docz1105/ComBack_Plus_Plus

ComBack++: A Multi-Language Dataset Providing End-to-End Support for Compiler Backend Development

ComBack++ is a large-scale, multi-platform and multi-language compiler backend code dataset. It is sourced from GCC and LLVM backends corresponding to 183 target platforms.

Dataset Information

• Source Data

  • GCC

  Category	Target Platform	C++ Function	C++ KLoC	Machine Description KLoC
  CPU	30	56,211	858.2	228.5
  MPU	35	8,713	243.8	87.1
  GPU	2	731	12.7	3.0
  VLIW	5	1,323	32.8	17.8
  DSP	4	542	17.2	6.1
  Virtual	5	558	10.5	0.1
  SUM	81	68,078	1175.2	342.6

  • LLVM

  Category	Target Platform	C++ Function	C++ KLoC	TableGen KLoC
  CPU	43	97,972	3,768.4	1,157.7
  MPU	30	12,552	190.7	24.8
  GPU	5	28,267	886.8	168.5
  VLIW	5	2,386	27.4	0.8
  DSP	7	11,584	296.8	129.4
  Virtual	12	10,538	199.4	12.9
  SUM	102	163,299	5,369.5	1,494.1

• Tasks

  • Statement-Level Completion: complete current statement.

  //Inputs:
  ...
  adjustReg(MBB,LastFrameDestroy, DL, SPReg, FPReg, -StackSize+RVFI->getVarArgsSaveSize() 
  //Ground Truth:
  MachineInstr::FrameDestroy);
  

  • Next-Statement Suggestion: predict the next statement.

  //Inputs:
  ...
  maxCallFrameSize = (maxCallFrameSize + AlignMask) & ~AlignMask;
  //Ground Truth:
  MFI -> setMaxCallFrameSize(maxCallFrameSize);
  

  • Code Generation: generate a function template with function description in natrual language.

  //Inputs:
  getPointerRegClass: Returns a TargetRegisterClass used for pointer values.
  //Ground Truth:
  TargetRegisterClass *RegisterInfo::getPointerRegClass(MachineFunction &MF ,unsigned Kind) {
     return Subtarget.is64Bit() ? <ISA_LIT> : <ISA_LIT>;
  }
  

  • Program Repair: automatically fixing bugs at identified locations.

  //Inputs:
  int64_t Imm; <BUGS> MCExpr::VariantKind VK; <BUGE> ...
  //Ground Truth:
  <FIXS> MCExpr::VariantKind VK = MCExpr::VK_None; <FIXE>
  

Organization

• ./*.jsonl: Take data of RISC-V,ARC,NVPTX both in GCC and LLVM as test set, split train/valid set in the ratio of 90%:10% of other targets excluding RI5CY(RI5CY is custmoized based on RISCV)

  Task	Programming Language	Train	Valid	Test
  Statement-Level Completion	C++	149,368 (16.41M)	16,480 (1.84M)	7,897 (0.82M)
  Statement-Level Completion	TableGen/MD	120,460 (4.46M)	13,295 (0.50M)	7,090 (0.25M)
  Next-Statement Suggestion	C++	137,155 (20.55M)	15,218 (2.32M)	8,109 (1.09M)
  Next-Statement Suggestion	TableGen/MD	124,528 (6.71M)	14,009 (0.76M)	2,726 (0.14M)
  Code Generation	C++	37,805 (10.69M)	4,137 (1.16M)	2,944 (0.73M)
  Program Repair	C++	40,578 (4.33M)	4,539 (0.48M)	2,623 (0.29M)
  Program Repair	TableGen/MD	25,983 (2.95M)	2,975 (0.33M)	1,814 (0.24M)

Citation

@article{zhong2025bepilot,
author = {Zhong, Ming and Sun, Xin and Lv, Fang and Wang, Lulin and Geng, Hongna and Qiu, Lei and Cui, Huimin and Feng, Xiaobing},
title = {BePilot: An AI Programming Assistant for Compiler Backend Development},
year = {2025},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
issn = {1049-331X},
url = {https://doi.org/10.1145/3764585},
doi = {10.1145/3764585},
note = {Just Accepted},
journal = {ACM Trans. Softw. Eng. Methodol.}
}