Auto merge of rust-lang#45476 - Xanewok:fingerprint-disambiguator, r=michaelwoerister

Use 128 bit instead of Symbol for crate disambiguator

As discussed on gitter, this changes `crate_disambiguator` from Strings to what they are represented as, a 128 bit number.

There's also one bit I think also needs to change, but wasn't 100% sure how: [create_root_def](https://github.com/rust-lang/rust/blob/f338dba29705e144bad8b2a675284538dd514896/src/librustc/hir/map/definitions.rs#L468-L482). Should I change `DefKey::root_parent_stable_hash` to accept `Fingerprint` as crate_disambiguator to quickly combine the hash of `crate_name` with the new 128 bit hash instead of a string for a disambiguator?

r? @michaelwoerister

EDIT: Are those 3 tests `mir-opt` failing, because the hash is different, because we calculate it a little bit differently (storing directly instead of hashing the hex-string representation)? Should it be updated like in rust-lang#45319?

Author: bors

src/librustc/hir/map/collector.rs

@@ -12,6 +12,7 @@ use super::*;
 
 use dep_graph::{DepGraph, DepKind, DepNodeIndex};
 use hir::intravisit::{Visitor, NestedVisitorMap};
+use session::CrateDisambiguator;
 use std::iter::repeat;
 use syntax::ast::{NodeId, CRATE_NODE_ID};
 use syntax_pos::Span;
@@ -118,7 +119,7 @@ impl<'a, 'hir> NodeCollector<'a, 'hir> {
     }
 
     pub(super) fn finalize_and_compute_crate_hash(self,
-                                                  crate_disambiguator: &str)
+                                                  crate_disambiguator: CrateDisambiguator)
                                                   -> Vec<MapEntry<'hir>> {
         let mut node_hashes: Vec<_> = self
             .hir_body_nodes
@@ -133,7 +134,7 @@ impl<'a, 'hir> NodeCollector<'a, 'hir> {
 
         self.dep_graph.with_task(DepNode::new_no_params(DepKind::Krate),
                                  &self.hcx,
-                                 (node_hashes, crate_disambiguator),
+                                 (node_hashes, crate_disambiguator.to_fingerprint()),
                                  identity_fn);
         self.map
     }

src/librustc/hir/map/def_collector.rs

@@ -10,6 +10,7 @@
 
 use hir::map::definitions::*;
 use hir::def_id::{CRATE_DEF_INDEX, DefIndex, DefIndexAddressSpace};
+use session::CrateDisambiguator;
 
 use syntax::ast::*;
 use syntax::ext::hygiene::Mark;
@@ -43,7 +44,9 @@ impl<'a> DefCollector<'a> {
         }
     }
 
-    pub fn collect_root(&mut self, crate_name: &str, crate_disambiguator: &str) {
+    pub fn collect_root(&mut self,
+                        crate_name: &str,
+                        crate_disambiguator: CrateDisambiguator) {
         let root = self.definitions.create_root_def(crate_name,
                                                     crate_disambiguator);
         assert_eq!(root, CRATE_DEF_INDEX);

src/librustc/hir/map/definitions.rs

@@ -22,6 +22,7 @@ use rustc_data_structures::fx::FxHashMap;
 use rustc_data_structures::indexed_vec::IndexVec;
 use rustc_data_structures::stable_hasher::StableHasher;
 use serialize::{Encodable, Decodable, Encoder, Decoder};
+use session::CrateDisambiguator;
 use std::fmt::Write;
 use std::hash::Hash;
 use syntax::ast;
@@ -231,7 +232,9 @@ impl DefKey {
         DefPathHash(hasher.finish())
     }
 
-    fn root_parent_stable_hash(crate_name: &str, crate_disambiguator: &str) -> DefPathHash {
+    fn root_parent_stable_hash(crate_name: &str,
+                               crate_disambiguator: CrateDisambiguator)
+                               -> DefPathHash {
         let mut hasher = StableHasher::new();
         // Disambiguate this from a regular DefPath hash,
         // see compute_stable_hash() above.
@@ -467,7 +470,7 @@ impl Definitions {
     /// Add a definition with a parent definition.
     pub fn create_root_def(&mut self,
                            crate_name: &str,
-                           crate_disambiguator: &str)
+                           crate_disambiguator: CrateDisambiguator)
                            -> DefIndex {
         let key = DefKey {
             parent: None,

src/librustc/hir/map/mod.rs

@@ -1014,8 +1014,8 @@ pub fn map_crate<'hir>(sess: &::session::Session,
                                                 hcx);
         intravisit::walk_crate(&mut collector, &forest.krate);
 
-        let crate_disambiguator = sess.local_crate_disambiguator().as_str();
-        collector.finalize_and_compute_crate_hash(&crate_disambiguator)
+        let crate_disambiguator = sess.local_crate_disambiguator();
+        collector.finalize_and_compute_crate_hash(crate_disambiguator)
     };
 
     if log_enabled!(::log::LogLevel::Debug) {

src/librustc/middle/cstore.rs

@@ -30,7 +30,7 @@ use hir::map::definitions::{Definitions, DefKey, DefPathTable};
 use hir::svh::Svh;
 use ich;
 use ty::{self, TyCtxt};
-use session::Session;
+use session::{Session, CrateDisambiguator};
 use session::search_paths::PathKind;
 use util::nodemap::NodeSet;
 
@@ -267,7 +267,7 @@ pub trait CrateStore {
     fn export_macros_untracked(&self, cnum: CrateNum);
     fn dep_kind_untracked(&self, cnum: CrateNum) -> DepKind;
     fn crate_name_untracked(&self, cnum: CrateNum) -> Symbol;
-    fn crate_disambiguator_untracked(&self, cnum: CrateNum) -> Symbol;
+    fn crate_disambiguator_untracked(&self, cnum: CrateNum) -> CrateDisambiguator;
     fn crate_hash_untracked(&self, cnum: CrateNum) -> Svh;
     fn struct_field_names_untracked(&self, def: DefId) -> Vec<ast::Name>;
     fn item_children_untracked(&self, did: DefId, sess: &Session) -> Vec<def::Export>;
@@ -338,7 +338,7 @@ impl CrateStore for DummyCrateStore {
     fn dep_kind_untracked(&self, cnum: CrateNum) -> DepKind { bug!("is_explicitly_linked") }
     fn export_macros_untracked(&self, cnum: CrateNum) { bug!("export_macros") }
     fn crate_name_untracked(&self, cnum: CrateNum) -> Symbol { bug!("crate_name") }
-    fn crate_disambiguator_untracked(&self, cnum: CrateNum) -> Symbol {
+    fn crate_disambiguator_untracked(&self, cnum: CrateNum) -> CrateDisambiguator {
         bug!("crate_disambiguator")
     }
     fn crate_hash_untracked(&self, cnum: CrateNum) -> Svh { bug!("crate_hash") }

src/librustc/session/mod.rs

@@ -12,6 +12,7 @@ pub use self::code_stats::{CodeStats, DataTypeKind, FieldInfo};
 pub use self::code_stats::{SizeKind, TypeSizeInfo, VariantInfo};
 
 use hir::def_id::{CrateNum, DefIndex};
+use ich::Fingerprint;
 
 use lint;
 use middle::allocator::AllocatorKind;
@@ -29,7 +30,6 @@ use syntax::json::JsonEmitter;
 use syntax::feature_gate;
 use syntax::parse;
 use syntax::parse::ParseSess;
-use syntax::symbol::Symbol;
 use syntax::{ast, codemap};
 use syntax::feature_gate::AttributeType;
 use syntax_pos::{Span, MultiSpan};
@@ -83,12 +83,12 @@ pub struct Session {
     pub plugin_attributes: RefCell<Vec<(String, AttributeType)>>,
     pub crate_types: RefCell<Vec<config::CrateType>>,
     pub dependency_formats: RefCell<dependency_format::Dependencies>,
-    /// The crate_disambiguator is constructed out of all the `-C metadata`
+        /// The crate_disambiguator is constructed out of all the `-C metadata`
     /// arguments passed to the compiler. Its value together with the crate-name
     /// forms a unique global identifier for the crate. It is used to allow
     /// multiple crates with the same name to coexist. See the
     /// trans::back::symbol_names module for more information.
-    pub crate_disambiguator: RefCell<Option<Symbol>>,
+    pub crate_disambiguator: RefCell<Option<CrateDisambiguator>>,
     pub features: RefCell<feature_gate::Features>,
 
     /// The maximum recursion limit for potentially infinitely recursive
@@ -165,9 +165,9 @@ enum DiagnosticBuilderMethod {
 }
 
 impl Session {
-    pub fn local_crate_disambiguator(&self) -> Symbol {
+    pub fn local_crate_disambiguator(&self) -> CrateDisambiguator {
         match *self.crate_disambiguator.borrow() {
-            Some(sym) => sym,
+            Some(value) => value,
             None => bug!("accessing disambiguator before initialization"),
         }
     }
@@ -471,14 +471,18 @@ impl Session {
 
     /// Returns the symbol name for the registrar function,
     /// given the crate Svh and the function DefIndex.
-    pub fn generate_plugin_registrar_symbol(&self, disambiguator: Symbol, index: DefIndex)
+    pub fn generate_plugin_registrar_symbol(&self, disambiguator: CrateDisambiguator,
+                                            index: DefIndex)
                                             -> String {
-        format!("__rustc_plugin_registrar__{}_{}", disambiguator, index.as_usize())
+        format!("__rustc_plugin_registrar__{}_{}", disambiguator.to_fingerprint().to_hex(),
+                                                   index.as_usize())
     }
 
-    pub fn generate_derive_registrar_symbol(&self, disambiguator: Symbol, index: DefIndex)
+    pub fn generate_derive_registrar_symbol(&self, disambiguator: CrateDisambiguator,
+                                            index: DefIndex)
                                             -> String {
-        format!("__rustc_derive_registrar__{}_{}", disambiguator, index.as_usize())
+        format!("__rustc_derive_registrar__{}_{}", disambiguator.to_fingerprint().to_hex(),
+                                                   index.as_usize())
     }
 
     pub fn sysroot<'a>(&'a self) -> &'a Path {
@@ -838,6 +842,26 @@ pub fn build_session_(sopts: config::Options,
     sess
 }
 
+/// Hash value constructed out of all the `-C metadata` arguments passed to the
+/// compiler. Together with the crate-name forms a unique global identifier for
+/// the crate.
+#[derive(Eq, PartialEq, Ord, PartialOrd, Hash, Debug, Clone, Copy, RustcEncodable, RustcDecodable)]
+pub struct CrateDisambiguator(Fingerprint);
+
+impl CrateDisambiguator {
+    pub fn to_fingerprint(self) -> Fingerprint {
+        self.0
+    }
+}
+
+impl From<Fingerprint> for CrateDisambiguator {
+    fn from(fingerprint: Fingerprint) -> CrateDisambiguator {
+        CrateDisambiguator(fingerprint)
+    }
+}
+
+impl_stable_hash_for!(tuple_struct CrateDisambiguator { fingerprint });
+
 /// Holds data on the current incremental compilation session, if there is one.
 #[derive(Debug)]
 pub enum IncrCompSession {

src/librustc/ty/context.rs

@@ -1249,7 +1249,7 @@ impl<'a, 'gcx, 'tcx> TyCtxt<'a, 'gcx, 'tcx> {
                 crate_name,
                 // Don't print the whole crate disambiguator. That's just
                 // annoying in debug output.
-                &(crate_disambiguator.as_str())[..4],
+                &(crate_disambiguator.to_fingerprint().to_hex())[..4],
                 self.def_path(def_id).to_string_no_crate())
     }
 

src/librustc/ty/maps/mod.rs

@@ -29,7 +29,7 @@ use middle::lang_items::{LanguageItems, LangItem};
 use middle::exported_symbols::SymbolExportLevel;
 use middle::trans::{CodegenUnit, Stats};
 use mir;
-use session::CompileResult;
+use session::{CompileResult, CrateDisambiguator};
 use session::config::OutputFilenames;
 use traits::Vtable;
 use traits::specialization_graph;
@@ -285,7 +285,7 @@ define_maps! { <'tcx>
     [] fn native_libraries: NativeLibraries(CrateNum) -> Rc<Vec<NativeLibrary>>,
     [] fn plugin_registrar_fn: PluginRegistrarFn(CrateNum) -> Option<DefId>,
     [] fn derive_registrar_fn: DeriveRegistrarFn(CrateNum) -> Option<DefId>,
-    [] fn crate_disambiguator: CrateDisambiguator(CrateNum) -> Symbol,
+    [] fn crate_disambiguator: CrateDisambiguator(CrateNum) -> CrateDisambiguator,
     [] fn crate_hash: CrateHash(CrateNum) -> Svh,
     [] fn original_crate_name: OriginalCrateName(CrateNum) -> Symbol,
 

src/librustc/ty/mod.rs

@@ -26,6 +26,7 @@ use middle::privacy::AccessLevels;
 use middle::resolve_lifetime::ObjectLifetimeDefault;
 use mir::Mir;
 use mir::GeneratorLayout;
+use session::CrateDisambiguator;
 use traits;
 use ty;
 use ty::subst::{Subst, Substs};
@@ -2556,7 +2557,7 @@ fn param_env<'a, 'tcx>(tcx: TyCtxt<'a, 'tcx, 'tcx>,
 }
 
 fn crate_disambiguator<'a, 'tcx>(tcx: TyCtxt<'a, 'tcx, 'tcx>,
-                                 crate_num: CrateNum) -> Symbol {
+                                 crate_num: CrateNum) -> CrateDisambiguator {
     assert_eq!(crate_num, LOCAL_CRATE);
     tcx.sess.local_crate_disambiguator()
 }

src/librustc_driver/driver.rs

@@ -14,7 +14,7 @@ use rustc::hir::lowering::lower_crate;
 use rustc::ich::Fingerprint;
 use rustc_data_structures::stable_hasher::StableHasher;
 use rustc_mir as mir;
-use rustc::session::{Session, CompileResult};
+use rustc::session::{Session, CompileResult, CrateDisambiguator};
 use rustc::session::CompileIncomplete;
 use rustc::session::config::{self, Input, OutputFilenames, OutputType};
 use rustc::session::search_paths::PathKind;
@@ -58,7 +58,6 @@ use syntax::{ast, diagnostics, visit};
 use syntax::attr;
 use syntax::ext::base::ExtCtxt;
 use syntax::parse::{self, PResult};
-use syntax::symbol::Symbol;
 use syntax::util::node_count::NodeCounter;
 use syntax;
 use syntax_ext;
@@ -633,12 +632,12 @@ pub fn phase_2_configure_and_expand<F>(sess: &Session,
 
     *sess.crate_types.borrow_mut() = collect_crate_types(sess, &krate.attrs);
 
-    let disambiguator = Symbol::intern(&compute_crate_disambiguator(sess));
+    let disambiguator = compute_crate_disambiguator(sess);
     *sess.crate_disambiguator.borrow_mut() = Some(disambiguator);
     rustc_incremental::prepare_session_directory(
         sess,
         &crate_name,
-        &disambiguator.as_str(),
+        disambiguator,
     );
 
     let dep_graph = if sess.opts.build_dep_graph() {
@@ -1312,16 +1311,13 @@ pub fn collect_crate_types(session: &Session, attrs: &[ast::Attribute]) -> Vec<c
         .collect()
 }
 
-pub fn compute_crate_disambiguator(session: &Session) -> String {
+pub fn compute_crate_disambiguator(session: &Session) -> CrateDisambiguator {
     use std::hash::Hasher;
 
     // The crate_disambiguator is a 128 bit hash. The disambiguator is fed
     // into various other hashes quite a bit (symbol hashes, incr. comp. hashes,
     // debuginfo type IDs, etc), so we don't want it to be too wide. 128 bits
     // should still be safe enough to avoid collisions in practice.
-    // FIXME(mw): It seems that the crate_disambiguator is used everywhere as
-    //            a hex-string instead of raw bytes. We should really use the
-    //            smaller representation.
     let mut hasher = StableHasher::<Fingerprint>::new();
 
     let mut metadata = session.opts.cg.metadata.clone();
@@ -1340,11 +1336,13 @@ pub fn compute_crate_disambiguator(session: &Session) -> String {
         hasher.write(s.as_bytes());
     }
 
-    // If this is an executable, add a special suffix, so that we don't get
-    // symbol conflicts when linking against a library of the same name.
+    // Also incorporate crate type, so that we don't get symbol conflicts when
+    // linking against a library of the same name, if this is an executable.
     let is_exe = session.crate_types.borrow().contains(&config::CrateTypeExecutable);
+    hasher.write(if is_exe { b"exe" } else { b"lib" });
+
+    CrateDisambiguator::from(hasher.finish())
 
-    format!("{}{}", hasher.finish().to_hex(), if is_exe { "-exe" } else {""})
 }
 
 pub fn build_output_filenames(input: &Input,

src/librustc_incremental/persist/fs.rs

@@ -115,7 +115,7 @@
 //! implemented.
 
 use rustc::hir::svh::Svh;
-use rustc::session::Session;
+use rustc::session::{Session, CrateDisambiguator};
 use rustc::util::fs as fs_util;
 use rustc_data_structures::{flock, base_n};
 use rustc_data_structures::fx::{FxHashSet, FxHashMap};
@@ -188,7 +188,7 @@ pub fn in_incr_comp_dir(incr_comp_session_dir: &Path, file_name: &str) -> PathBu
 /// The garbage collection will take care of it.
 pub fn prepare_session_directory(sess: &Session,
                                  crate_name: &str,
-                                 crate_disambiguator: &str) {
+                                 crate_disambiguator: CrateDisambiguator) {
     if sess.opts.incremental.is_none() {
         return
     }
@@ -614,21 +614,17 @@ fn string_to_timestamp(s: &str) -> Result<SystemTime, ()> {
 
 fn crate_path(sess: &Session,
               crate_name: &str,
-              crate_disambiguator: &str)
+              crate_disambiguator: CrateDisambiguator)
               -> PathBuf {
-    use std::hash::{Hasher, Hash};
-    use std::collections::hash_map::DefaultHasher;
 
     let incr_dir = sess.opts.incremental.as_ref().unwrap().clone();
 
-    // The full crate disambiguator is really long. A hash of it should be
+    // The full crate disambiguator is really long. 64 bits of it should be
     // sufficient.
-    let mut hasher = DefaultHasher::new();
-    crate_disambiguator.hash(&mut hasher);
+    let crate_disambiguator = crate_disambiguator.to_fingerprint().to_smaller_hash();
+    let crate_disambiguator = base_n::encode(crate_disambiguator, INT_ENCODE_BASE);
 
-    let crate_name = format!("{}-{}",
-                             crate_name,
-                             base_n::encode(hasher.finish(), INT_ENCODE_BASE));
+    let crate_name = format!("{}-{}", crate_name, crate_disambiguator);
     incr_dir.join(crate_name)
 }
 

src/librustc_metadata/creader.rs

@@ -19,7 +19,7 @@ use rustc::hir::def_id::{CrateNum, DefIndex, CRATE_DEF_INDEX};
 use rustc::hir::svh::Svh;
 use rustc::middle::allocator::AllocatorKind;
 use rustc::middle::cstore::DepKind;
-use rustc::session::Session;
+use rustc::session::{Session, CrateDisambiguator};
 use rustc::session::config::{Sanitizer, self};
 use rustc_back::PanicStrategy;
 use rustc::session::search_paths::PathKind;
@@ -626,7 +626,7 @@ impl<'a> CrateLoader<'a> {
     pub fn find_plugin_registrar(&mut self,
                                  span: Span,
                                  name: &str)
-                                 -> Option<(PathBuf, Symbol, DefIndex)> {
+                                 -> Option<(PathBuf, CrateDisambiguator, DefIndex)> {
         let ekrate = self.read_extension_crate(span, &ExternCrateInfo {
              name: Symbol::intern(name),
              ident: Symbol::intern(name),

src/librustc_metadata/cstore.rs

@@ -17,6 +17,7 @@ use rustc::hir::def_id::{CRATE_DEF_INDEX, CrateNum, DefIndex};
 use rustc::hir::map::definitions::DefPathTable;
 use rustc::hir::svh::Svh;
 use rustc::middle::cstore::{DepKind, ExternCrate, MetadataLoader};
+use rustc::session::CrateDisambiguator;
 use rustc_back::PanicStrategy;
 use rustc_data_structures::indexed_vec::IndexVec;
 use rustc::util::nodemap::{FxHashMap, FxHashSet, NodeMap};
@@ -171,7 +172,7 @@ impl CrateMetadata {
     pub fn hash(&self) -> Svh {
         self.root.hash
     }
-    pub fn disambiguator(&self) -> Symbol {
+    pub fn disambiguator(&self) -> CrateDisambiguator {
         self.root.disambiguator
     }