rishi/scryer-prolog
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

ensure '\0' atom is static & never inlined (#2880)

Browse Source
This commit is contained in:
Mark Thom
2025-04-12 18:48:47 -07:00
committed by Mark Thom
parent 907e64e185
commit 1297a24469
2 changed files with 16 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
build/static_string_indexing.rs
View File

@@ -87,7 +87,7 @@ impl<'ast> Visit<'ast> for StaticStrVisitor {
const INLINED_ATOM_MAX_LEN: usize = 6; const INLINED_ATOM_MAX_LEN: usize = 6;
fn static_string_index(string: &str, index: usize) -> u64 { fn static_string_index(string: &str, index: usize) -> u64 {
if 0 < string.len() && string.len() <= INLINED_ATOM_MAX_LEN { if !string.is_empty() && string.len() <= INLINED_ATOM_MAX_LEN && !string.contains('\u{0}') {
let mut string_buf: [u8; 8] = [0u8; 8]; let mut string_buf: [u8; 8] = [0u8; 8];
string_buf[..string.len()].copy_from_slice(string.as_bytes()); string_buf[..string.len()].copy_from_slice(string.as_bytes());
(u64::from_le_bytes(string_buf) << 1) | 1 (u64::from_le_bytes(string_buf) << 1) | 1
@@ -183,8 +183,7 @@ pub fn index_static_strings(instruction_rs_path: &std::path::Path) -> TokenStrea
}) })
.collect(); .collect();
let static_strs_len = static_strs.len(); // visitor.static_strs.len(); let static_strs_len = static_strs.len();
//let static_strs: &Vec<_> = &visitor.static_strs.into_iter().collect();
quote! { quote! {
static STRINGS: [&str; #static_strs_len] = [ static STRINGS: [&str; #static_strs_len] = [

27
src/atom_table.rs
View File

@@ -49,9 +49,9 @@ const_assert!(mem::size_of::<Atom>() == 8);
impl AtomCell { impl AtomCell {
#[inline] #[inline]
pub fn new_static(index: u64) -> Self { fn new_static(index: u64) -> Self {
// upper 23 bits of index must be 0 // upper 23 bits of index must be 0
debug_assert!(index & !((1 << 49) - 1) == 0); debug_assert_eq!(index & !((1 << 49) - 1), 0);
AtomCell::new() AtomCell::new()
.with_name(index) .with_name(index)
.with_arity(0u8) .with_arity(0u8)
@@ -62,7 +62,7 @@ impl AtomCell {
} }
#[inline] #[inline]
pub fn new_inlined(string: &str, arity: u8) -> Self { fn new_inlined(string: &str, arity: u8) -> Self {
debug_assert!(string.len() <= INLINED_ATOM_MAX_LEN); debug_assert!(string.len() <= INLINED_ATOM_MAX_LEN);
let mut string_buf: [u8; 8] = [0u8; 8]; let mut string_buf: [u8; 8] = [0u8; 8];
@@ -80,6 +80,10 @@ impl AtomCell {
#[inline] #[inline]
pub fn new_char_inlined(c: char) -> Self { pub fn new_char_inlined(c: char) -> Self {
if c == '\u{0}' {
return Self::new_static(NULL_ATOM.flat_index());
}
let mut char_buf = [0u8; 8]; let mut char_buf = [0u8; 8];
c.encode_utf8(&mut char_buf); c.encode_utf8(&mut char_buf);
@@ -135,6 +139,7 @@ include!(concat!(env!("OUT_DIR"), "/static_atoms.rs"));
// populate these in STRINGS so they can be used from build_functor // populate these in STRINGS so they can be used from build_functor
const _: Atom = atom!("."); const _: Atom = atom!(".");
const _: Atom = atom!("[]"); const _: Atom = atom!("[]");
const NULL_ATOM: Atom = atom!("\0");
impl<'a> From<&'a Atom> for Atom { impl<'a> From<&'a Atom> for Atom {
#[inline] #[inline]
@@ -222,16 +227,12 @@ pub enum AtomString<'a> {
Dynamic(AtomTableRef<str>), Dynamic(AtomTableRef<str>),
} }
#[inline(always)]
fn inlined_to_str(bytes: &[u8; 8]) -> &str { fn inlined_to_str(bytes: &[u8; 8]) -> &str {
// allow the '\0\' atom to be represented as the 0-valued inlined atom let slice_len = bytes
let slice_len = if bytes[0] == 0 { .iter()
1 .position(|&b| b == 0u8)
} else { .unwrap_or(INLINED_ATOM_MAX_LEN);
bytes
.iter()
.position(|&b| b == 0u8)
.unwrap_or(INLINED_ATOM_MAX_LEN)
};
unsafe { str::from_utf8_unchecked(&bytes[..slice_len]) } unsafe { str::from_utf8_unchecked(&bytes[..slice_len]) }
} }
@@ -467,7 +468,7 @@ impl AtomTable {
} }
pub fn build_with(atom_table: &AtomTable, string: &str) -> Atom { pub fn build_with(atom_table: &AtomTable, string: &str) -> Atom {
if 0 < string.len() && string.len() <= INLINED_ATOM_MAX_LEN && !string.contains('\u{0}') { if !string.is_empty() && string.len() <= INLINED_ATOM_MAX_LEN && !string.contains('\u{0}') {
return Atom::new_inlined(string); return Atom::new_inlined(string);
} }