From 08e4497c6e3b3f434c0bb255c3942648f153fe66 Mon Sep 17 00:00:00 2001
From: Lee Clagett <code@leeclagett.com>
Date: Sun, 24 Jan 2021 07:42:57 +0000
Subject: Improve cryptonote (block and tx) binary read performance

---
 src/serialization/binary_archive.h  | 90 +++++++++++++++++++------------------
 src/serialization/binary_utils.h    |  3 +-
 src/serialization/container.h       | 10 ++---
 src/serialization/crypto.h          |  6 +--
 src/serialization/debug_archive.h   |  1 +
 src/serialization/difficulty_type.h |  8 ++--
 src/serialization/json_archive.h    |  7 ++-
 src/serialization/pair.h            | 12 ++---
 src/serialization/serialization.h   | 36 +++++++--------
 src/serialization/string.h          |  2 +-
 src/serialization/variant.h         |  8 ++--
 11 files changed, 94 insertions(+), 89 deletions(-)

(limited to 'src/serialization')

diff --git a/src/serialization/binary_archive.h b/src/serialization/binary_archive.h
index 49ca8aa57..acda70039 100644
--- a/src/serialization/binary_archive.h
+++ b/src/serialization/binary_archive.h
@@ -36,9 +36,11 @@
 #include <cassert>
 #include <iostream>
 #include <iterator>
+#include <boost/endian/conversion.hpp>
 #include <boost/type_traits/make_unsigned.hpp>
 
 #include "common/varint.h"
+#include "span.h"
 #include "warnings.h"
 
 /* I have no clue what these lines means */
@@ -55,16 +57,15 @@ DISABLE_VS_WARNINGS(4244)
  * purpse is to define the functions used for the binary_archive. Its
  * a header, basically. I think it was declared simply to save typing...
  */
-template <class Stream, bool IsSaving>
+template <bool IsSaving>
 struct binary_archive_base
 {
-  typedef Stream stream_type;
-  typedef binary_archive_base<Stream, IsSaving> base_type;
+  typedef binary_archive_base<IsSaving> base_type;
   typedef boost::mpl::bool_<IsSaving> is_saving;
 
   typedef uint8_t variant_tag_type;
 
-  explicit binary_archive_base(stream_type &s) : stream_(s) { }
+  explicit binary_archive_base() { }
   
   /* definition of standard API functions */
   void tag(const char *) { }
@@ -72,12 +73,6 @@ struct binary_archive_base
   void end_object() { }
   void begin_variant() { }
   void end_variant() { }
-  /* I just want to leave a comment saying how this line really shows
-     flaws in the ownership model of many OOP languages, that is all. */
-  stream_type &stream() { return stream_; } 
-
-protected:
-  stream_type &stream_;
 };
 
 /* \struct binary_archive
@@ -95,15 +90,18 @@ struct binary_archive;
 
 
 template <>
-struct binary_archive<false> : public binary_archive_base<std::istream, false>
+struct binary_archive<false> : public binary_archive_base<false>
 {
+  explicit binary_archive(epee::span<const std::uint8_t> s)
+    : base_type(), bytes_(s), begin_(s.begin()), good_(true), varint_bug_backward_compatibility_(false)
+  {}
 
-  explicit binary_archive(stream_type &s) : base_type(s), varint_bug_backward_compatibility_(false) {
-    stream_type::pos_type pos = stream_.tellg();
-    stream_.seekg(0, std::ios_base::end);
-    eof_pos_ = stream_.tellg();
-    stream_.seekg(pos);
-  }
+  bool good() const noexcept { return good_; }
+  void set_fail() noexcept { good_ = false; }
+
+  //! If implementing as `std::istream`, reset stream error state after `peek()` call.
+  bool eof() const noexcept { return bytes_.empty(); }
+  std::size_t getpos() const noexcept { return bytes_.begin() - begin_; }
 
   template <class T>
   void serialize_int(T &v)
@@ -116,24 +114,24 @@ struct binary_archive<false> : public binary_archive_base<std::istream, false>
    * \brief serializes an unsigned integer
    */
   template <class T>
-  void serialize_uint(T &v, size_t width = sizeof(T))
+  void serialize_uint(T &v)
   {
-    T ret = 0;
-    unsigned shift = 0;
-    for (size_t i = 0; i < width; i++) {
-      //std::cerr << "tell: " << stream_.tellg() << " value: " << ret << std::endl;
-      char c;
-      stream_.get(c);
-      T b = (unsigned char)c;
-      ret += (b << shift);	// can this be changed to OR, i think it can.
-      shift += 8;
+    const std::size_t actual = bytes_.remove_prefix(sizeof(T));
+    good_ &= (actual == sizeof(T));
+    if (actual == sizeof(T))
+    {
+      std::memcpy(std::addressof(v), bytes_.data() - sizeof(T), sizeof(T));
+      boost::endian::little_to_native_inplace(v); // epee isn't templated
     }
-    v = ret;
+    else
+      v = 0; // ensures initialization
   }
   
   void serialize_blob(void *buf, size_t len, const char *delimiter="")
   {
-    stream_.read((char *)buf, len);
+    const std::size_t actual = bytes_.remove_prefix(len);
+    good_ &= (len == actual);
+    std::memcpy(buf, bytes_.data() - actual, actual);
   }
   
   template <class T>
@@ -145,9 +143,11 @@ struct binary_archive<false> : public binary_archive_base<std::istream, false>
   template <class T>
   void serialize_uvarint(T &v)
   {
-    typedef std::istreambuf_iterator<char> it;
-    if (tools::read_varint(it(stream_), it(), v) < 0)
-      stream_.setstate(std::ios_base::failbit);
+    auto current = bytes_.cbegin();
+    auto end = bytes_.cend();
+    good_ &= (0 <= tools::read_varint(current, end, v));
+    current = std::min(current, bytes_.cend());
+    bytes_ = {current, std::size_t(bytes_.cend() - current)};
   }
 
   void begin_array(size_t &s)
@@ -166,26 +166,26 @@ struct binary_archive<false> : public binary_archive_base<std::istream, false>
     serialize_int(t);
   }
 
-  size_t remaining_bytes() {
-    if (!stream_.good())
-      return 0;
-    //std::cerr << "tell: " << stream_.tellg() << std::endl;
-    assert(stream_.tellg() <= eof_pos_);
-    return eof_pos_ - stream_.tellg();
-  }
-
+  size_t remaining_bytes() const noexcept { return good() ? bytes_.size() : 0; }
   void enable_varint_bug_backward_compatibility() { varint_bug_backward_compatibility_ = true; }
   bool varint_bug_backward_compatibility_enabled() const { return varint_bug_backward_compatibility_; }
-
 protected:
-  std::streamoff eof_pos_;
+  epee::span<const std::uint8_t> bytes_;
+  std::uint8_t const* const begin_;
+  bool good_;
   bool varint_bug_backward_compatibility_;
 };
 
 template <>
-struct binary_archive<true> : public binary_archive_base<std::ostream, true>
+struct binary_archive<true> : public binary_archive_base<true>
 {
-  explicit binary_archive(stream_type &s) : base_type(s) { }
+  typedef std::ostream stream_type;
+  explicit binary_archive(stream_type &s) : base_type(), stream_(s) { }
+
+  bool good() const { return stream_.good(); }
+  void set_fail() { stream_.setstate(std::ios::failbit); }
+
+  std::streampos getpos() const { return stream_.tellp(); }
 
   template <class T>
   void serialize_int(T v)
@@ -234,6 +234,8 @@ struct binary_archive<true> : public binary_archive_base<std::ostream, true>
   }
 
   bool varint_bug_backward_compatibility_enabled() const { return false; }
+protected:
+  stream_type& stream_;
 };
 
 POP_WARNINGS
diff --git a/src/serialization/binary_utils.h b/src/serialization/binary_utils.h
index 8635585af..8444f220e 100644
--- a/src/serialization/binary_utils.h
+++ b/src/serialization/binary_utils.h
@@ -39,8 +39,7 @@ namespace serialization {
   template <class T>
     bool parse_binary(const std::string &blob, T &v)
     {
-      std::istringstream istr(blob);
-      binary_archive<false> iar(istr);
+      binary_archive<false> iar{epee::strspan<std::uint8_t>(blob)};
       return ::serialization::serialize(iar, v);
     }
 
diff --git a/src/serialization/container.h b/src/serialization/container.h
index a4997c8ae..77681ec93 100644
--- a/src/serialization/container.h
+++ b/src/serialization/container.h
@@ -67,13 +67,13 @@ bool do_serialize_container(Archive<false> &ar, C &v)
 {
   size_t cnt;
   ar.begin_array(cnt);
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   v.clear();
 
   // very basic sanity check
   if (ar.remaining_bytes() < cnt) {
-    ar.stream().setstate(std::ios::failbit);
+    ar.set_fail();
     return false;
   }
 
@@ -86,7 +86,7 @@ bool do_serialize_container(Archive<false> &ar, C &v)
     if (!::serialization::detail::serialize_container_element(ar, e))
       return false;
     ::serialization::detail::do_add(v, std::move(e));
-    if (!ar.stream().good())
+    if (!ar.good())
       return false;
   }
   ar.end_array();
@@ -100,13 +100,13 @@ bool do_serialize_container(Archive<true> &ar, C &v)
   ar.begin_array(cnt);
   for (auto i = v.begin(); i != v.end(); ++i)
   {
-    if (!ar.stream().good())
+    if (!ar.good())
       return false;
     if (i != v.begin())
       ar.delimit_array();
     if(!::serialization::detail::serialize_container_element(ar, (typename C::value_type&)*i))
       return false;
-    if (!ar.stream().good())
+    if (!ar.good())
       return false;
   }
   ar.end_array();
diff --git a/src/serialization/crypto.h b/src/serialization/crypto.h
index 42bd06e92..d635a9ee9 100644
--- a/src/serialization/crypto.h
+++ b/src/serialization/crypto.h
@@ -47,7 +47,7 @@ bool do_serialize(Archive<false> &ar, std::vector<crypto::signature> &v)
 
   // very basic sanity check
   if (ar.remaining_bytes() < cnt*sizeof(crypto::signature)) {
-    ar.stream().setstate(std::ios::failbit);
+    ar.set_fail();
     return false;
   }
 
@@ -55,7 +55,7 @@ bool do_serialize(Archive<false> &ar, std::vector<crypto::signature> &v)
   for (size_t i = 0; i < cnt; i++) {
     v.resize(i+1);
     ar.serialize_blob(&(v[i]), sizeof(crypto::signature), "");
-    if (!ar.stream().good())
+    if (!ar.good())
       return false;
   }
   return true;
@@ -70,7 +70,7 @@ bool do_serialize(Archive<true> &ar, std::vector<crypto::signature> &v)
   size_t cnt = v.size();
   for (size_t i = 0; i < cnt; i++) {
     ar.serialize_blob(&(v[i]), sizeof(crypto::signature), "");
-    if (!ar.stream().good())
+    if (!ar.good())
       return false;
   }
   ar.end_string();
diff --git a/src/serialization/debug_archive.h b/src/serialization/debug_archive.h
index b04d6ae19..e8ccb9a58 100644
--- a/src/serialization/debug_archive.h
+++ b/src/serialization/debug_archive.h
@@ -38,6 +38,7 @@ struct debug_archive : public json_archive<W> {
   typedef typename json_archive<W>::stream_type stream_type;
 
   debug_archive(stream_type &s) : json_archive<W>(s) { }
+  stream_type& stream() { return this->stream_; }
 };
 
 template <class T>
diff --git a/src/serialization/difficulty_type.h b/src/serialization/difficulty_type.h
index 56c0312e7..75d1fd13f 100644
--- a/src/serialization/difficulty_type.h
+++ b/src/serialization/difficulty_type.h
@@ -38,10 +38,10 @@ inline bool do_serialize(Archive<false>& ar, cryptonote::difficulty_type &diff)
 {
   uint64_t hi, lo;
   ar.serialize_varint(hi);
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   ar.serialize_varint(lo);
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   diff = hi;
   diff <<= 64;
@@ -52,13 +52,13 @@ inline bool do_serialize(Archive<false>& ar, cryptonote::difficulty_type &diff)
 template <template <bool> class Archive>
 inline bool do_serialize(Archive<true>& ar, cryptonote::difficulty_type &diff)
 {
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   const uint64_t hi = ((diff >> 64) & 0xffffffffffffffff).convert_to<uint64_t>();
   const uint64_t lo = (diff & 0xffffffffffffffff).convert_to<uint64_t>();
   ar.serialize_varint(hi);
   ar.serialize_varint(lo);
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   return true;
 }
diff --git a/src/serialization/json_archive.h b/src/serialization/json_archive.h
index 3f98b5101..bab6dfa94 100644
--- a/src/serialization/json_archive.h
+++ b/src/serialization/json_archive.h
@@ -58,6 +58,10 @@ struct json_archive_base
   json_archive_base(stream_type &s, bool indent = false)
   : stream_(s), indent_(indent), object_begin(false), depth_(0) { }
 
+  bool good() const { return stream_.good(); }
+  void set_fail() { stream_.setstate(std::ios::failbit); }
+  void clear_fail() { stream_.clear(); }
+
   void tag(const char *tag) {
     if (!object_begin)
       stream_ << ", ";
@@ -82,7 +86,6 @@ struct json_archive_base
 
   void begin_variant() { begin_object(); }
   void end_variant() { end_object(); }
-  Stream &stream() { return stream_; }
 
   bool varint_bug_backward_compatibility_enabled() const { return false; }
 
@@ -117,6 +120,8 @@ struct json_archive<true> : public json_archive_base<std::ostream, true>
 {
   json_archive(stream_type &s, bool indent = false) : base_type(s, indent), inner_array_size_(0) { }
 
+  std::streampos getpos() const { return stream_.tellp(); }
+
   template<typename T>
   static auto promote_to_printable_integer_type(T v) -> decltype(+v)
   {
diff --git a/src/serialization/pair.h b/src/serialization/pair.h
index 44aafa04d..2d9a89242 100644
--- a/src/serialization/pair.h
+++ b/src/serialization/pair.h
@@ -69,19 +69,19 @@ inline bool do_serialize(Archive<false>& ar, std::pair<F,S>& p)
 {
   size_t cnt;
   ar.begin_array(cnt);
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   if (cnt != 2)
     return false;
 
   if (!::serialization::detail::serialize_pair_element(ar, p.first))
     return false;
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   ar.delimit_array();
   if (!::serialization::detail::serialize_pair_element(ar, p.second))
     return false;
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
 
   ar.end_array();
@@ -92,16 +92,16 @@ template <template <bool> class Archive, class F, class S>
 inline bool do_serialize(Archive<true>& ar, std::pair<F,S>& p)
 {
   ar.begin_array(2);
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   if(!::serialization::detail::serialize_pair_element(ar, p.first))
     return false;
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   ar.delimit_array();
   if(!::serialization::detail::serialize_pair_element(ar, p.second))
     return false;
-  if (!ar.stream().good())
+  if (!ar.good())
     return false;
   ar.end_array();
   return true;
diff --git a/src/serialization/serialization.h b/src/serialization/serialization.h
index 3ebeed171..4acadeab3 100644
--- a/src/serialization/serialization.h
+++ b/src/serialization/serialization.h
@@ -213,7 +213,7 @@ inline bool do_serialize(Archive &ar, bool &v)
  * \brief self-explanatory
  */
 #define END_SERIALIZE()				\
-  return ar.stream().good();			\
+  return ar.good();				\
   }
 
 /*! \macro VALUE(f)
@@ -223,7 +223,7 @@ inline bool do_serialize(Archive &ar, bool &v)
   do {							\
     ar.tag(#f);						\
     bool r = ::do_serialize(ar, f);			\
-    if (!r || !ar.stream().good()) return false;	\
+    if (!r || !ar.good()) return false;			\
   } while(0);
 
 /*! \macro FIELD_N(t,f)
@@ -234,7 +234,7 @@ inline bool do_serialize(Archive &ar, bool &v)
   do {							\
     ar.tag(t);						\
     bool r = ::do_serialize(ar, f);			\
-    if (!r || !ar.stream().good()) return false;	\
+    if (!r || !ar.good()) return false;			\
   } while(0);
 
 /*! \macro FIELD(f)
@@ -245,7 +245,7 @@ inline bool do_serialize(Archive &ar, bool &v)
   do {							\
     ar.tag(#f);						\
     bool r = ::do_serialize(ar, f);			\
-    if (!r || !ar.stream().good()) return false;	\
+    if (!r || !ar.good()) return false;			\
   } while(0);
 
 /*! \macro FIELDS(f)
@@ -255,7 +255,7 @@ inline bool do_serialize(Archive &ar, bool &v)
 #define FIELDS(f)							\
   do {									\
     bool r = ::do_serialize(ar, f);					\
-    if (!r || !ar.stream().good()) return false;			\
+    if (!r || !ar.good()) return false;					\
   } while(0);
 
 /*! \macro VARINT_FIELD(f)
@@ -265,7 +265,7 @@ inline bool do_serialize(Archive &ar, bool &v)
   do {						\
     ar.tag(#f);					\
     ar.serialize_varint(f);			\
-    if (!ar.stream().good()) return false;	\
+    if (!ar.good()) return false;		\
   } while(0);
 
 /*! \macro VARINT_FIELD_N(t, f)
@@ -276,7 +276,7 @@ inline bool do_serialize(Archive &ar, bool &v)
   do {						\
     ar.tag(t);					\
     ar.serialize_varint(f);			\
-    if (!ar.stream().good()) return false;	\
+    if (!ar.good()) return false;		\
   } while(0);
 
 /*! \macro MAGIC_FIELD(m)
@@ -286,7 +286,7 @@ inline bool do_serialize(Archive &ar, bool &v)
   do {						\
     ar.tag("magic");				\
     ar.serialize_blob((void*)magic.data(), magic.size()); \
-    if (!ar.stream().good()) return false;	\
+    if (!ar.good()) return false;		\
     if (magic != m) return false;		\
   } while(0);
 
@@ -297,7 +297,7 @@ inline bool do_serialize(Archive &ar, bool &v)
   do {						\
     ar.tag("version");				\
     ar.serialize_varint(version);		\
-    if (!ar.stream().good()) return false;	\
+    if (!ar.good()) return false;		\
   } while(0);
 
 
@@ -339,10 +339,10 @@ namespace serialization {
      *
      * \brief self explanatory
      */
-    template<class Stream>
-    bool do_check_stream_state(Stream& s, boost::mpl::bool_<true>, bool noeof)
+    template<class Archive>
+    bool do_check_stream_state(Archive& ar, boost::mpl::bool_<true>, bool noeof)
     {
-      return s.good();
+      return ar.good();
     }
     /*! \fn do_check_stream_state
      *
@@ -350,15 +350,13 @@ namespace serialization {
      *
      * \detailed Also checks to make sure that the stream is not at EOF
      */
-    template<class Stream>
-    bool do_check_stream_state(Stream& s, boost::mpl::bool_<false>, bool noeof)
+    template<class Archive>
+    bool do_check_stream_state(Archive& ar, boost::mpl::bool_<false>, bool noeof)
     {
       bool result = false;
-      if (s.good())
+      if (ar.good())
 	{
-	  std::ios_base::iostate state = s.rdstate();
-	  result = noeof || EOF == s.peek();
-	  s.clear(state);
+	  result = noeof || ar.eof();
 	}
       return result;
     }
@@ -371,7 +369,7 @@ namespace serialization {
   template<class Archive>
   bool check_stream_state(Archive& ar, bool noeof = false)
   {
-    return detail::do_check_stream_state(ar.stream(), typename Archive::is_saving(), noeof);
+    return detail::do_check_stream_state(ar, typename Archive::is_saving(), noeof);
   }
 
   /*! \fn serialize
diff --git a/src/serialization/string.h b/src/serialization/string.h
index f5f69833a..f1f8f4ab0 100644
--- a/src/serialization/string.h
+++ b/src/serialization/string.h
@@ -39,7 +39,7 @@ inline bool do_serialize(Archive<false>& ar, std::string& str)
   ar.serialize_varint(size);
   if (ar.remaining_bytes() < size)
   {
-    ar.stream().setstate(std::ios::failbit);
+    ar.set_fail();
     return false;
   }
 
diff --git a/src/serialization/variant.h b/src/serialization/variant.h
index b08cf6bf0..6debb63d1 100644
--- a/src/serialization/variant.h
+++ b/src/serialization/variant.h
@@ -74,7 +74,7 @@ struct variant_reader
       current_type x;
       if(!::do_serialize(ar, x))
       {
-        ar.stream().setstate(std::ios::failbit);
+        ar.set_fail();
         return false;
       }
       v = x;
@@ -95,7 +95,7 @@ struct variant_reader<Archive, Variant, TBegin, TBegin>
 
   static inline bool read(Archive &ar, Variant &v, variant_tag_type t)
   {
-    ar.stream().setstate(std::ios::failbit);
+    ar.set_fail();
     return false;
   }
 };
@@ -116,7 +116,7 @@ struct serializer<Archive<false>, boost::variant<BOOST_VARIANT_ENUM_PARAMS(T)>>
        typename boost::mpl::begin<types>::type,
        typename boost::mpl::end<types>::type>::read(ar, v, t))
     {
-      ar.stream().setstate(std::ios::failbit);
+      ar.set_fail();
       return false;
     }
     ar.end_variant();
@@ -143,7 +143,7 @@ struct serializer<Archive<true>, boost::variant<BOOST_VARIANT_ENUM_PARAMS(T)>>
       ar.write_variant_tag(variant_serialization_traits<Archive<true>, T>::get_tag());
       if(!::do_serialize(ar, rv))
       {
-        ar.stream().setstate(std::ios::failbit);
+        ar.set_fail();
         return false;
       }
       ar.end_variant();
-- 
cgit v1.2.3