Uploaded image for project: 'IMPALA'
  1. IMPALA
  2. IMPALA-6137

ASAN heap-use-after-free in HdfsTextScanner::CheckForSplitDelimiter()

Attach filesAttach ScreenshotVotersWatch issueWatchersCreate sub-taskLinkCloneUpdate Comment AuthorReplace String in CommentUpdate Comment VisibilityDelete Comments
    XMLWordPrintableJSON

Details

    Description

      While working on the scanner memory management I found some latent issues with memory lifetime in the text scanners.

      Here's a problem ASAN uncovered on my private branch.

      ==8817==ERROR: AddressSanitizer: heap-use-after-free on address 0x6310008d4803 at pc 0x000001ba3453 bp 0x7faa70821a90 sp 0x7faa70821a88
      READ of size 1 at 0x6310008d4803 thread T12271
          #0 0x1ba3452 in impala::HdfsTextScanner::CheckForSplitDelimiter(bool*) /tmp/be/src/exec/hdfs-text-scanner.cc:705:10
          #1 0x1ba14a9 in impala::HdfsTextScanner::FinishScanRange(impala::RowBatch*) /tmp/be/src/exec/hdfs-text-scanner.cc:243:39
          #2 0x1ba6cee in impala::HdfsTextScanner::GetNextInternal(impala::RowBatch*) /tmp/be/src/exec/hdfs-text-scanner.cc:451:41
          #3 0x1b5353e in impala::HdfsScanner::ProcessSplit() /tmp/be/src/exec/hdfs-scanner.cc:120:21
          #4 0x1b19345 in impala::HdfsScanNode::ProcessSplit(std::vector<impala::FilterContext, std::allocator<impala::FilterContext> > const&, impala::MemPool*, impala::DiskIoMgr::ScanRange*) /tmp/be/s
      rc/exec/hdfs-scan-node.cc:532:21
          #5 0x1b18609 in impala::HdfsScanNode::ScannerThread() /tmp/be/src/exec/hdfs-scan-node.cc:441:16
          #6 0x160bc82 in boost::function0<void>::operator()() const /tmp/toolchain/boost-1.57.0-p3/include/boost/function/function_template.hpp:766:14
          #7 0x1a323e7 in impala::Thread::SuperviseThread(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*) /tmp/be/src/util/thread.cc:352:3
          #8 0x1a3d175 in void boost::_bi::list4<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::Promise<long>*> >
      ::operator()<void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list0>(boost::_bi::type<void>, void (*&)(std::string const&, std::strin
      g const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list0&, int) /tmp/toolchain/boost-1.57.0-p3/include/boost/bind/bind.hpp:457:9
          #9 0x1a3cff1 in boost::_bi::bind_t<void, void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list4<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::Promise<long>*> > >::operator()() /tmp/toolchain/boost-1.57.0-p3/include/boost/bind/bind_template.hpp:20:16
          #10 0x23db4d9 in thread_proxy (/home/tarmstrong/Impala/incubator-impala/be/build/debug/service/impalad+0x23db4d9)
          #11 0x7fad4fe1c6b9 in start_thread (/lib/x86_64-linux-gnu/libpthread.so.0+0x76b9)
          #12 0x7fad4f93c3dc in clone /build/glibc-bfm8X4/glibc-2.23/misc/../sysdeps/unix/sysv/linux/x86_64/clone.S:109
      
      0x6310008d4803 is located 3 bytes inside of 65536-byte region [0x6310008d4800,0x6310008e4800)
      freed by thread T12271 here:
          #0 0x12fc600 in __interceptor_free /data/jenkins/workspace/verify-impala-toolchain-package-build/label/ec2-package-ubuntu-16-04/toolchain/source/llvm/llvm-3.9.1.src/projects/compiler-rt/lib/asan/asan_malloc_linux.cc:47
          #1 0x16bcdbe in impala::DiskIoRequestContext::FreeBuffer(impala::DiskIoMgr::BufferDescriptor*) /tmp/be/src/runtime/disk-io-mgr-reader-context.cc:48:3
          #2 0x16a73fc in impala::DiskIoMgr::ReturnBuffer(std::unique_ptr<impala::DiskIoMgr::BufferDescriptor, std::default_delete<impala::DiskIoMgr::BufferDescriptor> >) /tmp/be/src/runtime/disk-io-mgr.cc:463:15
          #3 0x1d8fc18 in impala::ScannerContext::Stream::ReleaseCompletedResources(bool) /tmp/be/src/exec/scanner-context.cc:109:44
          #4 0x1d8fa1d in impala::ScannerContext::ReleaseCompletedResources(bool) /tmp/be/src/exec/scanner-context.cc:63:18
          #5 0x1b54838 in impala::HdfsScanner::CommitRows(int, impala::RowBatch*) /tmp/be/src/exec/hdfs-scanner.cc:195:15
          #6 0x1ba5622 in impala::HdfsTextScanner::ProcessRange(impala::RowBatch*, int*) /tmp/be/src/exec/hdfs-text-scanner.cc:403:41
          #7 0x1ba6a82 in impala::HdfsTextScanner::GetNextInternal(impala::RowBatch*) /tmp/be/src/exec/hdfs-text-scanner.cc:443:41
          #8 0x1b5353e in impala::HdfsScanner::ProcessSplit() /tmp/be/src/exec/hdfs-scanner.cc:120:21
          #9 0x1b19345 in impala::HdfsScanNode::ProcessSplit(std::vector<impala::FilterContext, std::allocator<impala::FilterContext> > const&, impala::MemPool*, impala::DiskIoMgr::ScanRange*) /tmp/be/src/exec/hdfs-scan-node.cc:532:21
          #10 0x1b18609 in impala::HdfsScanNode::ScannerThread() /tmp/be/src/exec/hdfs-scan-node.cc:441:16
          #11 0x160bc82 in boost::function0<void>::operator()() const /tmp/toolchain/boost-1.57.0-p3/include/boost/function/function_template.hpp:766:14
          #12 0x1a323e7 in impala::Thread::SuperviseThread(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*) /tmp/be/src/util/thread.cc:352:3
          #13 0x1a3d175 in void boost::_bi::list4<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::Promise<long>*> >::operator()<void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list0>(boost::_bi::type<void>, void (*&)(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list0&, int) /tmp/toolchain/boost-1.57.0-p3/include/boost/bind/bind.hpp:457:9
      

      Here's a different problem uncovered on master:

      ./buildall.sh -asan -skiptests -noclean -ninja -notests && start-impala-cluster.py --impalad_args=--disable_mem_pools=true  && impala-py.test -n4 --verbose tests/query_test/test_scanners.py tests/query_test/test_aggregation.py --workload_exploration_strategy=functional-query:exhaustive -k text
      
      ==11633==ERROR: AddressSanitizer: heap-use-after-free on address 0x6310010b882e at pc 0x0000012e8065 bp 0x7f3e8c708150 sp 0x7f3e8c707900
      READ of size 1 at 0x6310010b882e thread T899
          #0 0x12e8064 in __asan_memcpy /data/jenkins/workspace/verify-impala-toolchain-package-build/label/ec2-package-ubuntu-16-04/toolchain/source/llvm/llvm-3.9.1.src/projects/compiler-rt/lib/asan/asan_interceptors.cc:413
          #1 0x1709a72 in impala::Tuple::DeepCopyVarlenData(impala::TupleDescriptor const&, char**, int*, bool) /tmp/be/src/runtime/tuple.cc:143:5
          #2 0x16eedb5 in impala::RowBatch::SerializeInternal(long, impala::FixedSizeHashTable<impala::Tuple*, int>*, impala::TRowBatch*) /tmp/be/src/runtime/row-batch.cc:281:14
          #3 0x16ed1c8 in impala::RowBatch::Serialize(impala::TRowBatch*, bool) /tmp/be/src/runtime/row-batch.cc:188:5
          #4 0x16ecf7b in impala::RowBatch::Serialize(impala::TRowBatch*) /tmp/be/src/runtime/row-batch.cc:161:10
          #5 0x22d3ea9 in impala::DataStreamSender::SerializeBatch(impala::RowBatch*, impala::TRowBatch*, int) /tmp/be/src/runtime/data-stream-sender.cc:518:46
          #6 0x22d67d6 in impala::DataStreamSender::Send(impala::RuntimeState*, impala::RowBatch*) /tmp/be/src/runtime/data-stream-sender.cc:429:41
          #7 0x17352ae in impala::FragmentInstanceState::ExecInternal() /tmp/be/src/runtime/fragment-instance-state.cc:275:48
          #8 0x17324ac in impala::FragmentInstanceState::Exec() /tmp/be/src/runtime/fragment-instance-state.cc:89:14
          #9 0x16dc212 in impala::QueryState::ExecFInstance(impala::FragmentInstanceState*) /tmp/be/src/runtime/query-state.cc:380:24
          #10 0x160e6e2 in boost::function0<void>::operator()() const /tmp/toolchain/boost-1.57.0-p3/include/boost/function/function_template.hpp:766:14
          #11 0x1a390c7 in impala::Thread::SuperviseThread(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*) /tmp/be/src/util/thread.cc:352:3
          #12 0x1a43e55 in void boost::_bi::list4<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::Promise<long>*> >::operator()<void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list0>(boost::_bi::type<void>, void (*&)(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list0&, int) /tmp/toolchain/boost-1.57.0-p3/include/boost/bind/bind.hpp:457:9
          #13 0x1a43cd1 in boost::_bi::bind_t<void, void (*)(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*), boost::_bi::list4<boost::_bi::value<std::string>, boost::_bi::value<std::string>, boost::_bi::value<boost::function<void ()> >, boost::_bi::value<impala::Promise<long>*> > >::operator()() /tmp/toolchain/boost-1.57.0-p3/include/boost/bind/bind_template.hpp:20:16
          #14 0x23e24a9 in thread_proxy (/home/tarmstrong/Impala/incubator-impala/be/build/debug/service/impalad+0x23e24a9)
          #15 0x7f414dab86b9 in start_thread (/lib/x86_64-linux-gnu/libpthread.so.0+0x76b9)
          #16 0x7f414d5d83dc in clone /build/glibc-bfm8X4/glibc-2.23/misc/../sysdeps/unix/sysv/linux/x86_64/clone.S:109
      
      0x6310010b882e is located 46 bytes inside of 65536-byte region [0x6310010b8800,0x6310010c8800)
      freed by thread T911 here:
          #0 0x1336960 in operator delete[](void*) /data/jenkins/workspace/verify-impala-toolchain-package-build/label/ec2-package-ubuntu-16-04/toolchain/source/llvm/llvm-3.9.1.src/projects/compiler-rt/lib/asan/asan_new_delete.cc:114
          #1 0x16acae0 in impala::DiskIoMgr::FreeBufferMemory(impala::DiskIoMgr::BufferDescriptor*) /tmp/be/src/runtime/disk-io-mgr.cc:811:7
          #2 0x16ac545 in impala::DiskIoMgr::ReturnBuffer(std::unique_ptr<impala::DiskIoMgr::BufferDescriptor, std::default_delete<impala::DiskIoMgr::BufferDescriptor> >) /tmp/be/src/runtime/disk-io-mgr.cc:688:7
          #3 0x1d96df4 in impala::ScannerContext::Stream::ReleaseCompletedResources(impala::RowBatch*, bool) /tmp/be/src/exec/scanner-context.cc:117:46
          #4 0x1d96b50 in impala::ScannerContext::ReleaseCompletedResources(impala::RowBatch*, bool) /tmp/be/src/exec/scanner-context.cc:63:18
          #5 0x1ba71fe in impala::HdfsTextScanner::Close(impala::RowBatch*) /tmp/be/src/exec/hdfs-text-scanner.cc:169:15
          #6 0x7f3ed973c141 in impala::HdfsLzoTextScanner::Close(impala::RowBatch*) (/home/tarmstrong/Impala/Impala-lzo/build/libimpalalzo.so+0x1a141)
          #7 0x1b5c0bc in impala::HdfsScanner::Close() /tmp/be/src/exec/hdfs-scanner.cc:129:3
          #8 0x1b21c94 in impala::HdfsScanNode::ProcessSplit(std::vector<impala::FilterContext, std::allocator<impala::FilterContext> > const&, impala::MemPool*, impala::DiskIoMgr::ScanRange*) /tmp/be/src/exec/hdfs-scan-node.cc:551:12
          #9 0x1b20c95 in impala::HdfsScanNode::ScannerThread() /tmp/be/src/exec/hdfs-scan-node.cc:442:16
          #10 0x160e6e2 in boost::function0<void>::operator()() const /tmp/toolchain/boost-1.57.0-p3/include/boost/function/function_template.hpp:766:14
          #11 0x1a390c7 in impala::Thread::SuperviseThread(std::string const&, std::string const&, boost::function<void ()>, impala::Promise<long>*) /tmp/be/src/util/thread.cc:352:3
      
      

      Attachments

        Activity

          This comment will be Viewable by All Users Viewable by All Users
          Cancel

          People

            tarmstrong Tim Armstrong
            tarmstrong Tim Armstrong
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Slack

                Issue deployment