[IMPALA-9306] HdfsOrcScanner crashes in orc::readFully

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: Impala 3.4.0
• Component/s: Backend
• Epic Link: Basic ORC Support
• Epic Color: ghx-label-8
• Assignee: Quanlong Huang
• Reporter: Quanlong Huang
• Created: 19/Jan/20 11:20
• Updated: 02/Mar/20 02:57
• Resolved: 22/Jan/20 11:30

Description

Hit a crash after running test_scanners_fuzz.py for 317 times (took 20h). Using the latest Impala and latest ORC library:

• Impala git-hash: d66610837e53965cb969b78116aec58164bb8548
• ORC git-hash: 35385bdff1374417831628b72468e03558c969f7

Stacktrace:

Thread 295 (crashed)
 0  impalad!orc::readFully(char*, long, orc::SeekableInputStream*) [ColumnReader.cc : 522 + 0x0]
 1  impalad!orc::StringDictionaryColumnReader::StringDictionaryColumnReader(orc::Type const&, orc::StripeStreams&) [ColumnReader.cc : 596 + 0x11]
 2  impalad!orc::buildReader(orc::Type const&, orc::StripeStreams&) [ColumnReader.cc : 1756 + 0x1b]
 3  impalad!orc::StructColumnReader::StructColumnReader(orc::Type const&, orc::StripeStreams&) [ColumnReader.cc : 876 + 0x10]
 4  impalad!orc::buildReader(orc::Type const&, orc::StripeStreams&) [ColumnReader.cc : 1787 + 0x1b]
 5  impalad!orc::RowReaderImpl::startNextStripe() [Reader.cc : 917 + 0x12]
 6  impalad!orc::RowReaderImpl::next(orc::ColumnVectorBatch&) [Reader.cc : 932 + 0x5]
 7  impalad!impala::HdfsOrcScanner::AssembleRows(impala::RowBatch*) [hdfs-orc-scanner.cc : 618 + 0x40]
 8  impalad!impala::HdfsOrcScanner::GetNextInternal(impala::RowBatch*) [hdfs-orc-scanner.cc : 516 + 0x20]
 9  impalad!impala::HdfsOrcScanner::ProcessSplit() [hdfs-orc-scanner.cc : 435 + 0x39]
10  impalad!impala::HdfsScanNode::ProcessSplit(std::vector<impala::FilterContext, std::allocator<impala::FilterContext> > const&, impala::MemPool*, impala::io::ScanRange*, long*) [hdfs-scan-node.cc : 514 + 0x28]
11  impalad!impala::HdfsScanNode::ScannerThread(bool, long) [hdfs-scan-node.cc : 416 + 0x2a]
12  impalad!impala::HdfsScanNode::ThreadTokenAvailableCb(impala::ThreadResourcePool*)::{lambda()#1}::operator()() const + 0x30
13  impalad!boost::detail::function::void_function_obj_invoker0<impala::HdfsScanNode::ThreadTokenAvailableCb(impala::ThreadResourcePool*)::<lambda()>, void>::invoke [function_template.hpp : 153 + 0xc]
14  impalad!boost::function0<void>::operator()() const [function_template.hpp : 767 + 0x11]
15  impalad!impala::Thread::SuperviseThread(std::string const&, std::string const&, boost::function<void ()>, impala::ThreadDebugInfo const*, impala::Promise<long, (impala::PromiseMode)0>*) [thread.cc : 360 + 0xf]
16  impalad!void boost::_bi::list5<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::ThreadDebugInfo*>, boost::_bi::value<impala::Promise<long, (impala::PromiseMode)0>*> >::operator()<void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::ThreadDebugInfo const*, impala::Promise<long, (impala::PromiseMode)0>*), boost::_bi::list0>(boost::_bi::type<void>, void (*&)(std::string const&, std::string const&, boost::function<void ()>, impala::ThreadDebugInfo const*, impala::Promise<long, (impala::PromiseMode)0>*), boost::_bi::list0&, int) [bind.hpp : 525 + 0x15]
17  impalad!boost::_bi::bind_t<void, void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::ThreadDebugInfo const*, impala::Promise<long, (impala::PromiseMode)0>*), boost::_bi::list5<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::ThreadDebugInfo*>, boost::_bi::value<impala::Promise<long, (impala::PromiseMode)0>*> > >::operator()() [bind_template.hpp : 20 + 0x22]
18  impalad!boost::detail::thread_data<boost::_bi::bind_t<void, void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::ThreadDebugInfo const*, impala::Promise<long, (impala::PromiseMode)0>*), boost::_bi::list5<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::ThreadDebugInfo*>, boost::_bi::value<impala::Promise<long, (impala::PromiseMode)0>*> > > >::run() [thread.hpp : 116 + 0x12]
19  impalad!thread_proxy + 0xda
20  libpthread-2.23.so + 0x76ba
21  libc-2.23.so + 0x10741d

How to reproduce
Build Impala using the latest ORC library according to https://cwiki.apache.org/confluence/display/IMPALA/Compile+Impala+with+the+latest+ORC+library
Create table by:

CREATE TABLE uncomp_dst_alltypes (
  id INT,                 
  bool_col BOOLEAN,       
  tinyint_col TINYINT,    
  smallint_col SMALLINT,  
  int_col INT,            
  bigint_col BIGINT,      
  float_col FLOAT,        
  double_col DOUBLE,      
  date_string_col STRING, 
  string_col STRING,      
  timestamp_col TIMESTAMP,
  year INT,               
  month INT               
)                          
STORED AS ORC;

Load the attached orc file and run the query

select count(*) from (select distinct * from uncomp_dst_alltypes) q;

Issue Links

• blocks: IMPALA-6772 Enable test_scanners_fuzz for ORC format
• is caused by: ORC-591 orc::readFully crash due to null pointer variable