing Examples - #### different_types ```hb Color := struct { @@ -589,4 +589,36 @@ main := fn(): int { color := MAGENTA return color.r } -``` \ No newline at end of file +``` + +#### comptime_function_from_another_file +```hb +.{min} := @use("math.hb") +a := min(100, 50) + +main := fn(): int { + return a +} + + +// in module: math.hb + + +SIZEOF_INT := 32 +SHIFT := SIZEOF_INT - 1 +min := fn(a: int, b: int): int { + c := a - b + return b + (c & c >> SHIFT) +} +``` + +#### using_structs_in_registers +```hb +ColorBGRA := struct {b: u8, g: u8, r: u8, a: u8} +MAGENTA := ColorBGRA.{b: 205, g: 0, r: 205, a: 255} + +main := fn(): int { + color := MAGENTA + return color.r +} +``` diff --git a/hblang/src/codegen.rs b/hblang/src/codegen.rs index c628df1..9d2121d 100644 --- a/hblang/src/codegen.rs +++ b/hblang/src/codegen.rs @@ -3,7 +3,7 @@ use { crate::{ ident::{self, Ident}, instrs::{self, *}, - lexer::TokenKind, + lexer::{self, TokenKind}, log, parser::{self, find_symbol, idfl, CtorField, Expr, ExprRef, FileId, Pos}, HashMap, @@ -1935,11 +1935,11 @@ impl Codegen { if let Some(val) = val { let size = self.ci.ret.map_or(17, |ty| self.tys.size_of(ty)); let loc = match size { - 0 => Loc::default(), - 1..=16 => Loc::reg(1), - _ => Loc::reg(self.ci.ret_reg.as_ref()).into_derefed(), + 0 => None, + 1..=16 => Some(Loc::reg(1)), + _ => Some(Loc::reg(self.ci.ret_reg.as_ref()).into_derefed()), }; - let ty = self.expr_ctx(val, Ctx { ty: self.ci.ret, loc: Some(loc) })?.ty; + let ty = self.expr_ctx(val, Ctx { ty: self.ci.ret, loc })?.ty; match self.ci.ret { None => self.ci.ret = Some(ty), Some(ret) => _ = self.assert_ty(pos, ty, ret), @@ -2626,6 +2626,15 @@ impl Codegen { fn alloc_ret(&mut self, ret: ty::Id, ctx: Ctx) -> Loc { let size = self.tys.size_of(ret); + if size == 0 { + debug_assert!(ctx.loc.is_none(), "{}", self.ty_display(ret)); + return Loc::default(); + } + + if ctx.loc.is_some() && size < 16 { + return ctx.loc.unwrap(); + } + match size { 0 => Loc::default(), 1..=8 => Loc::reg(1), @@ -3030,6 +3039,8 @@ impl Codegen { let ret_loc = unsafe { self.output.code.as_mut_ptr().add(offset) }; self.ct.vm.write_reg(1, ret_loc as u64); + self.ci.free_loc(ret.loc); + Global { ty: ret.ty, offset: offset as _ } } @@ -3133,7 +3144,7 @@ impl Codegen { } fn report_log(&self, pos: Pos, msg: impl std::fmt::Display) { - let (line, col) = self.cfile().nlines.line_col(pos); + let (line, col) = lexer::line_col(self.cfile().file.as_bytes(), pos); println!("{}:{}:{}: {}", self.cfile().path, line, col, msg); } @@ -3261,7 +3272,7 @@ mod tests { let mut codegen = super::Codegen { files: module_map .iter() - .map(|&(path, content)| parser::Ast::new(path, content, &loader)) + .map(|&(path, content)| parser::Ast::new(path, content.to_owned(), &loader)) .collect(), ..Default::default() }; @@ -3338,7 +3349,7 @@ mod tests { struct_patterns => README; arrays => README; struct_return_from_module_function => README; - comptime_pointers => README; + //comptime_pointers => README; sort_something_viredly => README; hex_octal_binary_literals => README; comptime_function_from_another_file => README; diff --git a/hblang/src/lexer.rs b/hblang/src/lexer.rs index 9b4a1ae..76a3345 100644 --- a/hblang/src/lexer.rs +++ b/hblang/src/lexer.rs @@ -1,5 +1,3 @@ -use std::simd::cmp::SimdPartialEq; - const fn ascii_mask(chars: &[u8]) -> u128 { let mut eq = 0; let mut i = 0; @@ -398,112 +396,3 @@ pub fn line_col(bytes: &[u8], pos: u32) -> (usize, usize) { .map(|(line, col)| (line + 1, col + 1)) .unwrap_or((1, 1)) } - -pub struct LineMap { - lines: Box<[u8]>, -} - -impl LineMap { - pub fn line_col(&self, mut pos: u32) -> (usize, usize) { - let mut line = 1; - - let mut iter = self.lines.iter().copied(); - - loop { - let mut acc = 0; - let mut idx = 0; - loop { - let len = iter.next().unwrap(); - acc |= ((len & 0x7F) as u32) << (7 * idx); - idx += 1; - if len & 0x80 == 0 { - break; - } - } - - if pos < acc { - break; - } - pos = pos.saturating_sub(acc); - line += 1; - } - - (line, pos as usize + 1) - } - - pub fn new(input: &str) -> Self { - let bytes = input.as_bytes(); - let (start, simd_mid, end) = bytes.as_simd::<16>(); - - let query = std::simd::u8x16::splat(b'\n'); - - let nl_count = start.iter().map(|&b| (b == b'\n') as usize).sum::() - + simd_mid.iter().map(|s| s.simd_eq(query).to_bitmask().count_ones()).sum::() - as usize - + end.iter().map(|&b| (b == b'\n') as usize).sum::(); - - let mut lines = Vec::with_capacity(nl_count); - let mut last_nl = 0; - - let handle_rem = |offset: usize, bytes: &[u8], last_nl: &mut usize, lines: &mut Vec| { - bytes - .iter() - .copied() - .enumerate() - .filter_map(|(i, b)| (b == b'\n').then_some(i + offset)) - .for_each(|i| { - lines.push((i - *last_nl + 1) as u8); - *last_nl = i + 1; - }); - }; - - handle_rem(0, start, &mut last_nl, &mut lines); - - for (i, simd) in simd_mid.iter().enumerate() { - let mask = simd.simd_eq(query); - let mut mask = mask.to_bitmask(); - while mask != 0 { - let idx = mask.trailing_zeros() as usize + i * 16 + start.len(); - let mut len = idx - last_nl + 1; - while len >= 0x80 { - lines.push(0x80 | (len & 0x7F) as u8); - len >>= 7; - } - lines.push(len as u8); - last_nl = idx + 1; - mask &= mask - 1; - } - } - - handle_rem(bytes.len() - end.len(), end, &mut last_nl, &mut lines); - - Self { lines: Box::from(lines) } - } -} - -#[cfg(test)] -mod test { - #[test] - fn test_smh() { - let example = include_str!("../README.md"); - - let nlines = super::LineMap::new(example); - - fn slow_nline_search(str: &str, mut pos: usize) -> (usize, usize) { - ( - str.lines() - .take_while(|l| match pos.checked_sub(l.len() + 1) { - Some(nl) => (pos = nl, true).1, - None => false, - }) - .count() - + 1, - pos + 1, - ) - } - - for i in 0..example.len() { - assert_eq!(slow_nline_search(example, i), nlines.line_col(i as _)); - } - } -} diff --git a/hblang/src/lib.rs b/hblang/src/lib.rs index 7059548..9ce7957 100644 --- a/hblang/src/lib.rs +++ b/hblang/src/lib.rs @@ -424,7 +424,7 @@ pub fn parse_from_fs(extra_threads: usize, root: &str) -> io::Result> { let mut file = std::fs::File::open(path)?; file.read_to_end(buffer)?; let src = std::str::from_utf8(buffer).map_err(InvalidFileData)?; - Ok(Ast::new(path, src, &loader)) + Ok(Ast::new(path, src.to_owned(), &loader)) }; let thread = || { diff --git a/hblang/src/parser.rs b/hblang/src/parser.rs index 541e66f..8355084 100644 --- a/hblang/src/parser.rs +++ b/hblang/src/parser.rs @@ -2,7 +2,7 @@ use { crate::{ codegen, ident::{self, Ident}, - lexer::{self, Lexer, LineMap, Token, TokenKind}, + lexer::{self, Lexer, Token, TokenKind}, log, }, std::{ @@ -1090,7 +1090,7 @@ pub struct AstInner { exprs: *const [Expr<'static>], pub path: Box, - pub nlines: LineMap, + pub file: Box, pub symbols: T, } @@ -1102,11 +1102,11 @@ impl AstInner<[Symbol]> { .0 } - fn new(content: &str, path: &str, loader: Loader) -> NonNull { + fn new(content: String, path: &str, loader: Loader) -> NonNull { let arena = Arena::default(); let mut syms = Vec::new(); let mut parser = Parser::new(&arena, &mut syms, loader); - let exprs = parser.file(content, path) as *const [Expr<'static>]; + let exprs = parser.file(&content, path) as *const [Expr<'static>]; syms.sort_unstable_by_key(|s| s.name); @@ -1121,7 +1121,7 @@ impl AstInner<[Symbol]> { mem: arena.chunk.into_inner(), exprs, path: path.into(), - nlines: LineMap::new(content), + file: content.into(), symbols: (), }); std::ptr::addr_of_mut!((*inner).symbols) @@ -1137,7 +1137,7 @@ impl AstInner<[Symbol]> { pub struct Ast(NonNull>); impl Ast { - pub fn new(path: &str, content: &str, loader: Loader) -> Self { + pub fn new(path: &str, content: String, loader: Loader) -> Self { Self(AstInner::new(content, path, loader)) } @@ -1168,7 +1168,7 @@ impl std::fmt::Display for Ast { impl Default for Ast { fn default() -> Self { - Self(AstInner::new("", "", &no_loader)) + Self(AstInner::new(String::new(), "", &no_loader)) } } @@ -1344,7 +1344,7 @@ impl Drop for ArenaChunk { #[cfg(test)] pub mod test { pub fn format(ident: &str, input: &str) { - let ast = super::Ast::new(ident, input, &|_, _| Ok(0)); + let ast = super::Ast::new(ident, input.to_owned(), &|_, _| Ok(0)); let mut output = Vec::new(); crate::format_to(&ast, input, &mut output).unwrap(); diff --git a/hblang/tests/codegen_tests_arithmetic.txt b/hblang/tests/codegen_tests_arithmetic.txt index de488f4..8773e5c 100644 --- a/hblang/tests/codegen_tests_arithmetic.txt +++ b/hblang/tests/codegen_tests_arithmetic.txt @@ -1,3 +1,3 @@ -code size: 189 +code size: 193 ret: 1 status: Ok(())