optimization: add single byte writes

Single byte writes are used very often (used for every (u)int8_t or std::byte or bool and for every VarInt's first byte which is also needed for every (pre)Vector).
It makes sense to avoid the generalized serialization infrastructure that isn't needed:
* AutoFile write doesn't need to allocate 4k buffer for a single byte now;
* `VectorWriter` and `DataStream` avoids memcpy/insert calls.

> cmake -B build -DBUILD_BENCH=ON -DCMAKE_BUILD_TYPE=Release && cmake --build build -j$(nproc) && build/bin/bench_bitcoin -filter='SizeComputerBlock|SerializeBlock' --min-time=10000

> C compiler ............................ AppleClang 16.0.0.16000026

|            ns/block |             block/s |    err% |     total | benchmark
|--------------------:|--------------------:|--------:|----------:|:----------
|          174,569.19 |            5,728.39 |    0.6% |     10.89 | `SerializeBlock`
|           10,241.16 |           97,645.21 |    0.0% |     11.00 | `SizeComputerBlock`

> C++ compiler .......................... GNU 13.3.0

|            ns/block |             block/s |    err% |       ins/block |       cyc/block |    IPC |      bra/block |   miss% |     total | benchmark
|--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:----------
|          615,000.56 |            1,626.01 |    0.0% |    8,015,883.64 |    2,208,340.88 |  3.630 |   1,517,035.62 |    0.5% |     10.56 | `SerializeBlock`
|           25,676.76 |           38,945.72 |    0.0% |      159,390.03 |       92,202.10 |  1.729 |      42,131.03 |    0.9% |     11.00 | `SizeComputerBlock`

src/bench/checkblock.cpp

@@ -52,7 +52,7 @@ static void DeserializeBlock(benchmark::Bench& bench)
 {
     DataStream stream(benchmark::data::block413567);
     std::byte a{0};
-    stream.write({&a, 1}); // Prevent compaction
+    stream.write(std::span{&a, 1}); // Prevent compaction
 
     bench.unit("block").run([&] {
         CBlock block;
@@ -66,7 +66,7 @@ static void DeserializeAndCheckBlock(benchmark::Bench& bench)
 {
     DataStream stream(benchmark::data::block413567);
     std::byte a{0};
-    stream.write({&a, 1}); // Prevent compaction
+    stream.write(std::span{&a, 1}); // Prevent compaction
 
     ArgsManager bench_args;
     const auto chainParams = CreateChainParams(bench_args, ChainType::MAIN);

src/bench/rpc_blockchain.cpp

@@ -33,7 +33,7 @@ struct TestBlockAndIndex {
     {
         DataStream stream{benchmark::data::block413567};
         std::byte a{0};
-        stream.write({&a, 1}); // Prevent compaction
+        stream.write(std::span{&a, 1}); // Prevent compaction
 
         stream >> TX_WITH_WITNESS(block);
 

src/crypto/sha256.cpp

@@ -723,6 +723,21 @@ CSHA256& CSHA256::Write(const unsigned char* data, size_t len)
     }
     return *this;
 }
+CSHA256& CSHA256::Write(unsigned char data)
+{
+    size_t bufsize = bytes % 64;
+
+    // Add the single byte to the buffer
+    buf[bufsize] = data;
+    bytes += 1;
+
+    if (bufsize == 63) {
+        // Process the buffer if full
+        Transform(s, buf, 1);
+    }
+
+    return *this;
+}
 
 void CSHA256::Finalize(unsigned char hash[OUTPUT_SIZE])
 {

src/crypto/sha256.h

@@ -22,6 +22,7 @@ public:
 
     CSHA256();
     CSHA256& Write(const unsigned char* data, size_t len);
+    CSHA256& Write(unsigned char data);
     void Finalize(unsigned char hash[OUTPUT_SIZE]);
     CSHA256& Reset();
 };

src/hash.h

@@ -38,6 +38,10 @@ public:
         sha.Write(input.data(), input.size());
         return *this;
     }
+    CHash256& Write(std::span<const unsigned char, 1> input) {
+        sha.Write(input[0]);
+        return *this;
+    }
 
     CHash256& Reset() {
         sha.Reset();
@@ -63,6 +67,10 @@ public:
         sha.Write(input.data(), input.size());
         return *this;
     }
+    CHash160& Write(std::span<const unsigned char, 1> input) {
+        sha.Write(input[0]);
+        return *this;
+    }
 
     CHash160& Reset() {
         sha.Reset();
@@ -107,6 +115,10 @@ public:
     {
         ctx.Write(UCharCast(src.data()), src.size());
     }
+    void write(std::span<const std::byte, 1> src)
+    {
+        ctx.Write(*UCharCast(&src[0]));
+    }
 
     /** Compute the double-SHA256 hash of all data written to this object.
      *
@@ -160,13 +172,18 @@ public:
         m_source.read(dst);
         this->write(dst);
     }
+    void read(std::span<std::byte, 1> dst)
+    {
+        m_source.read(dst);
+        this->write(std::span<const std::byte, 1>{dst});
+    }
 
     void ignore(size_t num_bytes)
     {
         std::byte data[1024];
         while (num_bytes > 0) {
             size_t now = std::min<size_t>(num_bytes, 1024);
-            read({data, now});
+            read(std::span{data, now});
             num_bytes -= now;
         }
     }
@@ -194,6 +211,11 @@ public:
         m_source.write(src);
         HashWriter::write(src);
     }
+    void write(std::span<const std::byte, 1> src)
+    {
+        m_source.write(src);
+        HashWriter::write(src);
+    }
 
     template <typename T>
     HashedSourceWriter& operator<<(const T& obj)

src/serialize.h

@@ -1112,6 +1112,10 @@ public:
     {
         this->nSize += src.size();
     }
+    void write(std::span<const std::byte, 1>)
+    {
+        this->nSize += 1;
+    }
 
     /** Pretend _nSize bytes are written, without specifying them. */
     void seek(size_t _nSize)
@@ -1161,7 +1165,9 @@ public:
     template <typename U> ParamsStream& operator<<(const U& obj) { ::Serialize(*this, obj); return *this; }
     template <typename U> ParamsStream& operator>>(U&& obj) { ::Unserialize(*this, obj); return *this; }
     void write(std::span<const std::byte> src) { GetStream().write(src); }
+    void write(std::span<const std::byte, 1> src) { GetStream().write(src); }
     void read(std::span<std::byte> dst) { GetStream().read(dst); }
+    void read(std::span<std::byte, 1> dst) { GetStream().read(dst); }
     void ignore(size_t num) { GetStream().ignore(num); }
     bool eof() const { return GetStream().eof(); }
     size_t size() const { return GetStream().size(); }

src/streams.cpp

@@ -64,6 +64,13 @@ void AutoFile::read(std::span<std::byte> dst)
     }
 }
 
+void AutoFile::read(std::span<std::byte, 1> dst)
+{
+    if (detail_fread(dst) != 1) {
+        throw std::ios_base::failure(feof() ? "AutoFile::read: end of file" : "AutoFile::read: fread failed");
+    }
+}
+
 void AutoFile::ignore(size_t nSize)
 {
     if (!m_file) throw std::ios_base::failure("AutoFile::ignore: file handle is nullptr");
@@ -97,6 +104,12 @@ void AutoFile::write(std::span<const std::byte> src)
     }
 }
 
+void AutoFile::write(std::span<const std::byte, 1> src)
+{
+    std::byte temp_byte = src[0];
+    write_buffer(std::span(&temp_byte, 1));
+}
+
 void AutoFile::write_buffer(std::span<std::byte> src)
 {
     if (!m_file) throw std::ios_base::failure("AutoFile::write_buffer: file handle is nullptr");

src/streams.h

@@ -83,6 +83,17 @@ public:
         }
         nPos += src.size();
     }
+    void write(std::span<const std::byte, 1> src)
+    {
+        assert(nPos <= vchData.size());
+        const auto byte{*UCharCast(&src[0])};
+        if (nPos < vchData.size()) {
+            vchData[nPos] = byte;
+        } else {
+            vchData.push_back(byte);
+        }
+        nPos += 1;
+    }
     template <typename T>
     VectorWriter& operator<<(const T& obj)
     {
@@ -254,6 +265,10 @@ public:
         // Write to the end of the buffer
         vch.insert(vch.end(), src.begin(), src.end());
     }
+    void write(std::span<const value_type, 1> src)
+    {
+        vch.push_back(src[0]);
+    }
 
     template<typename T>
     DataStream& operator<<(const T& obj)
@@ -453,8 +468,10 @@ public:
     // Stream subset
     //
     void read(std::span<std::byte> dst);
+    void read(std::span<std::byte, 1> dst);
     void ignore(size_t nSize);
     void write(std::span<const std::byte> src);
+    void write(std::span<const std::byte, 1> src);
 
     template <typename T>
     AutoFile& operator<<(const T& obj)

src/test/crypto_tests.cpp

@@ -1079,7 +1079,7 @@ BOOST_AUTO_TEST_CASE(sha256d64)
             in[j] = m_rng.randbits(8);
         }
         for (int j = 0; j < i; ++j) {
-            CHash256().Write({in + 64 * j, 64}).Finalize({out1 + 32 * j, 32});
+            CHash256().Write(std::span{in + 64 * j, 64}).Finalize({out1 + 32 * j, 32});
         }
         SHA256D64(out2, in, i);
         BOOST_CHECK(memcmp(out1, out2, 32 * i) == 0);

src/test/fuzz/autofile.cpp

@@ -29,14 +29,14 @@ FUZZ_TARGET(autofile)
             [&] {
                 std::array<std::byte, 4096> arr{};
                 try {
-                    auto_file.read({arr.data(), fuzzed_data_provider.ConsumeIntegralInRange<size_t>(0, 4096)});
+                    auto_file.read(std::span{arr.data(), fuzzed_data_provider.ConsumeIntegralInRange<size_t>(0, 4096)});
                 } catch (const std::ios_base::failure&) {
                 }
             },
             [&] {
                 const std::array<std::byte, 4096> arr{};
                 try {
-                    auto_file.write({arr.data(), fuzzed_data_provider.ConsumeIntegralInRange<size_t>(0, 4096)});
+                    auto_file.write(std::span{arr.data(), fuzzed_data_provider.ConsumeIntegralInRange<size_t>(0, 4096)});
                 } catch (const std::ios_base::failure&) {
                 }
             },

src/test/streams_tests.cpp

@@ -26,9 +26,9 @@ BOOST_AUTO_TEST_CASE(xor_file)
     {
         // Check errors for missing file
         AutoFile xor_file{raw_file("rb"), xor_pat};
-        BOOST_CHECK_EXCEPTION(xor_file << std::byte{}, std::ios_base::failure, HasReason{"AutoFile::write: file handle is nullpt"});
-        BOOST_CHECK_EXCEPTION(xor_file >> std::byte{}, std::ios_base::failure, HasReason{"AutoFile::read: file handle is nullpt"});
-        BOOST_CHECK_EXCEPTION(xor_file.ignore(1), std::ios_base::failure, HasReason{"AutoFile::ignore: file handle is nullpt"});
+        BOOST_CHECK_EXCEPTION(xor_file << std::byte{}, std::ios_base::failure, HasReason{"file handle is nullpt"});
+        BOOST_CHECK_EXCEPTION(xor_file >> std::byte{}, std::ios_base::failure, HasReason{"file handle is nullpt"});
+        BOOST_CHECK_EXCEPTION(xor_file.ignore(1), std::ios_base::failure, HasReason{"file handle is nullpt"});
     }
     {
 #ifdef __MINGW64__