Skip to content

Remove Frame mutex and use DataStack bump allocator for LocalsPlus #7333

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
youknowone merged 14 commits into from
Mar 4, 2026
Merged

Remove Frame mutex and use DataStack bump allocator for LocalsPlus #7333

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
f869c35
Remove PyMutex<FrameState> from Frame, use UnsafeCell fields directly
youknowone Mar 3, 2026
85df41e
Add thread-local DataStack for bump-allocating frame data
youknowone Mar 3, 2026
e286597
Unify FastLocals and BoxVec stack into LocalsPlus
youknowone Mar 3, 2026
9dbe95d
Use DataStack for LocalsPlus in non-generator function calls
youknowone Mar 3, 2026
a5981f0
Fix clippy: import Layout from core::alloc instead of alloc::alloc
youknowone Mar 3, 2026
938fdf7
Fix vectorcall compatibility with LocalsPlus API
youknowone Mar 3, 2026
fa49f5d
Add datastack, nlocalsplus, ncells, tstate to cspell dictionary
youknowone Mar 3, 2026
6a6e90d
Fix DataStack pop() for non-monotonic allocation addresses
youknowone Mar 4, 2026
2cc6b58
Fix stale comments: release_datastack -> materialize_localsplus
youknowone Mar 4, 2026
f300ff3
Fix non-threading mode for parallel test execution
youknowone Mar 4, 2026
927269c
Fix CallAllocAndEnterInit to use LocalsPlus stack API
youknowone Mar 4, 2026
e8d94bb
Use checked arithmetic in LocalsPlus and DataStack allocators
youknowone Mar 4, 2026
16778fc
Address code review: checked arithmetic, threading feature deps, Send…
youknowone Mar 4, 2026
22df0c6
Clean up comments: remove redundant/stale remarks, fix CPython refere…
youknowone Mar 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Use checked arithmetic in LocalsPlus and DataStack allocators
  • Loading branch information
@youknowone
youknowone committed Mar 4, 2026
commit e8d94bb74e58c2bab2d47f023374202cbb019b7c
5 changes: 4 additions & 1 deletion crates/common/src/lock.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,10 @@ cfg_if::cfg_if! {
pub use std::sync::LazyLock;
} else {
Comment thread
coderabbitai[bot] marked this conversation as resolved.
pub struct LazyLock<T, F = fn() -> T>(core::cell::LazyCell<T, F>);
// SAFETY: Without std, there can be no threads.
// SAFETY: This branch is only active when both "std" and "threading"
// features are absent — i.e., truly single-threaded no_std environments
// (e.g., embedded or bare-metal WASM). Without std, the Rust runtime
// cannot spawn threads, so Sync is trivially satisfied.
unsafe impl<T, F> Sync for LazyLock<T, F> {}
Comment thread
coderabbitai[bot] marked this conversation as resolved.

impl<T, F: FnOnce() -> T> LazyLock<T, F> {
Expand Down
11 changes: 8 additions & 3 deletions crates/vm/src/datastack.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,10 +102,15 @@ impl DataStack {
#[inline(never)]
fn push_slow(&mut self, aligned_size: usize) -> *mut u8 {
let mut chunk_size = MIN_CHUNK_SIZE;
let needed =
aligned_size + MINIMUM_OVERHEAD + core::mem::size_of::<DataStackChunk>() + ALIGN;
let needed = aligned_size
.checked_add(MINIMUM_OVERHEAD)
.and_then(|v| v.checked_add(core::mem::size_of::<DataStackChunk>()))
.and_then(|v| v.checked_add(ALIGN))
.expect("DataStack chunk size overflow");
while chunk_size < needed {
chunk_size *= 2;
chunk_size = chunk_size
.checked_mul(2)
.expect("DataStack chunk size overflow");
}
// Save current position in old chunk.
unsafe {
Expand Down
91 changes: 59 additions & 32 deletions crates/vm/src/frame.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -163,10 +163,13 @@ const _: () = {
impl LocalsPlus {
/// Create a new heap-backed LocalsPlus. All slots start as None (0).
fn new(nlocalsplus: usize, stacksize: usize) -> Self {
let capacity = nlocalsplus + stacksize;
let capacity = nlocalsplus
.checked_add(stacksize)
.expect("LocalsPlus capacity overflow");
let nlocalsplus_u32 = u32::try_from(nlocalsplus).expect("nlocalsplus exceeds u32");
Self {
data: LocalsPlusData::Heap(vec![0usize; capacity].into_boxed_slice()),
nlocalsplus: nlocalsplus as u32,
nlocalsplus: nlocalsplus_u32,
stack_top: 0,
}
}
Expand All @@ -177,14 +180,19 @@ impl LocalsPlus {
/// The caller must call `materialize_localsplus()` when the frame finishes
/// to migrate data to the heap, then `datastack_pop()` to free the memory.
fn new_on_datastack(nlocalsplus: usize, stacksize: usize, vm: &VirtualMachine) -> Self {
let capacity = nlocalsplus + stacksize;
let byte_size = capacity * core::mem::size_of::<usize>();
let capacity = nlocalsplus
.checked_add(stacksize)
.expect("LocalsPlus capacity overflow");
let byte_size = capacity
.checked_mul(core::mem::size_of::<usize>())
.expect("LocalsPlus byte size overflow");
let nlocalsplus_u32 = u32::try_from(nlocalsplus).expect("nlocalsplus exceeds u32");
let ptr = vm.datastack_push(byte_size) as *mut usize;
// Zero-initialize all slots (0 = None for both PyObjectRef and PyStackRef).
unsafe { core::ptr::write_bytes(ptr, 0, capacity) };
Self {
data: LocalsPlusData::DataStack { ptr, capacity },
nlocalsplus: nlocalsplus as u32,
nlocalsplus: nlocalsplus_u32,
stack_top: 0,
Comment thread
coderabbitai[bot] marked this conversation as resolved.
}
}
Expand Down Expand Up @@ -2362,18 +2370,16 @@ impl ExecutingFrame<'_> {
// Same as LoadFast but explicitly checks for unbound locals
// (LoadFast in RustPython already does this check)
let idx = idx.get(arg) as usize;
let x = self.localsplus.fastlocals()[idx]
.clone()
.ok_or_else(|| {
vm.new_exception_msg(
vm.ctx.exceptions.unbound_local_error.to_owned(),
format!(
"local variable '{}' referenced before assignment",
self.code.varnames[idx]
)
.into(),
let x = self.localsplus.fastlocals()[idx].clone().ok_or_else(|| {
vm.new_exception_msg(
vm.ctx.exceptions.unbound_local_error.to_owned(),
format!(
"local variable '{}' referenced before assignment",
self.code.varnames[idx]
)
})?;
.into(),
)
})?;
self.push_value(x);
Ok(None)
}
Expand Down Expand Up @@ -2413,18 +2419,16 @@ impl ExecutingFrame<'_> {
// lifetime issues at yield/exception points are resolved.
Instruction::LoadFastBorrow { var_num: idx } => {
let idx = idx.get(arg) as usize;
let x = self.localsplus.fastlocals()[idx]
.clone()
.ok_or_else(|| {
vm.new_exception_msg(
vm.ctx.exceptions.unbound_local_error.to_owned(),
format!(
"local variable '{}' referenced before assignment",
self.code.varnames[idx]
)
.into(),
let x = self.localsplus.fastlocals()[idx].clone().ok_or_else(|| {
vm.new_exception_msg(
vm.ctx.exceptions.unbound_local_error.to_owned(),
format!(
"local variable '{}' referenced before assignment",
self.code.varnames[idx]
)
})?;
.into(),
)
})?;
self.push_value(x);
Ok(None)
}
Expand Down Expand Up @@ -4312,7 +4316,10 @@ impl ExecutingFrame<'_> {
let nargs: u32 = arg.into();
let callable = self.nth_value(nargs + 1);
let stack_len = self.localsplus.stack_len();
let self_or_null_is_some = self.localsplus.stack_index(stack_len - nargs as usize - 1).is_some();
let self_or_null_is_some = self
.localsplus
.stack_index(stack_len - nargs as usize - 1)
.is_some();
if !self_or_null_is_some
&& cached_version != 0
&& let Some(cls) = callable.downcast_ref::<PyType>()
Expand Down Expand Up @@ -5994,11 +6001,21 @@ impl ExecutingFrame<'_> {
args_vec.push(self_val);
}
for stack_idx in args_start..stack_len {
let val = self.localsplus.stack_index_mut(stack_idx).take().unwrap().to_pyobj();
let val = self
.localsplus
.stack_index_mut(stack_idx)
.take()
.unwrap()
.to_pyobj();
args_vec.push(val);
}

let callable_obj = self.localsplus.stack_index_mut(callable_idx).take().unwrap().to_pyobj();
let callable_obj = self
.localsplus
.stack_index_mut(callable_idx)
.take()
.unwrap()
.to_pyobj();
self.localsplus.stack_truncate(callable_idx);

// invoke_vectorcall falls back to FuncArgs if no vectorcall slot
Expand Down Expand Up @@ -6049,11 +6066,21 @@ impl ExecutingFrame<'_> {
args_vec.push(self_val);
}
for stack_idx in args_start..stack_len {
let val = self.localsplus.stack_index_mut(stack_idx).take().unwrap().to_pyobj();
let val = self
.localsplus
.stack_index_mut(stack_idx)
.take()
.unwrap()
.to_pyobj();
args_vec.push(val);
}

let callable_obj = self.localsplus.stack_index_mut(callable_idx).take().unwrap().to_pyobj();
let callable_obj = self
.localsplus
.stack_index_mut(callable_idx)
.take()
.unwrap()
.to_pyobj();
self.localsplus.stack_truncate(callable_idx);

// invoke_vectorcall falls back to FuncArgs if no vectorcall slot
Expand Down