【Hackathon 9th No.4】fix index_add 0-size

[enkilee]

PR Category

Execute Infrastructure

PR Types

Bug fixes

Description

 paddle.index_add(Tensor([10],"float32"), Tensor([0],"int32"), 0, Tensor([4],"float32"), ) 
 (External) CUDA error(700), an illegal memory access was encountered. 
  [Hint: 'cudaErrorIllegalAddress'. The device encountered a load or store instruction on an invalid memory address. This leaves the process in an inconsistentstate and any further CUDA work will return the same error. To continue using CUDA, the process must be terminated and relaunched. ] (at /home/aistudio/Paddle/paddle/fluid/pybind/eager_functions.cc:1435)

terminate called after throwing an instance of 'common::enforce::EnforceNotMet'
  what():  (External) CUDA error(700), an illegal memory access was encountered. 
  [Hint: 'cudaErrorIllegalAddress'. The device encountered a load or store instruction on an invalid memory address. This leaves the process in an inconsistentstate and any further CUDA work will return the same error. To continue using CUDA, the process must be terminated and relaunched. ] (at /home/aistudio/Paddle/paddle/phi/core/platform/device/gpu/gpu_info.cc:348)



--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
0   std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release()
1   phi::DenseTensor::~DenseTensor()
2   std::_Sp_counted_deleter<phi::Allocation*, std::function<void (phi::Allocation*)>, std::allocator<void>, (__gnu_cxx::_Lock_policy)2>::_M_dispose()
3   paddle::memory::allocation::CUDAAllocator::FreeImpl(phi::Allocation*)

----------------------
Error Message Summary:
----------------------
FatalError: `Process abort signal` is detected by the operating system.
  [TimeInfo: *** Aborted at 1755826917 (unix time) try "date -d @1755826917" if you are using GNU date ***]
  [SignalInfo: *** SIGABRT (@0x3e800000e48) received by PID 3656 (TID 0x7f57ee946740) from PID 3656 ***]

Aborted (core dumped)

1. check src_dim_idx in index_add_cuda_kernel
2. add PADDLE_ENFORCE_NOT_NULL

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[enkilee]

/re-run all-failed

[DanielSun11]

请补充下paddleapitest回测的结果。

#72637
paddleapitest地址 https://github.com/PFCCLab/PaddleAPITest
出错配置存放的位置：

• CPU执行的测试：https://github.com/PFCCLab/PaddleAPITest/tree/main/report/0size_tensor_cpu
• GPU执行的测试： https://github.com/PFCCLab/PaddleAPITest/tree/main/report/0size_tensor_gpu
  参考运行命令：
  python engine.py --accuracy=True --api_config_file= your config file

[codecov-commenter]

Codecov Report

❌ Patch coverage is 88.88889% with 1 line in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (develop@955e32c). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
paddle/phi/infermeta/binary.cc	88.88%	1 Missing ⚠️

❌ Your patch status has failed because the patch coverage (88.88%) is below the target coverage (90.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@            Coverage Diff             @@
##             develop   #74831   +/-   ##
==========================================
  Coverage           ?   88.88%           
==========================================
  Files              ?        1           
  Lines              ?        9           
  Branches           ?        0           
==========================================
  Hits               ?        8           
  Misses             ?        1           
  Partials           ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[enkilee]

@DanielSun11 please review if u are free. thx.

[DanielSun11]

LGTM