Auto merge of rust-lang#122207 - Urgau:cfg-check-cfg-cache, r=<try>

Add cache-based fast path for cfgs and check-cfgs

This PR add a fast path for testing if a cfg is enabled and is expected.

This cache reduce the number of lookup from 2/3 to just 1 (in the good path).

Currently:
 - when there are no expecteds cfgs, we do 2 lookup (one for finding if the cfg is enabled and the second for trying to get the expected values)
 - when there are some expecteds cfgs, we do 3 lookup in the good case (same as above + one to see if the value is in the expected values) and 2 in the "bad" case

With this PR we will only do 1 lookup if the cfg is expected (aka the good case) no matter if the cfg is actually enabled or not. See the PR changes/commits for more details.

The perf runs done below shows that there is no to a slightly improvement, which is expected as we do less work, but we also don't have any benchmark with many cfgs to be assertive.

Author: bors

compiler/rustc_attr/src/builtin.rs

@@ -526,38 +526,40 @@ pub fn cfg_matches(
 ) -> bool {
     eval_condition(cfg, sess, features, &mut |cfg| {
         try_gate_cfg(cfg.name, cfg.span, sess, features);
-        match sess.psess.check_config.expecteds.get(&cfg.name) {
-            Some(ExpectedValues::Some(values)) if !values.contains(&cfg.value) => {
-                sess.psess.buffer_lint_with_diagnostic(
-                    UNEXPECTED_CFGS,
-                    cfg.span,
-                    lint_node_id,
-                    if let Some(value) = cfg.value {
-                        format!("unexpected `cfg` condition value: `{value}`")
-                    } else {
-                        format!("unexpected `cfg` condition value: (none)")
-                    },
-                    BuiltinLintDiag::UnexpectedCfgValue(
-                        (cfg.name, cfg.name_span),
-                        cfg.value.map(|v| (v, cfg.value_span.unwrap())),
-                    ),
-                );
-            }
-            None if sess.psess.check_config.exhaustive_names => {
-                sess.psess.buffer_lint_with_diagnostic(
-                    UNEXPECTED_CFGS,
-                    cfg.span,
-                    lint_node_id,
-                    format!("unexpected `cfg` condition name: `{}`", cfg.name),
-                    BuiltinLintDiag::UnexpectedCfgName(
-                        (cfg.name, cfg.name_span),
-                        cfg.value.map(|v| (v, cfg.value_span.unwrap())),
-                    ),
-                );
+        sess.psess.config_cache.get(&(cfg.name, cfg.value)).copied().unwrap_or_else(|| {
+            match sess.psess.check_config.expecteds.get(&cfg.name) {
+                Some(ExpectedValues::Some(values)) if !values.contains(&cfg.value) => {
+                    sess.psess.buffer_lint_with_diagnostic(
+                        UNEXPECTED_CFGS,
+                        cfg.span,
+                        lint_node_id,
+                        if let Some(value) = cfg.value {
+                            format!("unexpected `cfg` condition value: `{value}`")
+                        } else {
+                            format!("unexpected `cfg` condition value: (none)")
+                        },
+                        BuiltinLintDiag::UnexpectedCfgValue(
+                            (cfg.name, cfg.name_span),
+                            cfg.value.map(|v| (v, cfg.value_span.unwrap())),
+                        ),
+                    );
+                }
+                None if sess.psess.check_config.exhaustive_names => {
+                    sess.psess.buffer_lint_with_diagnostic(
+                        UNEXPECTED_CFGS,
+                        cfg.span,
+                        lint_node_id,
+                        format!("unexpected `cfg` condition name: `{}`", cfg.name),
+                        BuiltinLintDiag::UnexpectedCfgName(
+                            (cfg.name, cfg.name_span),
+                            cfg.value.map(|v| (v, cfg.value_span.unwrap())),
+                        ),
+                    );
+                }
+                _ => { /* not unexpected */ }
             }
-            _ => { /* not unexpected */ }
-        }
-        sess.psess.config.contains(&(cfg.name, cfg.value))
+            sess.psess.config.contains(&(cfg.name, cfg.value))
+        })
     })
 }
 

compiler/rustc_interface/src/interface.rs

@@ -20,7 +20,7 @@ use rustc_query_impl::QueryCtxt;
 use rustc_query_system::query::print_query_stack;
 use rustc_session::config::{self, Cfg, CheckCfg, ExpectedValues, Input, OutFileName};
 use rustc_session::filesearch::{self, sysroot_candidates};
-use rustc_session::parse::ParseSess;
+use rustc_session::parse::{CfgCache, ParseSess};
 use rustc_session::{lint, CompilerIO, EarlyDiagCtxt, Session};
 use rustc_span::source_map::{FileLoader, RealFileLoader, SourceMapInputs};
 use rustc_span::symbol::sym;
@@ -297,6 +297,35 @@ pub(crate) fn parse_check_cfg(dcx: &DiagCtxt, specs: Vec<String>) -> CheckCfg {
     check_cfg
 }
 
+/// Create the config cache of `--cfg` and `--check-cfg` for internal use.
+///
+/// The cache is used to reduce the number of hashmap lookups by
+/// pre-computing the maximum possible configs[^1] from `--check-cfg` and `--cfg`.
+///
+/// It heavily favours expected cfgs (if any are provided), since if the
+/// user provides a list of expected cfgs, they are likely to care more
+/// about expected cfgs than unexpected cfgs.
+///
+/// [^1]: See [`CfgCache`] for an explanation of "maximum possible".
+fn config_cache(config: &Cfg, check_cfg: &CheckCfg) -> CfgCache {
+    let mut config_cache = CfgCache::default();
+
+    if check_cfg.expecteds.is_empty() {
+        config_cache.extend(config.iter().map(|cfg| (*cfg, true)));
+    } else {
+        #[allow(rustc::potential_query_instability)]
+        for (name, expected) in &check_cfg.expecteds {
+            if let ExpectedValues::Some(values) = expected {
+                config_cache.extend(
+                    values.iter().map(|value| ((*name, *value), config.contains(&(*name, *value)))),
+                );
+            }
+        }
+    }
+
+    config_cache
+}
+
 /// The compiler configuration
 pub struct Config {
     /// Command line options
@@ -459,6 +488,8 @@ pub fn run_compiler<R: Send>(config: Config, f: impl FnOnce(&Compiler) -> R + Se
             check_cfg.fill_well_known(&sess.target);
             sess.psess.check_config = check_cfg;
 
+            sess.psess.config_cache = config_cache(&sess.psess.config, &sess.psess.check_config);
+
             if let Some(psess_created) = config.psess_created {
                 psess_created(&mut sess.psess);
             }

compiler/rustc_session/src/parse.rs

@@ -199,12 +199,24 @@ pub fn add_feature_diagnostics_for_issue<G: EmissionGuarantee>(
     }
 }
 
+/// Config cache that tries to combine expected cfgs and enabled cfgs.
+///
+/// If a config is found, this means it is definitely an expected config
+/// and the value represents whenever that config is enabled.
+///
+/// If a config is not found, it doesn't imply anything! The config still may
+/// be enabled and it may also be expected. You should only rely on the
+/// cache having a config and not "NOT having it". You should always fallback
+/// to a manual lookup in `ParseSess::config` and `ParseSess::check_cfg`.
+pub type CfgCache = FxHashMap<(Symbol, Option<Symbol>), bool>;
+
 /// Info about a parsing session.
 pub struct ParseSess {
     pub dcx: DiagCtxt,
     pub unstable_features: UnstableFeatures,
     pub config: Cfg,
     pub check_config: CheckCfg,
+    pub config_cache: CfgCache,
     pub edition: Edition,
     /// Places where raw identifiers were used. This is used to avoid complaining about idents
     /// clashing with keywords in new editions.
@@ -253,6 +265,7 @@ impl ParseSess {
             unstable_features: UnstableFeatures::from_environment(None),
             config: Cfg::default(),
             check_config: CheckCfg::default(),
+            config_cache: CfgCache::default(),
             edition: ExpnId::root().expn_data().edition,
             raw_identifier_spans: Default::default(),
             bad_unicode_identifiers: Lock::new(Default::default()),