Fix for 32 bit archs for encoded Arrays (#7427)

* tentative fix for 32 bit archs
* removed wrong cast to size_t from bf_iterator::set_value()
* fix inverted condition in 'unsigned_to_num_bits'
* fix inverted condition in 'unsigned_to_num_bits'

---------

Co-authored-by: Nicola Cabiddu <nicola.cabiddu@mongodb.com>

src/realm/array_direct.hpp

@@ -301,9 +301,9 @@ class bf_iterator {
     {
         auto in_word_position = field_position & 0x3F;
         auto first_word = first_word_ptr[0];
-        uint64_t mask = 0 - 1ULL;
+        uint64_t mask = 0ULL - 1ULL;
         if (field_size < 64) {
-            mask = static_cast<size_t>((1ULL << field_size) - 1);
+            mask = (1ULL << field_size) - 1;
             value &= mask;
         }
         // zero out field in first word:
@@ -389,16 +389,16 @@ inline void write_bitfield(uint64_t* data_area, size_t field_position, size_t wi
     *it = value;
 }
 
-inline int64_t sign_extend_field_by_mask(size_t sign_mask, uint64_t value)
+inline int64_t sign_extend_field_by_mask(uint64_t sign_mask, uint64_t value)
 {
-    uint64_t sign_extension = 0 - (value & sign_mask);
+    uint64_t sign_extension = 0ULL - (value & sign_mask);
     return value | sign_extension;
 }
 
 inline int64_t sign_extend_value(size_t width, uint64_t value)
 {
     uint64_t sign_mask = 1ULL << (width - 1);
-    uint64_t sign_extension = 0 - (value & sign_mask);
+    uint64_t sign_extension = 0ULL - (value & sign_mask);
     return value | sign_extension;
 }
 

src/realm/array_encode.hpp

@@ -40,7 +40,7 @@ class ArrayEncode {
     // init from mem B
     inline size_t size() const;
     inline size_t width() const;
-    inline size_t width_mask() const;
+    inline uint64_t width_mask() const;
     inline NodeHeader::Encoding get_encoding() const;
 
     // get/set
@@ -71,8 +71,7 @@ class ArrayEncode {
     using Encoding = NodeHeader::Encoding;
     Encoding m_encoding{NodeHeader::Encoding::WTypBits}; // this is not ok .... probably
     size_t m_v_width = 0, m_v_size = 0, m_ndx_width = 0, m_ndx_size = 0;
-    size_t m_v_mask = 0;
-    size_t m_ndx_mask = 0;
+    uint64_t m_v_mask = 0, m_ndx_mask = 0;
 
     friend class ArrayPacked;
     friend class ArrayFlex;
@@ -93,7 +92,7 @@ inline size_t ArrayEncode::width() const
     return m_v_width;
 }
 
-inline size_t ArrayEncode::width_mask() const
+inline uint64_t ArrayEncode::width_mask() const
 {
     return m_v_mask;
 }

src/realm/array_flex.cpp

@@ -102,13 +102,13 @@ int64_t ArrayFlex::get(const Array& arr, size_t ndx) const
 }
 
 int64_t ArrayFlex::get(const char* data, size_t ndx, size_t v_width, size_t v_size, size_t ndx_width, size_t ndx_size,
-                       size_t mask) const
+                       uint64_t mask) const
 {
     return do_get((uint64_t*)data, ndx, v_width, v_size, ndx_width, ndx_size, mask);
 }
 
 int64_t ArrayFlex::do_get(uint64_t* data, size_t ndx, size_t v_width, size_t ndx_width, size_t v_size,
-                          size_t ndx_size, size_t mask) const
+                          size_t ndx_size, uint64_t mask) const
 {
     if (ndx >= ndx_size)
         return realm::not_found;
@@ -176,7 +176,7 @@ bool ArrayFlex::find_all(const Array& arr, int64_t value, size_t start, size_t e
 }
 
 template <typename Cond, bool v>
-inline size_t ArrayFlex::parallel_subword_find(const Array& arr, uint64_t value, size_t width_mask, size_t offset,
+inline size_t ArrayFlex::parallel_subword_find(const Array& arr, uint64_t value, uint64_t width_mask, size_t offset,
                                                uint_least8_t width, size_t start, size_t end) const
 {
     const auto MSBs = populate(width, width_mask);

src/realm/array_flex.hpp

@@ -33,10 +33,9 @@ class ArrayFlex {
     // encoding/decoding
     void init_array(char* h, uint8_t flags, size_t v_width, size_t ndx_width, size_t v_size, size_t ndx_size) const;
     void copy_data(const Array&, const std::vector<int64_t>&, const std::vector<size_t>&) const;
-    std::vector<int64_t> fetch_all_values(const Array& h) const;
     // getters/setters
     int64_t get(const Array&, size_t) const;
-    int64_t get(const char*, size_t, size_t, size_t, size_t, size_t, size_t) const;
+    int64_t get(const char*, size_t, size_t, size_t, size_t, size_t, uint64_t) const;
     void get_chunk(const Array& h, size_t ndx, int64_t res[8]) const;
     void set_direct(const Array&, size_t, int64_t) const;
 
@@ -46,11 +45,11 @@ class ArrayFlex {
     int64_t sum(const Array&, size_t, size_t) const;
 
 private:
-    int64_t do_get(uint64_t*, size_t, size_t, size_t, size_t, size_t, size_t) const;
+    int64_t do_get(uint64_t*, size_t, size_t, size_t, size_t, size_t, uint64_t) const;
     bool find_all_match(size_t start, size_t end, size_t baseindex, QueryStateBase* state) const;
 
-    template <typename Cond, bool = true> // true int64_t other uint64_t
-    inline size_t parallel_subword_find(const Array&, uint64_t, size_t, size_t, uint_least8_t, size_t, size_t) const;
+    template <typename Cond, bool = true> // true int64_t, false uint64_t
+    inline size_t parallel_subword_find(const Array&, uint64_t, uint64_t, size_t, uint_least8_t, size_t, size_t) const;
 
     bool find_eq(const Array&, int64_t, size_t, size_t, size_t, QueryStateBase*) const;
     bool find_neq(const Array&, int64_t, size_t, size_t, size_t, QueryStateBase*) const;

src/realm/array_packed.cpp

@@ -81,12 +81,12 @@ int64_t ArrayPacked::get(const Array& arr, size_t ndx) const
     return do_get((uint64_t*)arr.m_data, ndx, w, sz, arr.get_encoder().width_mask());
 }
 
-int64_t ArrayPacked::get(const char* data, size_t ndx, size_t width, size_t sz, size_t mask) const
+int64_t ArrayPacked::get(const char* data, size_t ndx, size_t width, size_t sz, uint64_t mask) const
 {
     return do_get((uint64_t*)data, ndx, width, sz, mask);
 }
 
-int64_t ArrayPacked::do_get(uint64_t* data, size_t ndx, size_t v_width, size_t v_size, size_t mask) const
+int64_t ArrayPacked::do_get(uint64_t* data, size_t ndx, size_t v_width, size_t v_size, uint64_t mask) const
 {
     if (ndx >= v_size)
         return realm::not_found;

src/realm/array_packed.hpp

@@ -37,7 +37,7 @@ class ArrayPacked {
     void copy_data(const Array&, Array&) const;
     // get or set
     int64_t get(const Array&, size_t) const;
-    int64_t get(const char*, size_t, size_t, size_t, size_t) const;
+    int64_t get(const char*, size_t, size_t, size_t, uint64_t) const;
     void get_chunk(const Array&, size_t, int64_t res[8]) const;
     void set_direct(const Array&, size_t, int64_t) const;
 
@@ -46,7 +46,7 @@ class ArrayPacked {
     int64_t sum(const Array&, size_t, size_t) const;
 
 private:
-    int64_t do_get(uint64_t*, size_t, size_t, size_t, size_t) const;
+    int64_t do_get(uint64_t*, size_t, size_t, size_t, uint64_t) const;
     bool find_all_match(size_t start, size_t end, size_t baseindex, QueryStateBase* state) const;
 
     template <typename Cond>

src/realm/node_header.hpp

@@ -174,7 +174,15 @@ class NodeHeader {
 
     static size_t unsigned_to_num_bits(uint64_t value)
     {
-        return 1 + log2(static_cast<size_t>(value));
+        if constexpr (sizeof(size_t) == sizeof(uint64_t))
+            return 1 + log2(value);
+        uint32_t high = value >> 32;
+        if (high)
+            return 33 + log2(high);
+        uint32_t low = value & 0xFFFFFFFFUL;
+        if (low)
+            return 1 + log2(low);
+        return 0;
     }
 
     static size_t signed_to_num_bits(int64_t value)