| 일 | 월 | 화 | 수 | 목 | 금 | 토 |
|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | ||
| 6 | 7 | 8 | 9 | 10 | 11 | 12 |
| 13 | 14 | 15 | 16 | 17 | 18 | 19 |
| 20 | 21 | 22 | 23 | 24 | 25 | 26 |
| 27 | 28 | 29 | 30 |
- LLVM
- initial-exec
- TLS
- linux thread
- android inject
- Android
- on stack replacement
- apm
- 난독화
- v8 tracing
- tracing
- Injection
- Linux custom packer
- custom packer
- OSR
- uftrace
- v8 optimizing
- linux debugging
- 안티디버깅
- pinpoint
- anti debugging
- thread local storage
- tracerpid
- Linux packer
- Obfuscator
- LLVM 난독화
- on-stack replacement
- LLVM Obfuscator
- so inject
- pthread
- Today
- Total
목록전체 글 (154)
Why should I know this?
Prompt:issue_206909_scalar_splat_masked_store_formula.md이 문서에서 다루는 최적화 패턴에 대해 이해 및 검증하고, 응용해서 새로운 최적화 가능성이 있는 패턴을 제시해봐 결과GLM-5.2-xhigh DeepSeek4-pro-xhigh HY3 Mimo v2.5 pro-xhigh
1. LLVM MLIR 과 BACKEND 포함 컴파일cmake -G Ninja -S llvm -B /home/m/MLIR \ -DCMAKE_BUILD_TYPE=Release \ -DLLVM_ENABLE_ASSERTIONS=ON \ -DLLVM_ENABLE_PROJECTS="clang;mlir" \ -DLLVM_TARGETS_TO_BUILD="X86;AArch64;RISCV;AMDGPU;NVPTX" \ -DLLVM_BUILD_TESTS=ON \ -DMLIR_INCLUDE_TESTS=ON AMDGPU = AMD, NVPTX=NVIDIA MLIR 을 빌드 후 BACKEND 결과를 만들고 성능 측정하는 과정 설명 예시 MLIR:// cmul.mlirmodule { func.func @comple..
https://github.com/llvm/llvm-project/issues/190773#event-24751520769 [WebAssembly] Type legalization can form cycles when re-promoting SIMD f16 compare masks · Issue #190773 · llvm/llvm-projectSummary When compiling SIMD f16 vector compares for WebAssembly, SelectionDAG type legalization can enter a non-converging/cyclic legalization path. The problem seems to happen when a v4f16 compare...githu..
https://github.com/llvm/llvm-project/pull/191210 [DSE] Refine mayThrowBetween for cross-BB dead store checks by ParkHanbum · Pull Request #191210 · llvm/llvm-projectmayThrowBetween() currently bails out for cross-BB cases if any unmodeled throwing block exists in the function, which is overly conservative and can block valid DSE. Handle same-BB cases by scanni...github.com 기존의 이슈는, DSE 과정에서 kill..
https://github.com/llvm/llvm-project/issues/186572 [X86][AVX10.2] Optimize FP_TO_INT_SAT for non-standard saturation widths to avoid generic expansion penalty · Issue #186572 ·Background & Problem When lowering fp_to_sint_sat or fp_to_uint_sat with a non-standard saturation width (e.g., i24, i48) that does not perfectly match the destination width, the standard fallback ...github.com 이번에 얘기하고 싶..