move the LargeAssignments lint logic into its own file

This commit is contained in:
Ralf Jung 2024-04-14 18:09:40 +02:00
parent a3269e920c
commit 80dc3d14c9
2 changed files with 161 additions and 140 deletions

View File

@ -205,6 +205,8 @@
//! this is not implemented however: a mono item will be produced
//! regardless of whether it is actually needed or not.
mod move_check;
use rustc_data_structures::fx::{FxHashMap, FxHashSet};
use rustc_data_structures::sync::{par_for_each_in, LRef, MTLock};
use rustc_hir as hir;
@ -227,7 +229,6 @@ use rustc_middle::ty::{
};
use rustc_middle::ty::{GenericArgKind, GenericArgs};
use rustc_session::config::EntryFnType;
use rustc_session::lint::builtin::LARGE_ASSIGNMENTS;
use rustc_session::Limit;
use rustc_span::source_map::{dummy_spanned, respan, Spanned};
use rustc_span::symbol::{sym, Ident};
@ -236,9 +237,9 @@ use rustc_target::abi::Size;
use std::path::PathBuf;
use crate::errors::{
self, EncounteredErrorWhileInstantiating, LargeAssignmentsLint, NoOptimizedMir, RecursionLimit,
TypeLengthLimit,
self, EncounteredErrorWhileInstantiating, NoOptimizedMir, RecursionLimit, TypeLengthLimit,
};
use move_check::MoveCheckState;
#[derive(PartialEq)]
pub enum MonoItemCollectionStrategy {
@ -667,11 +668,8 @@ struct MirUsedCollector<'a, 'tcx> {
/// Note that this contains *not-monomorphized* items!
used_mentioned_items: &'a mut FxHashSet<MentionedItem<'tcx>>,
instance: Instance<'tcx>,
/// Spans for move size lints already emitted. Helps avoid duplicate lints.
move_size_spans: Vec<Span>,
visiting_call_terminator: bool,
/// Set of functions for which it is OK to move large data into.
skip_move_check_fns: Option<Vec<DefId>>,
move_check: move_check::MoveCheckState,
}
impl<'a, 'tcx> MirUsedCollector<'a, 'tcx> {
@ -687,124 +685,6 @@ impl<'a, 'tcx> MirUsedCollector<'a, 'tcx> {
)
}
fn check_operand_move_size(&mut self, operand: &mir::Operand<'tcx>, location: Location) {
let limit = self.tcx.move_size_limit();
if limit.0 == 0 {
return;
}
// This function is called by visit_operand() which visits _all_
// operands, including TerminatorKind::Call operands. But if
// check_fn_args_move_size() has been called, the operands have already
// been visited. Do not visit them again.
if self.visiting_call_terminator {
return;
}
let source_info = self.body.source_info(location);
debug!(?source_info);
if let Some(too_large_size) = self.operand_size_if_too_large(limit, operand) {
self.lint_large_assignment(limit.0, too_large_size, location, source_info.span);
};
}
fn check_fn_args_move_size(
&mut self,
callee_ty: Ty<'tcx>,
args: &[Spanned<mir::Operand<'tcx>>],
fn_span: Span,
location: Location,
) {
let limit = self.tcx.move_size_limit();
if limit.0 == 0 {
return;
}
if args.is_empty() {
return;
}
// Allow large moves into container types that themselves are cheap to move
let ty::FnDef(def_id, _) = *callee_ty.kind() else {
return;
};
if self
.skip_move_check_fns
.get_or_insert_with(|| build_skip_move_check_fns(self.tcx))
.contains(&def_id)
{
return;
}
debug!(?def_id, ?fn_span);
for arg in args {
// Moving args into functions is typically implemented with pointer
// passing at the llvm-ir level and not by memcpy's. So always allow
// moving args into functions.
let operand: &mir::Operand<'tcx> = &arg.node;
if let mir::Operand::Move(_) = operand {
continue;
}
if let Some(too_large_size) = self.operand_size_if_too_large(limit, operand) {
self.lint_large_assignment(limit.0, too_large_size, location, arg.span);
};
}
}
fn operand_size_if_too_large(
&mut self,
limit: Limit,
operand: &mir::Operand<'tcx>,
) -> Option<Size> {
let ty = operand.ty(self.body, self.tcx);
let ty = self.monomorphize(ty);
let Ok(layout) = self.tcx.layout_of(ty::ParamEnv::reveal_all().and(ty)) else {
return None;
};
if layout.size.bytes_usize() > limit.0 {
debug!(?layout);
Some(layout.size)
} else {
None
}
}
fn lint_large_assignment(
&mut self,
limit: usize,
too_large_size: Size,
location: Location,
span: Span,
) {
let source_info = self.body.source_info(location);
debug!(?source_info);
for reported_span in &self.move_size_spans {
if reported_span.overlaps(span) {
return;
}
}
let lint_root = source_info.scope.lint_root(&self.body.source_scopes);
debug!(?lint_root);
let Some(lint_root) = lint_root else {
// This happens when the issue is in a function from a foreign crate that
// we monomorphized in the current crate. We can't get a `HirId` for things
// in other crates.
// FIXME: Find out where to report the lint on. Maybe simply crate-level lint root
// but correct span? This would make the lint at least accept crate-level lint attributes.
return;
};
self.tcx.emit_node_span_lint(
LARGE_ASSIGNMENTS,
lint_root,
span,
LargeAssignmentsLint { span, size: too_large_size.bytes(), limit: limit as u64 },
);
self.move_size_spans.push(span);
}
/// Evaluates a *not yet monomorphized* constant.
fn eval_constant(
&mut self,
@ -1367,19 +1247,6 @@ fn assoc_fn_of_type<'tcx>(tcx: TyCtxt<'tcx>, def_id: DefId, fn_ident: Ident) ->
return None;
}
fn build_skip_move_check_fns(tcx: TyCtxt<'_>) -> Vec<DefId> {
let fns = [
(tcx.lang_items().owned_box(), "new"),
(tcx.get_diagnostic_item(sym::Rc), "new"),
(tcx.get_diagnostic_item(sym::Arc), "new"),
];
fns.into_iter()
.filter_map(|(def_id, fn_name)| {
def_id.and_then(|def_id| assoc_fn_of_type(tcx, def_id, Ident::from_str(fn_name)))
})
.collect::<Vec<_>>()
}
/// Scans the MIR in order to find function calls, closures, and drop-glue.
///
/// Anything that's found is added to `output`. Furthermore the "mentioned items" of the MIR are returned.
@ -1409,9 +1276,8 @@ fn collect_items_of_instance<'tcx>(
used_items,
used_mentioned_items: &mut used_mentioned_items,
instance,
move_size_spans: vec![],
visiting_call_terminator: false,
skip_move_check_fns: None,
move_check: MoveCheckState::new(),
};
if mode == CollectionMode::UsedItems {

View File

@ -0,0 +1,155 @@
use rustc_session::lint::builtin::LARGE_ASSIGNMENTS;
use super::*;
use crate::errors::LargeAssignmentsLint;
pub(super) struct MoveCheckState {
/// Spans for move size lints already emitted. Helps avoid duplicate lints.
move_size_spans: Vec<Span>,
/// Set of functions for which it is OK to move large data into.
skip_move_check_fns: Option<Vec<DefId>>,
}
impl MoveCheckState {
pub(super) fn new() -> Self {
MoveCheckState { move_size_spans: vec![], skip_move_check_fns: None }
}
}
impl<'a, 'tcx> MirUsedCollector<'a, 'tcx> {
pub(super) fn check_operand_move_size(
&mut self,
operand: &mir::Operand<'tcx>,
location: Location,
) {
let limit = self.tcx.move_size_limit();
if limit.0 == 0 {
return;
}
// This function is called by visit_operand() which visits _all_
// operands, including TerminatorKind::Call operands. But if
// check_fn_args_move_size() has been called, the operands have already
// been visited. Do not visit them again.
if self.visiting_call_terminator {
return;
}
let source_info = self.body.source_info(location);
debug!(?source_info);
if let Some(too_large_size) = self.operand_size_if_too_large(limit, operand) {
self.lint_large_assignment(limit.0, too_large_size, location, source_info.span);
};
}
pub(super) fn check_fn_args_move_size(
&mut self,
callee_ty: Ty<'tcx>,
args: &[Spanned<mir::Operand<'tcx>>],
fn_span: Span,
location: Location,
) {
let limit = self.tcx.move_size_limit();
if limit.0 == 0 {
return;
}
if args.is_empty() {
return;
}
// Allow large moves into container types that themselves are cheap to move
let ty::FnDef(def_id, _) = *callee_ty.kind() else {
return;
};
if self
.move_check
.skip_move_check_fns
.get_or_insert_with(|| build_skip_move_check_fns(self.tcx))
.contains(&def_id)
{
return;
}
debug!(?def_id, ?fn_span);
for arg in args {
// Moving args into functions is typically implemented with pointer
// passing at the llvm-ir level and not by memcpy's. So always allow
// moving args into functions.
let operand: &mir::Operand<'tcx> = &arg.node;
if let mir::Operand::Move(_) = operand {
continue;
}
if let Some(too_large_size) = self.operand_size_if_too_large(limit, operand) {
self.lint_large_assignment(limit.0, too_large_size, location, arg.span);
};
}
}
fn operand_size_if_too_large(
&mut self,
limit: Limit,
operand: &mir::Operand<'tcx>,
) -> Option<Size> {
let ty = operand.ty(self.body, self.tcx);
let ty = self.monomorphize(ty);
let Ok(layout) = self.tcx.layout_of(ty::ParamEnv::reveal_all().and(ty)) else {
return None;
};
if layout.size.bytes_usize() > limit.0 {
debug!(?layout);
Some(layout.size)
} else {
None
}
}
fn lint_large_assignment(
&mut self,
limit: usize,
too_large_size: Size,
location: Location,
span: Span,
) {
let source_info = self.body.source_info(location);
debug!(?source_info);
for reported_span in &self.move_check.move_size_spans {
if reported_span.overlaps(span) {
return;
}
}
let lint_root = source_info.scope.lint_root(&self.body.source_scopes);
debug!(?lint_root);
let Some(lint_root) = lint_root else {
// This happens when the issue is in a function from a foreign crate that
// we monomorphized in the current crate. We can't get a `HirId` for things
// in other crates.
// FIXME: Find out where to report the lint on. Maybe simply crate-level lint root
// but correct span? This would make the lint at least accept crate-level lint attributes.
return;
};
self.tcx.emit_node_span_lint(
LARGE_ASSIGNMENTS,
lint_root,
span,
LargeAssignmentsLint { span, size: too_large_size.bytes(), limit: limit as u64 },
);
self.move_check.move_size_spans.push(span);
}
}
fn build_skip_move_check_fns(tcx: TyCtxt<'_>) -> Vec<DefId> {
let fns = [
(tcx.lang_items().owned_box(), "new"),
(tcx.get_diagnostic_item(sym::Rc), "new"),
(tcx.get_diagnostic_item(sym::Arc), "new"),
];
fns.into_iter()
.filter_map(|(def_id, fn_name)| {
def_id.and_then(|def_id| assoc_fn_of_type(tcx, def_id, Ident::from_str(fn_name)))
})
.collect::<Vec<_>>()
}