pytorch/caffe2/utils
Digant Desai f2d95765e4 [pthreadpool] Set max threadlimit to tsan limit (#89453)
Summary:
This will make sure we don't run into an internal assert for clang tsan which has a cap of 63 on concurrently held lock count.
Seems like it is failing with 64 since the comparison is `<`, so setting it to 63 here.

```
llvm-project/compiler-rt/lib/sanitizer_common/sanitizer_deadlock_detector.h:67 "((n_all_locks_)) < (((sizeof(all_locks_with_contexts_)/sizeof((all_locks_with_contexts_)[0]))))"
```

Created from CodeHub with https://fburl.com/edit-in-codehub

Test Plan:
CI

Sandcastle run

Reviewed By: kimishpatel, salilsdesai

Differential Revision: D41444710

Pull Request resolved: https://github.com/pytorch/pytorch/pull/89453
Approved by: https://github.com/mcr229
2022-12-08 02:02:53 +00:00
..
hip
math
threadpool [pthreadpool] Set max threadlimit to tsan limit (#89453) 2022-12-08 02:02:53 +00:00
bench_utils.cc
bench_utils.h
cast.h
cast_test.cc
cblas.h
CMakeLists.txt
conversions.h
cpu_neon.h
cpuid.cc
cpuid.h
cpuid_test.cc
cub_namespace.cuh
eigen_utils.h
fatal_signal_asan_no_sig_test.cc
filler.h
fixed_divisor.h
fixed_divisor_test.cc
GpuAtomics.cuh
GpuBitonicSort.cuh
GpuDefs.cuh
GpuScanUtils.cuh
knob_patcher.cc
knob_patcher.h
knobs.cc
knobs.h
knobs_test.cc
map_utils.h
math-detail.h
math.h
math_cpu.cc
math_gpu.cu
math_gpu_test.cc
math_test.cc
murmur_hash3.cc
murmur_hash3.h
proto_convert.cc
proto_convert.h
proto_utils.cc
proto_utils.h
proto_utils_test.cc
proto_wrap.cc
proto_wrap.h
signal_handler.cc
signal_handler.h
simple_queue.h
simple_queue_test.cc
smart_tensor_printer.cc
smart_tensor_printer.h
smart_tensor_printer_test.cc
string_utils.cc
string_utils.h
zmq_helper.h