Skip to content

Commit

Permalink
[Fix](multi-catalog) Fix some undefined behaviors. (apache#37845)
Browse files Browse the repository at this point in the history
## Proposed changes

- Null pointer of type 'doris::StringRef' in orc reader. The root cause
is error will throw when `num_values == 0` in
`_decode_string_non_dict_encoded_column`.
```
/var/local/ldb-toolchain/bin/../lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/stl_vector.h:1046:9: runtime error: reference binding to null pointer of type 'doris::StringRef'
    #0 0x562516fa9770 in std::vector<doris::StringRef, std::allocator<doris::StringRef> >::operator[](unsigned long) /var/local/ldb-toolchain/bin/../lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/stl_vector.h:1046:2
    #1 0x562516fa9770 in doris::Status doris::vectorized::OrcReader::_decode_string_non_dict_encoded_column<false>(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, COW<doris::vectorized::IColumn>::mutable_ptr<doris::vectorized::IColumn> const&, orc::TypeKind const&, orc::EncodedStringVectorBatch*, unsigned long) /root/doris/be/src/vec/exec/format/orc/vorc_reader.cpp:1164:39
    #2 0x562516f9c08b in doris::Status doris::vectorized::OrcReader::_decode_string_column<false>(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, COW<doris::vectorized::IColumn>::mutable_ptr<doris::vectorized::IColumn> const&, orc::TypeKind const&, orc::ColumnVectorBatch*, unsigned long) /root/doris/be/src/vec/exec/format/orc/vorc_reader.cpp:1116:16
    #3 0x562516f91d73 in doris::Status doris::vectorized::OrcReader::_fill_doris_data_column<false>(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, COW<doris::vectorized::IColumn>::mutable_ptr<doris::vectorized::IColumn>&, std::shared_ptr<doris::vectorized::IDataType const> const&, orc::Type const*, orc::ColumnVectorBatch*, unsigned long) /root/doris/be/src/vec/exec/format/orc/vorc_reader.cpp:1357:16
    #4 0x562516c79a0c in doris::Status doris::vectorized::OrcReader::_orc_column_to_doris_column<false>(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, COW<doris::vectorized::IColumn>::immutable_ptr<doris::vectorized::IColumn>&, std::shared_ptr<doris::vectorized::IDataType const> const&, orc::Type const*, orc::ColumnVectorBatch*, unsigned long) /root/doris/be/src/vec/exec/format/orc/vorc_reader.cpp:1524:5
    #5 0x562516f9339a in doris::Status doris::vectorized::OrcReader::_fill_doris_data_column<false>(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, COW<doris::vectorized::IColumn>::mutable_ptr<doris::vectorized::IColumn>&, std::shared_ptr<doris::vectorized::IDataType const> const&, orc::Type const*, orc::ColumnVectorBatch*, unsigned long) /root/doris/be/src/vec/exec/format/orc/vorc_reader.cpp:1402:9
    #6 0x562516c79a0c in doris::Status doris::vectorized::OrcReader::_orc_column_to_doris_column<false>(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, COW<doris::vectorized::IColumn>::immutable_ptr<doris::vectorized::IColumn>&, std::shared_ptr<doris::vectorized::IDataType const> const&, orc::Type const*, orc::ColumnVectorBatch*, unsigned long) /root/doris/be/src/vec/exec/format/orc/vorc_reader.cpp:1524:5
...
```

- Shift exponent 128 is too large for 128-bit type 'ValueCopyType' (aka
'__int128') in parquet reader. The root cause is error will throw when
`len == 0`.
```
/root/doris/be/src/vec/exec/format/parquet/parquet_column_convert.h:413:27: runtime error: shift exponent 128 is too large for 128-bit type 'ValueCopyType' (aka '__int128')
    #0 0x56251760fbc7 in doris::vectorized::parquet::StringToDecimal<doris::vectorized::Decimal128V3, (doris::vectorized::DecimalScaleParams::ScaleType)1>::physical_convert(COW<doris::vectorized::IColumn>::immutable_ptr<doris::vectorized::IColumn>&, COW<doris::vectorized::IColumn>::immutable_ptr<doris::vectorized::IColumn>&) /root/doris/be/src/vec/exec/format/parquet/parquet_column_convert.h:413:27
    #1 0x562517290dc4 in doris::vectorized::parquet::PhysicalToLogicalConverter::convert(COW<doris::vectorized::IColumn>::immutable_ptr<doris::vectorized::IColumn>&, doris::TypeDescriptor, std::shared_ptr<doris::vectorized::IDataType const> const&, COW<doris::vectorized::IColumn>::immutable_ptr<doris::vectorized::IColumn>&, bool) /root/doris/be/src/vec/exec/format/parquet/parquet_column_convert.h:209:9
    #2 0x562517284a6d in doris::vectorized::ScalarColumnReader::read_column_data(COW<doris::vectorized::IColumn>::immutable_ptr<doris::vectorized::IColumn>&, std::shared_ptr<doris::vectorized::IDataType const>&, doris::vectorized::ColumnSelectVector&, unsigned long, unsigned long*, bool*, bool) /root/doris/be/src/vec/exec/format/parquet/vparquet_column_reader.cpp:569:24
    #3 0x56251725ae7e in doris::vectorized::RowGroupReader::_read_column_data(doris::vectorized::Block*, std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&, unsigned long, unsigned long*, bool*, doris::vectorized::ColumnSelectVector&) /root/doris/be/src/vec/exec/format/parquet/vparquet_group_reader.cpp:421:13
    #4 0x56251724d6d2 in doris::vectorized::RowGroupReader::next_batch(doris::vectorized::Block*, unsigned long, unsigned long*, bool*) /root/doris/be/src/vec/exec/format/parquet/vparquet_group_reader.cpp:321:9
    #5 0x56251708eb97 in doris::vectorized::ParquetReader::get_next_block(doris::vectorized::Block*, unsigned long*, bool*) /root/doris/be/src/vec/exec/format/parquet/vparquet_reader.cpp:530:36
    #6 0x56253036772d in doris::vectorized::VFileScanner::_get_block_wrapped(doris::RuntimeState*, doris::vectorized::Block*, bool*) /root/doris/be/src/vec/exec/scan/vfile_scanner.cpp:311:13
    #7 0x562530366549 in doris::vectorized::VFileScanner::_get_block_impl(doris::RuntimeState*, doris::vectorized::Block*, bool*) /root/doris/be/src/vec/exec/scan/vfile_scanner.cpp:253:17
    #8 0x5625176e79c8 in doris::vectorized::VScanner::get_block(doris::RuntimeState*, doris::vectorized::Block*, bool*) /root/doris/be/src/vec/exec/scan/vscanner.cpp:117:17
    #9 0x5625176e6fc1 in doris::vectorized::VScanner::get_block_after_projects(doris::RuntimeState*, doris::vectorized::Block*, bool*) /root/doris/be/src/vec/exec/scan/vscanner.cpp:84:12
    #10 0x562517698047 in doris::vectorized::ScannerScheduler::_scanner_scan(std::shared_ptr<doris::vectorized::ScannerContext>, std::shared_ptr<doris::vectorized::ScanTask>) /root/doris/be/src/vec/exec/scan/scanner_scheduler.cpp:250:5
    #11 0x56251769bc1f in doris::vectorized::ScannerScheduler::submit(std::shared_ptr<doris::vectorized::ScannerContext>, std::shared_ptr<doris::vectorized::ScanTask>)::$_1::operator()() const::'lambda'()::operator()() const::'lambda'()::operator()() const /root/doris/be/src/vec/exec/scan/scanner_scheduler.cpp:172:25
    #12 0x56251769bc1f in doris::vectorized::ScannerScheduler::submit(std::shared_ptr<doris::vectorized::ScannerContext>, std::shared_ptr<doris::vectorized::ScanTask>)::$_1::operator()() const::'lambda'()::operator()() const /root/doris/be/src/vec/exec/scan/scanner_scheduler.cpp:171:35
...
```
  • Loading branch information
kaka11chen authored Jul 16, 2024
1 parent a1b0264 commit eb17b2e
Show file tree
Hide file tree
Showing 2 changed files with 16 additions and 14 deletions.
4 changes: 2 additions & 2 deletions be/src/vec/exec/format/orc/vorc_reader.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -1583,7 +1583,7 @@ Status OrcReader::get_next_block_impl(Block* block, size_t* read_rows, bool* eof
_decimal_scale_params_index = 0;
try {
rr = _row_reader->nextBatch(*_batch, block);
if (rr == 0) {
if (rr == 0 || _batch->numElements == 0) {
*eof = true;
*read_rows = 0;
return Status::OK();
Expand Down Expand Up @@ -1653,7 +1653,7 @@ Status OrcReader::get_next_block_impl(Block* block, size_t* read_rows, bool* eof
_decimal_scale_params_index = 0;
try {
rr = _row_reader->nextBatch(*_batch, block);
if (rr == 0) {
if (rr == 0 || _batch->numElements == 0) {
*eof = true;
*read_rows = 0;
return Status::OK();
Expand Down
26 changes: 14 additions & 12 deletions be/src/vec/exec/format/parquet/parquet_column_convert.h
Original file line number Diff line number Diff line change
Expand Up @@ -408,18 +408,20 @@ class StringToDecimal : public PhysicalToLogicalConverter {
// When Decimal in parquet is stored in byte arrays, binary and fixed,
// the unscaled number must be encoded as two's complement using big-endian byte order.
ValueCopyType value = 0;
memcpy(reinterpret_cast<char*>(&value), buf + offset[i - 1], len);
value = BitUtil::big_endian_to_host(value);
value = value >> ((sizeof(value) - len) * 8);
if constexpr (ScaleType == DecimalScaleParams::SCALE_UP) {
value *= scale_params.scale_factor;
} else if constexpr (ScaleType == DecimalScaleParams::SCALE_DOWN) {
value /= scale_params.scale_factor;
} else if constexpr (ScaleType == DecimalScaleParams::NO_SCALE) {
// do nothing
} else {
LOG(FATAL) << "__builtin_unreachable";
__builtin_unreachable();
if (len > 0) {
memcpy(reinterpret_cast<char*>(&value), buf + offset[i - 1], len);
value = BitUtil::big_endian_to_host(value);
value = value >> ((sizeof(value) - len) * 8);
if constexpr (ScaleType == DecimalScaleParams::SCALE_UP) {
value *= scale_params.scale_factor;
} else if constexpr (ScaleType == DecimalScaleParams::SCALE_DOWN) {
value /= scale_params.scale_factor;
} else if constexpr (ScaleType == DecimalScaleParams::NO_SCALE) {
// do nothing
} else {
LOG(FATAL) << "__builtin_unreachable";
__builtin_unreachable();
}
}
auto& v = reinterpret_cast<DecimalType&>(data[start_idx + i]);
v = (DecimalType)value;
Expand Down

0 comments on commit eb17b2e

Please sign in to comment.