[CI] Threaded predict error.

[trivialfis]

https://github.com/dmlc/xgboost/actions/runs/12339183942/job/34435083920

 [ RUN      ] Learner.MultiThreadedPredict
=================================================================
==3895==ERROR: AddressSanitizer: heap-use-after-free on address 0x603008e17ec8 at pc 0x55b82804543a bp 0x7f6d7f5fde60 sp 0x7f6d7f5fde50
READ of size 4 at 0x603008e17ec8 thread T15161
    #0 0x55b828045439 in xgboost::gbm::GBTree::PredictBatchImpl(xgboost::DMatrix*, xgboost::PredictionCacheEntry*, bool, int, int) const /workspace/src/gbm/gbtree.cc:490
    #1 0x55b828047051 in xgboost::gbm::GBTree::PredictBatch(xgboost::DMatrix*, xgboost::PredictionCacheEntry*, bool, int, int) /workspace/src/gbm/gbtree.cc:528
    #2 0x55b82814bf5f in xgboost::LearnerImpl::PredictRaw(xgboost::DMatrix*, xgboost::PredictionCacheEntry*, bool, unsigned int, unsigned int) const /workspace/src/learner.cc:1473
    #3 0x55b828148ecd in xgboost::LearnerImpl::Predict(std::shared_ptr<xgboost::DMatrix>, bool, xgboost::HostDeviceVector<float>*, int, int, bool, bool, bool, bool, bool) /workspace/src/learner.cc:1399
    #4 0x55b82992aadd in operator() /workspace/tests/cpp/test_learner.cc:274
    #5 0x55b82997275e in __invoke_impl<void, xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/invoke.h:60
    #6 0x55b829972265 in __invoke<xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/invoke.h:95
    #7 0x55b829971a6c in _M_invoke<0> /usr/include/c++/10/thread:264
    #8 0x55b8299706b4 in operator() /usr/include/c++/10/thread:271
    #9 0x55b82996bb31 in _M_run /usr/include/c++/10/thread:215
    #10 0x7f6e5c0ed252  (/lib/x86_64-linux-gnu/libstdc++.so.6+0xdc252)
    #11 0x7f6e5bd73ac2  (/lib/x86_64-linux-gnu/libc.so.6+0x94ac2)
    #12 0x7f6e5be04a03 in __clone (/lib/x86_64-linux-gnu/libc.so.6+0x125a03)

0x603008e17ec8 is located 24 bytes inside of 32-byte region [0x603008e17eb0,0x603008e17ed0)
freed by thread T15161 here:
    #0 0x7f6e5c95824f in operator delete(void*, unsigned long) ../../../../src/libsanitizer/asan/asan_new_delete.cpp:172
    #1 0x55b8281ba2e9 in __gnu_cxx::new_allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> >::deallocate(std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2>*, unsigned long) /usr/include/c++/10/ext/new_allocator.h:139
    #2 0x55b8281b8514 in std::allocator_traits<std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> > >::deallocate(std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> >&, std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2>*, unsigned long) /usr/include/c++/10/bits/alloc_traits.h:492
    #3 0x55b8281b4818 in std::__allocated_ptr<std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> > >::~__allocated_ptr() /usr/include/c++/10/bits/allocated_ptr.h:73
    #4 0x55b8281bc991 in std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2>::_M_destroy() (/workspace/build/testxgboost+0x9a2b991)
    #5 0x55b827138d9c in std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release() /usr/include/c++/10/bits/shared_ptr_base.h:174
    #6 0x55b82710aaab in std::__shared_count<(__gnu_cxx::_Lock_policy)2>::~__shared_count() /usr/include/c++/10/bits/shared_ptr_base.h:736
    #7 0x55b8280f9c09 in std::__shared_ptr<xgboost::PredictionCacheEntry, (__gnu_cxx::_Lock_policy)2>::~__shared_ptr() /usr/include/c++/10/bits/shared_ptr_base.h:1188
    #8 0x55b8280f9c70 in std::shared_ptr<xgboost::PredictionCacheEntry>::~shared_ptr() /usr/include/c++/10/bits/shared_ptr.h:121
    #9 0x55b8280f9e86 in xgboost::PredictionContainer::Cache(std::shared_ptr<xgboost::DMatrix>, xgboost::DeviceOrd) /workspace/include/xgboost/predictor.h:55
    #10 0x55b828148d30 in xgboost::LearnerImpl::Predict(std::shared_ptr<xgboost::DMatrix>, bool, xgboost::HostDeviceVector<float>*, int, int, bool, bool, bool, bool, bool) /workspace/src/learner.cc:1398
    #11 0x55b82992aadd in operator() /workspace/tests/cpp/test_learner.cc:274
    #12 0x55b82997275e in __invoke_impl<void, xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/invoke.h:60
    #13 0x55b829972265 in __invoke<xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/invoke.h:95
    #14 0x55b829971a6c in _M_invoke<0> /usr/include/c++/10/thread:264
    #15 0x55b8299706b4 in operator() /usr/include/c++/10/thread:271
    #16 0x55b82996bb31 in _M_run /usr/include/c++/10/thread:215
    #17 0x7f6e5c0ed252  (/lib/x86_64-linux-gnu/libstdc++.so.6+0xdc252)

previously allocated by thread T15161 here:
    #0 0x7f6e5c9571e7 in operator new(unsigned long) ../../../../src/libsanitizer/asan/asan_new_delete.cpp:99
    #1 0x55b8281ba2a2 in __gnu_cxx::new_allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> >::allocate(unsigned long, void const*) /usr/include/c++/10/ext/new_allocator.h:121
    #2 0x55b8281b83dc in std::allocator_traits<std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> > >::allocate(std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> >&, unsigned long) /usr/include/c++/10/bits/alloc_traits.h:460
    #3 0x55b8281b46af in std::__allocated_ptr<std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> > > std::__allocate_guarded<std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> > >(std::allocator<std::_Sp_counted_ptr_inplace<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>, (__gnu_cxx::_Lock_policy)2> >&) /usr/include/c++/10/bits/allocated_ptr.h:97
    #4 0x55b8281adb47 in std::__shared_count<(__gnu_cxx::_Lock_policy)2>::__shared_count<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>>(xgboost::PredictionCacheEntry*&, std::_Sp_alloc_shared_tag<std::allocator<xgboost::PredictionCacheEntry> >) (/workspace/build/testxgboost+0x9a1cb47)
    #5 0x55b8281a4057 in std::__shared_ptr<xgboost::PredictionCacheEntry, (__gnu_cxx::_Lock_policy)2>::__shared_ptr<std::allocator<xgboost::PredictionCacheEntry>>(std::_Sp_alloc_shared_tag<std::allocator<xgboost::PredictionCacheEntry> >) (/workspace/build/testxgboost+0x9a13057)
    #6 0x55b828195c8b in std::shared_ptr<xgboost::PredictionCacheEntry>::shared_ptr<std::allocator<xgboost::PredictionCacheEntry>>(std::_Sp_alloc_shared_tag<std::allocator<xgboost::PredictionCacheEntry> >) (/workspace/build/testxgboost+0x9a04c8b)
    #7 0x55b828180435 in std::shared_ptr<xgboost::PredictionCacheEntry> std::allocate_shared<xgboost::PredictionCacheEntry, std::allocator<xgboost::PredictionCacheEntry>>(std::allocator<xgboost::PredictionCacheEntry> const&) /usr/include/c++/10/bits/shared_ptr.h:862
    #8 0x55b828166374 in std::shared_ptr<xgboost::PredictionCacheEntry> std::make_shared<xgboost::PredictionCacheEntry>() (/workspace/build/testxgboost+0x99d5374)
    #9 0x55b82814f898 in std::shared_ptr<xgboost::PredictionCacheEntry> xgboost::DMatrixCache<xgboost::PredictionCacheEntry>::CacheItem<>(std::shared_ptr<xgboost::DMatrix>) (/workspace/build/testxgboost+0x99be898)
    #10 0x55b8280f9dba in xgboost::PredictionContainer::Cache(std::shared_ptr<xgboost::DMatrix>, xgboost::DeviceOrd) /workspace/include/xgboost/predictor.h:55
    #11 0x55b828148d30 in xgboost::LearnerImpl::Predict(std::shared_ptr<xgboost::DMatrix>, bool, xgboost::HostDeviceVector<float>*, int, int, bool, bool, bool, bool, bool) /workspace/src/learner.cc:1398
    #12 0x55b82992aadd in operator() /workspace/tests/cpp/test_learner.cc:274
    #13 0x55b82997275e in __invoke_impl<void, xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/invoke.h:60
    #14 0x55b829972265 in __invoke<xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/invoke.h:95
    #15 0x55b829971a6c in _M_invoke<0> /usr/include/c++/10/thread:264
    #16 0x55b8299706b4 in operator() /usr/include/c++/10/thread:271
    #17 0x55b82996bb31 in _M_run /usr/include/c++/10/thread:215
    #18 0x7f6e5c0ed252  (/lib/x86_64-linux-gnu/libstdc++.so.6+0xdc252)

Thread T15161 created by T0 here:
    #0 0x7f6e5c8f9685 in __interceptor_pthread_create ../../../../src/libsanitizer/asan/asan_interceptors.cpp:216
    #1 0x7f6e5c0ed328 in std::thread::_M_start_thread(std::unique_ptr<std::thread::_State, std::default_delete<std::thread::_State> >, void (*)()) (/lib/x86_64-linux-gnu/libstdc++.so.6+0xdc328)
    #2 0x55b82994b5a3 in construct<std::thread, xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/ext/new_allocator.h:156
    #3 0x55b829947414 in construct<std::thread, xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/alloc_traits.h:512
    #4 0x55b82994179d in emplace_back<xgboost::Learner_MultiThreadedPredict_Test::TestBody()::<lambda()> > /usr/include/c++/10/bits/vector.tcc:115
    #5 0x55b82992c698 in xgboost::Learner_MultiThreadedPredict_Test::TestBody() /workspace/tests/cpp/test_learner.cc:279
    #6 0x55b829d24937 in void testing::internal::HandleSehExceptionsInMethodIfSupported<testing::Test, void>(testing::Test*, void (testing::Test::*)(), char const*) /workspace/build/googletest-src/googletest/src/gtest.cc:2607
    #7 0x55b829d06af6 in void testing::internal::HandleExceptionsInMethodIfSupported<testing::Test, void>(testing::Test*, void (testing::Test::*)(), char const*) /workspace/build/googletest-src/googletest/src/gtest.cc:2643
    #8 0x55b829c4da6d in testing::Test::Run() /workspace/build/googletest-src/googletest/src/gtest.cc:2682
    #9 0x55b829c51394 in testing::TestInfo::Run() /workspace/build/googletest-src/googletest/src/gtest.cc:2861
    #10 0x55b829c556f9 in testing::TestSuite::Run() /workspace/build/googletest-src/googletest/src/gtest.cc:3015
    #11 0x55b829c9cc5b in testing::internal::UnitTestImpl::RunAllTests() /workspace/build/googletest-src/googletest/src/gtest.cc:5855
    #12 0x55b829d2a878 in bool testing::internal::HandleSehExceptionsInMethodIfSupported<testing::internal::UnitTestImpl, bool>(testing::internal::UnitTestImpl*, bool (testing::internal::UnitTestImpl::*)(), char const*) /workspace/build/googletest-src/googletest/src/gtest.cc:2607
    #13 0x55b829d0b799 in bool testing::internal::HandleExceptionsInMethodIfSupported<testing::internal::UnitTestImpl, bool>(testing::internal::UnitTestImpl*, bool (testing::internal::UnitTestImpl::*)(), char const*) /workspace/build/googletest-src/googletest/src/gtest.cc:2643
    #14 0x55b829c902f8 in testing::UnitTest::Run() /workspace/build/googletest-src/googletest/src/gtest.cc:5438
    #15 0x55b8299d502b in RUN_ALL_TESTS() /workspace/build/googletest-src/googletest/include/gtest/gtest.h:2490
    #16 0x55b8299d4bb2 in main /workspace/tests/cpp/test_main.cc:19
    #17 0x7f6e5bd08d8f  (/lib/x86_64-linux-gnu/libc.so.6+0x29d8f)

SUMMARY: AddressSanitizer: heap-use-after-free /workspace/src/gbm/gbtree.cc:490 in xgboost::gbm::GBTree::PredictBatchImpl(xgboost::DMatrix*, xgboost::PredictionCacheEntry*, bool, int, int) const
Shadow bytes around the buggy address:
  0x0c06811baf80: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
  0x0c06811baf90: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
  0x0c06811bafa0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
  0x0c06811bafb0: fa fa fa fa fa fa fa fa fd fd fd fa fa fa fd fd
  0x0c06811bafc0: fd fa fa fa fa fa fa fa fa fa fd fd fd fa fa fa
=>0x0c06811bafd0: 00 00 00 fa fa fa fd fd fd[fd]fa fa fd fd fd fa
  0x0c06811bafe0: fa fa fd fd fd fa fa fa 00 00 00 00 fa fa 00 00
  0x0c06811baff0: 00 fa fa fa fd fd fd fa fa fa fd fd fd fa fa fa
  0x0c06811bb000: fd fd fd fd fa fa fd fd fd fd fa fa fa fa fa fa
  0x0c06811bb010: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
  0x0c06811bb020: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
Shadow byte legend (one shadow byte represents 8 application bytes):
  Addressable:           00
  Partially addressable: 01 02 03 04 05 06 07 
  Heap left redzone:       fa
  Freed heap region:       fd
  Stack left redzone:      f1
  Stack mid redzone:       f2
  Stack right redzone:     f3
  Stack after return:      f5
  Stack use after scope:   f8
  Global redzone:          f9
  Global init order:       f6
  Poisoned by user:        f7
  Container overflow:      fc
  Array cookie:            ac
  Intra object redzone:    bb
  ASan internal:           fe
  Left alloca redzone:     ca
  Right alloca redzone:    cb
  Shadow gap:              cc
==3895==ABORTING