Auto merge of #83273 - cjgillot:endecode, r=michaelwoerister

Simplify encoder and decoder

Extracted from https://github.com/rust-lang/rust/pull/83036 and https://github.com/rust-lang/rust/pull/82780.

compiler/rustc_data_structures/src/fingerprint.rs

@@ -1,8 +1,5 @@
 use crate::stable_hasher;
-use rustc_serialize::{
+use rustc_serialize::{Decodable, Encodable};
-    opaque::{self, EncodeResult, FileEncodeResult},
-    Decodable, Encodable,
-};
 use std::hash::{Hash, Hasher};
 use std::mem::{self, MaybeUninit};
 
@@ -63,16 +60,6 @@ pub fn combine_commutative(self, other: Fingerprint) -> Fingerprint {
     pub fn to_hex(&self) -> String {
         format!("{:x}{:x}", self.0, self.1)
     }
-
-    pub fn decode_opaque(decoder: &mut opaque::Decoder<'_>) -> Result<Fingerprint, String> {
-        let mut bytes: [MaybeUninit<u8>; 16] = MaybeUninit::uninit_array();
-
-        decoder.read_raw_bytes(&mut bytes)?;
-
-        let [l, r]: [u64; 2] = unsafe { mem::transmute(bytes) };
-
-        Ok(Fingerprint(u64::from_le(l), u64::from_le(r)))
-    }
 }
 
 impl std::fmt::Display for Fingerprint {
@@ -130,55 +117,22 @@ fn finish(hasher: stable_hasher::StableHasher) -> Self {
 impl_stable_hash_via_hash!(Fingerprint);
 
 impl<E: rustc_serialize::Encoder> Encodable<E> for Fingerprint {
+    #[inline]
     fn encode(&self, s: &mut E) -> Result<(), E::Error> {
-        s.encode_fingerprint(self)
+        let bytes: [u8; 16] = unsafe { mem::transmute([self.0.to_le(), self.1.to_le()]) };
-    }
+        s.emit_raw_bytes(&bytes)?;
-}
-
-impl<D: rustc_serialize::Decoder> Decodable<D> for Fingerprint {
-    fn decode(d: &mut D) -> Result<Self, D::Error> {
-        d.decode_fingerprint()
-    }
-}
-
-pub trait FingerprintEncoder: rustc_serialize::Encoder {
-    fn encode_fingerprint(&mut self, f: &Fingerprint) -> Result<(), Self::Error>;
-}
-
-pub trait FingerprintDecoder: rustc_serialize::Decoder {
-    fn decode_fingerprint(&mut self) -> Result<Fingerprint, Self::Error>;
-}
-
-impl<E: rustc_serialize::Encoder> FingerprintEncoder for E {
-    default fn encode_fingerprint(&mut self, _: &Fingerprint) -> Result<(), E::Error> {
-        panic!("Cannot encode `Fingerprint` with `{}`", std::any::type_name::<E>());
-    }
-}
-
-impl FingerprintEncoder for opaque::Encoder {
-    fn encode_fingerprint(&mut self, f: &Fingerprint) -> EncodeResult {
-        let bytes: [u8; 16] = unsafe { mem::transmute([f.0.to_le(), f.1.to_le()]) };
-        self.emit_raw_bytes(&bytes);
         Ok(())
     }
 }
 
-impl FingerprintEncoder for opaque::FileEncoder {
+impl<D: rustc_serialize::Decoder> Decodable<D> for Fingerprint {
-    fn encode_fingerprint(&mut self, f: &Fingerprint) -> FileEncodeResult {
+    #[inline]
-        let bytes: [u8; 16] = unsafe { mem::transmute([f.0.to_le(), f.1.to_le()]) };
+    fn decode(d: &mut D) -> Result<Self, D::Error> {
-        self.emit_raw_bytes(&bytes)
+        let mut bytes: [MaybeUninit<u8>; 16] = MaybeUninit::uninit_array();
-    }
+        d.read_raw_bytes(&mut bytes)?;
-}
 
-impl<D: rustc_serialize::Decoder> FingerprintDecoder for D {
+        let [l, r]: [u64; 2] = unsafe { mem::transmute(bytes) };
-    default fn decode_fingerprint(&mut self) -> Result<Fingerprint, D::Error> {
+        Ok(Fingerprint(u64::from_le(l), u64::from_le(r)))
-        panic!("Cannot decode `Fingerprint` with `{}`", std::any::type_name::<D>());
-    }
-}
-
-impl FingerprintDecoder for opaque::Decoder<'_> {
-    fn decode_fingerprint(&mut self) -> Result<Fingerprint, String> {
-        Fingerprint::decode_opaque(self)
     }
 }
 

compiler/rustc_incremental/src/persist/file_format.rs

@@ -15,6 +15,7 @@
 use std::path::Path;
 
 use rustc_serialize::opaque::{FileEncodeResult, FileEncoder};
+use rustc_serialize::Encoder;
 
 /// The first few bytes of files generated by incremental compilation.
 const FILE_MAGIC: &[u8] = b"RSIC";

compiler/rustc_metadata/src/rmeta/decoder.rs

@@ -7,7 +7,6 @@
 use rustc_ast as ast;
 use rustc_attr as attr;
 use rustc_data_structures::captures::Captures;
-use rustc_data_structures::fingerprint::{Fingerprint, FingerprintDecoder};
 use rustc_data_structures::fx::FxHashMap;
 use rustc_data_structures::svh::Svh;
 use rustc_data_structures::sync::{Lock, LockGuard, Lrc, OnceCell};
@@ -351,12 +350,6 @@ fn decode(d: &mut DecodeContext<'a, 'tcx>) -> Result<DefIndex, String> {
     }
 }
 
-impl<'a, 'tcx> FingerprintDecoder for DecodeContext<'a, 'tcx> {
-    fn decode_fingerprint(&mut self) -> Result<Fingerprint, String> {
-        Fingerprint::decode_opaque(&mut self.opaque)
-    }
-}
-
 impl<'a, 'tcx> Decodable<DecodeContext<'a, 'tcx>> for SyntaxContext {
     fn decode(decoder: &mut DecodeContext<'a, 'tcx>) -> Result<SyntaxContext, String> {
         let cdata = decoder.cdata();

compiler/rustc_metadata/src/rmeta/encoder.rs

@@ -1,7 +1,6 @@
 use crate::rmeta::table::{FixedSizeEncoding, TableBuilder};
 use crate::rmeta::*;
 
-use rustc_data_structures::fingerprint::{Fingerprint, FingerprintEncoder};
 use rustc_data_structures::fx::{FxHashMap, FxIndexSet};
 use rustc_data_structures::stable_hasher::StableHasher;
 use rustc_data_structures::sync::{join, par_iter, Lrc, ParallelIterator};
@@ -116,6 +115,7 @@ fn emit_unit(&mut self) -> Result<(), Self::Error> {
         emit_f32(f32);
         emit_char(char);
         emit_str(&str);
+        emit_raw_bytes(&[u8]);
     }
 }
 
@@ -307,12 +307,6 @@ fn encode(&self, s: &mut EncodeContext<'a, 'tcx>) -> opaque::EncodeResult {
     }
 }
 
-impl<'a, 'tcx> FingerprintEncoder for EncodeContext<'a, 'tcx> {
-    fn encode_fingerprint(&mut self, f: &Fingerprint) -> Result<(), Self::Error> {
-        self.opaque.encode_fingerprint(f)
-    }
-}
-
 impl<'a, 'tcx> TyEncoder<'tcx> for EncodeContext<'a, 'tcx> {
     const CLEAR_CROSS_CRATE: bool = true;
 
@@ -2064,10 +2058,10 @@ pub(super) fn encode_metadata(tcx: TyCtxt<'_>) -> EncodedMetadata {
 
 fn encode_metadata_impl(tcx: TyCtxt<'_>) -> EncodedMetadata {
     let mut encoder = opaque::Encoder::new(vec![]);
-    encoder.emit_raw_bytes(METADATA_HEADER);
+    encoder.emit_raw_bytes(METADATA_HEADER).unwrap();
 
     // Will be filled with the root position after encoding everything.
-    encoder.emit_raw_bytes(&[0, 0, 0, 0]);
+    encoder.emit_raw_bytes(&[0, 0, 0, 0]).unwrap();
 
     let source_map_files = tcx.sess.source_map().files();
     let source_file_cache = (source_map_files[0].clone(), 0);

compiler/rustc_metadata/src/rmeta/table.rs

@@ -2,6 +2,7 @@
 
 use rustc_index::vec::Idx;
 use rustc_serialize::opaque::Encoder;
+use rustc_serialize::Encoder as _;
 use std::convert::TryInto;
 use std::marker::PhantomData;
 use std::num::NonZeroUsize;
@@ -172,7 +173,7 @@ pub(crate) fn set(&mut self, i: I, value: T) {
 
     pub(crate) fn encode(&self, buf: &mut Encoder) -> Lazy<Table<I, T>> {
         let pos = buf.position();
-        buf.emit_raw_bytes(&self.bytes);
+        buf.emit_raw_bytes(&self.bytes).unwrap();
         Lazy::from_position_and_meta(NonZeroUsize::new(pos as usize).unwrap(), self.bytes.len())
     }
 }

compiler/rustc_middle/src/ty/codec.rs

@@ -472,6 +472,11 @@ impl<$($typaram ),*> Decoder for $DecoderName<$($typaram),*> {
                     read_str -> Cow<'_, str>;
                 }
 
+                #[inline]
+                fn read_raw_bytes(&mut self, bytes: &mut [std::mem::MaybeUninit<u8>]) -> Result<(), Self::Error> {
+                    self.opaque.read_raw_bytes(bytes)
+                }
+
                 fn error(&mut self, err: &str) -> Self::Error {
                     self.opaque.error(err)
                 }

compiler/rustc_middle/src/ty/query/on_disk_cache.rs

@@ -4,7 +4,6 @@
 use crate::ty::codec::{RefDecodable, TyDecoder, TyEncoder};
 use crate::ty::context::TyCtxt;
 use crate::ty::{self, Ty};
-use rustc_data_structures::fingerprint::{Fingerprint, FingerprintDecoder, FingerprintEncoder};
 use rustc_data_structures::fx::{FxHashMap, FxHashSet, FxIndexSet};
 use rustc_data_structures::sync::{HashMapExt, Lock, Lrc, OnceCell};
 use rustc_data_structures::thin_vec::ThinVec;
@@ -17,7 +16,7 @@
 use rustc_query_system::dep_graph::DepContext;
 use rustc_query_system::query::QueryContext;
 use rustc_serialize::{
-    opaque::{self, FileEncodeResult, FileEncoder},
+    opaque::{self, FileEncodeResult, FileEncoder, IntEncodedWithFixedSize},
     Decodable, Decoder, Encodable, Encoder,
 };
 use rustc_session::{CrateDisambiguator, Session};
@@ -913,12 +912,6 @@ fn decode(d: &mut CacheDecoder<'a, 'tcx>) -> Result<Self, String> {
     }
 }
 
-impl<'a, 'tcx> FingerprintDecoder for CacheDecoder<'a, 'tcx> {
-    fn decode_fingerprint(&mut self) -> Result<Fingerprint, Self::Error> {
-        Fingerprint::decode_opaque(&mut self.opaque)
-    }
-}
-
 impl<'a, 'tcx> Decodable<CacheDecoder<'a, 'tcx>> for &'tcx FxHashSet<LocalDefId> {
     fn decode(d: &mut CacheDecoder<'a, 'tcx>) -> Result<Self, String> {
         RefDecodable::decode(d)
@@ -1011,12 +1004,6 @@ fn encode_tagged<T: Encodable<Self>, V: Encodable<Self>>(
     }
 }
 
-impl<'a, 'tcx, E: OpaqueEncoder> FingerprintEncoder for CacheEncoder<'a, 'tcx, E> {
-    fn encode_fingerprint(&mut self, f: &Fingerprint) -> Result<(), E::Error> {
-        self.encoder.encode_fingerprint(f)
-    }
-}
-
 impl<'a, 'tcx, E> Encodable<CacheEncoder<'a, 'tcx, E>> for SyntaxContext
 where
     E: 'a + OpaqueEncoder,
@@ -1167,6 +1154,7 @@ fn emit_unit(&mut self) -> Result<(), Self::Error> {
         emit_f32(f32);
         emit_char(char);
         emit_str(&str);
+        emit_raw_bytes(&[u8]);
     }
 }
 
@@ -1180,42 +1168,6 @@ fn encode(&self, e: &mut CacheEncoder<'a, 'tcx, FileEncoder>) -> FileEncodeResul
     }
 }
 
-// An integer that will always encode to 8 bytes.
-struct IntEncodedWithFixedSize(u64);
-
-impl IntEncodedWithFixedSize {
-    pub const ENCODED_SIZE: usize = 8;
-}
-
-impl<E: OpaqueEncoder> Encodable<E> for IntEncodedWithFixedSize {
-    fn encode(&self, e: &mut E) -> Result<(), E::Error> {
-        let start_pos = e.position();
-        for i in 0..IntEncodedWithFixedSize::ENCODED_SIZE {
-            ((self.0 >> (i * 8)) as u8).encode(e)?;
-        }
-        let end_pos = e.position();
-        assert_eq!((end_pos - start_pos), IntEncodedWithFixedSize::ENCODED_SIZE);
-        Ok(())
-    }
-}
-
-impl<'a> Decodable<opaque::Decoder<'a>> for IntEncodedWithFixedSize {
-    fn decode(decoder: &mut opaque::Decoder<'a>) -> Result<IntEncodedWithFixedSize, String> {
-        let mut value: u64 = 0;
-        let start_pos = decoder.position();
-
-        for i in 0..IntEncodedWithFixedSize::ENCODED_SIZE {
-            let byte: u8 = Decodable::decode(decoder)?;
-            value |= (byte as u64) << (i * 8);
-        }
-
-        let end_pos = decoder.position();
-        assert_eq!((end_pos - start_pos), IntEncodedWithFixedSize::ENCODED_SIZE);
-
-        Ok(IntEncodedWithFixedSize(value))
-    }
-}
-
 pub fn encode_query_results<'a, 'tcx, CTX, Q>(
     tcx: CTX,
     encoder: &mut CacheEncoder<'a, 'tcx, FileEncoder>,

compiler/rustc_serialize/src/json.rs

@@ -188,6 +188,7 @@
 use std::io;
 use std::io::prelude::*;
 use std::mem::swap;
+use std::mem::MaybeUninit;
 use std::num::FpCategory as Fp;
 use std::ops::Index;
 use std::str::FromStr;
@@ -553,6 +554,12 @@ fn emit_char(&mut self, v: char) -> EncodeResult {
     fn emit_str(&mut self, v: &str) -> EncodeResult {
         escape_str(self.writer, v)
     }
+    fn emit_raw_bytes(&mut self, s: &[u8]) -> Result<(), Self::Error> {
+        for &c in s.iter() {
+            self.emit_u8(c)?;
+        }
+        Ok(())
+    }
 
     fn emit_enum<F>(&mut self, _name: &str, f: F) -> EncodeResult
     where
@@ -879,6 +886,12 @@ fn emit_char(&mut self, v: char) -> EncodeResult {
     fn emit_str(&mut self, v: &str) -> EncodeResult {
         escape_str(self.writer, v)
     }
+    fn emit_raw_bytes(&mut self, s: &[u8]) -> Result<(), Self::Error> {
+        for &c in s.iter() {
+            self.emit_u8(c)?;
+        }
+        Ok(())
+    }
 
     fn emit_enum<F>(&mut self, _name: &str, f: F) -> EncodeResult
     where
@@ -2354,6 +2367,14 @@ fn read_str(&mut self) -> DecodeResult<Cow<'_, str>> {
         expect!(self.pop(), String).map(Cow::Owned)
     }
 
+    fn read_raw_bytes(&mut self, s: &mut [MaybeUninit<u8>]) -> Result<(), Self::Error> {
+        for c in s.iter_mut() {
+            let h = self.read_u8()?;
+            unsafe { *c.as_mut_ptr() = h };
+        }
+        Ok(())
+    }
+
     fn read_enum<T, F>(&mut self, _name: &str, f: F) -> DecodeResult<T>
     where
         F: FnOnce(&mut Decoder) -> DecodeResult<T>,

compiler/rustc_serialize/src/lib.rs

@@ -17,6 +17,8 @@
 #![feature(vec_spare_capacity)]
 #![feature(core_intrinsics)]
 #![feature(int_bits_const)]
+#![feature(maybe_uninit_array_assume_init)]
+#![feature(maybe_uninit_uninit_array)]
 #![feature(maybe_uninit_slice)]
 #![feature(new_uninit)]
 #![cfg_attr(test, feature(test))]

compiler/rustc_serialize/src/opaque.rs

@@ -1,5 +1,5 @@
 use crate::leb128::{self, max_leb128_len};
-use crate::serialize;
+use crate::serialize::{self, Decoder as _, Encoder as _};
 use std::borrow::Cow;
 use std::fs::File;
 use std::io::{self, Write};
@@ -30,11 +30,6 @@ pub fn into_inner(self) -> Vec<u8> {
     pub fn position(&self) -> usize {
         self.data.len()
     }
-
-    #[inline]
-    pub fn emit_raw_bytes(&mut self, s: &[u8]) {
-        self.data.extend_from_slice(s);
-    }
 }
 
 macro_rules! write_leb128 {
@@ -154,7 +149,12 @@ fn emit_char(&mut self, v: char) -> EncodeResult {
     #[inline]
     fn emit_str(&mut self, v: &str) -> EncodeResult {
         self.emit_usize(v.len())?;
-        self.emit_raw_bytes(v.as_bytes());
+        self.emit_raw_bytes(v.as_bytes())
+    }
+
+    #[inline]
+    fn emit_raw_bytes(&mut self, s: &[u8]) -> EncodeResult {
+        self.data.extend_from_slice(s);
         Ok(())
     }
 }
@@ -208,11 +208,6 @@ pub fn position(&self) -> usize {
         self.flushed + self.buffered
     }
 
-    #[inline]
-    pub fn emit_raw_bytes(&mut self, s: &[u8]) -> FileEncodeResult {
-        self.write_all(s)
-    }
-
     pub fn flush(&mut self) -> FileEncodeResult {
         // This is basically a copy of `BufWriter::flush`. If `BufWriter` ever
         // offers a raw buffer access API, we can use it, and remove this.
@@ -508,6 +503,11 @@ fn emit_str(&mut self, v: &str) -> FileEncodeResult {
         self.emit_usize(v.len())?;
         self.emit_raw_bytes(v.as_bytes())
     }
+
+    #[inline]
+    fn emit_raw_bytes(&mut self, s: &[u8]) -> FileEncodeResult {
+        self.write_all(s)
+    }
 }
 
 // -----------------------------------------------------------------------------
@@ -539,26 +539,6 @@ pub fn set_position(&mut self, pos: usize) {
     pub fn advance(&mut self, bytes: usize) {
         self.position += bytes;
     }
-
-    #[inline]
-    pub fn read_raw_bytes(&mut self, s: &mut [MaybeUninit<u8>]) -> Result<(), String> {
-        let start = self.position;
-        let end = start + s.len();
-        assert!(end <= self.data.len());
-
-        // SAFETY: Both `src` and `dst` point to at least `s.len()` elements:
-        // `src` points to at least `s.len()` elements by above assert, and
-        // `dst` points to `s.len()` elements by derivation from `s`.
-        unsafe {
-            let src = self.data.as_ptr().add(start);
-            let dst = s.as_mut_ptr() as *mut u8;
-            ptr::copy_nonoverlapping(src, dst, s.len());
-        }
-
-        self.position = end;
-
-        Ok(())
-    }
 }
 
 macro_rules! read_leb128 {
@@ -677,6 +657,26 @@ fn read_str(&mut self) -> Result<Cow<'_, str>, Self::Error> {
     fn error(&mut self, err: &str) -> Self::Error {
         err.to_string()
     }
+
+    #[inline]
+    fn read_raw_bytes(&mut self, s: &mut [MaybeUninit<u8>]) -> Result<(), String> {
+        let start = self.position;
+        let end = start + s.len();
+        assert!(end <= self.data.len());
+
+        // SAFETY: Both `src` and `dst` point to at least `s.len()` elements:
+        // `src` points to at least `s.len()` elements by above assert, and
+        // `dst` points to `s.len()` elements by derivation from `s`.
+        unsafe {
+            let src = self.data.as_ptr().add(start);
+            let dst = s.as_mut_ptr() as *mut u8;
+            ptr::copy_nonoverlapping(src, dst, s.len());
+        }
+
+        self.position = end;
+
+        Ok(())
+    }
 }
 
 // Specializations for contiguous byte sequences follow. The default implementations for slices
@@ -689,8 +689,7 @@ fn error(&mut self, err: &str) -> Self::Error {
 impl serialize::Encodable<Encoder> for [u8] {
     fn encode(&self, e: &mut Encoder) -> EncodeResult {
         serialize::Encoder::emit_usize(e, self.len())?;
-        e.emit_raw_bytes(self);
+        e.emit_raw_bytes(self)
-        Ok(())
     }
 }
 
@@ -718,3 +717,46 @@ fn decode(d: &mut Decoder<'a>) -> Result<Self, String> {
         Ok(v)
     }
 }
+
+// An integer that will always encode to 8 bytes.
+pub struct IntEncodedWithFixedSize(pub u64);
+
+impl IntEncodedWithFixedSize {
+    pub const ENCODED_SIZE: usize = 8;
+}
+
+impl serialize::Encodable<Encoder> for IntEncodedWithFixedSize {
+    #[inline]
+    fn encode(&self, e: &mut Encoder) -> EncodeResult {
+        let _start_pos = e.position();
+        e.emit_raw_bytes(&self.0.to_le_bytes())?;
+        let _end_pos = e.position();
+        debug_assert_eq!((_end_pos - _start_pos), IntEncodedWithFixedSize::ENCODED_SIZE);
+        Ok(())
+    }
+}
+
+impl serialize::Encodable<FileEncoder> for IntEncodedWithFixedSize {
+    #[inline]
+    fn encode(&self, e: &mut FileEncoder) -> FileEncodeResult {
+        let _start_pos = e.position();
+        e.emit_raw_bytes(&self.0.to_le_bytes())?;
+        let _end_pos = e.position();
+        debug_assert_eq!((_end_pos - _start_pos), IntEncodedWithFixedSize::ENCODED_SIZE);
+        Ok(())
+    }
+}
+
+impl<'a> serialize::Decodable<Decoder<'a>> for IntEncodedWithFixedSize {
+    #[inline]
+    fn decode(decoder: &mut Decoder<'a>) -> Result<IntEncodedWithFixedSize, String> {
+        let mut bytes = MaybeUninit::uninit_array();
+        let _start_pos = decoder.position();
+        decoder.read_raw_bytes(&mut bytes)?;
+        let _end_pos = decoder.position();
+        debug_assert_eq!((_end_pos - _start_pos), IntEncodedWithFixedSize::ENCODED_SIZE);
+
+        let value = u64::from_le_bytes(unsafe { MaybeUninit::array_assume_init(bytes) });
+        Ok(IntEncodedWithFixedSize(value))
+    }
+}

compiler/rustc_serialize/src/serialize.rs

@@ -7,6 +7,7 @@
 use std::borrow::Cow;
 use std::cell::{Cell, RefCell};
 use std::marker::PhantomData;
+use std::mem::MaybeUninit;
 use std::path;
 use std::rc::Rc;
 use std::sync::Arc;
@@ -33,6 +34,7 @@ pub trait Encoder {
     fn emit_f32(&mut self, v: f32) -> Result<(), Self::Error>;
     fn emit_char(&mut self, v: char) -> Result<(), Self::Error>;
     fn emit_str(&mut self, v: &str) -> Result<(), Self::Error>;
+    fn emit_raw_bytes(&mut self, s: &[u8]) -> Result<(), Self::Error>;
 
     // Compound types:
     #[inline]
@@ -224,6 +226,7 @@ pub trait Decoder {
     fn read_f32(&mut self) -> Result<f32, Self::Error>;
     fn read_char(&mut self) -> Result<char, Self::Error>;
     fn read_str(&mut self) -> Result<Cow<'_, str>, Self::Error>;
+    fn read_raw_bytes(&mut self, s: &mut [MaybeUninit<u8>]) -> Result<(), Self::Error>;
 
     // Compound types:
     #[inline]