Bug 15491 – std.parallelism conflicts with thread_detachInstance

Hello, I get a crash during shutdown when having both a TaskPool and a real-time thread detached through thread_detachInstance. Here is the simplest program I could create to reproduce the bug: module main; import core.atomic; import core.thread; import core.sync.semaphore; import std.parallelism; shared bool running = true; void threadMain() { while (running.atomicLoad) { Thread.sleep(dur!"msecs"(10)); } } void main() { scope auto tasks = new TaskPool(4); scope (exit) tasks.finish(true); scope auto sync = new Semaphore(0); auto entry = { sync.notify(); threadMain(); }; auto t = new Thread(entry).start(); sync.wait(); thread_detachInstance(t); running.atomicStore(false); t.join(); } If you comment out the two lines creating and finishing the TaskPool, the program runs correctly. Otherwise it exits with the following stack trace: core.exception.RangeError@src/core/thread.d(1186): Range violation ---------------- 4 dmd_runYU3dOO 0x000000010bf4a045 _d_arraybounds + 97 5 dmd_runYU3dOO 0x000000010bf4b116 core.thread.__array + 38 6 dmd_runYU3dOO 0x000000010bf4c108 pure nothrow @nogc @safe int core.thread.Thread.getAll().__foreachbody1(ref core.thread.Thread) + 48 7 dmd_runYU3dOO 0x000000010bf4c177 int core.thread.Thread.opApply(scope int delegate(ref core.thread.Thread)) + 87 8 dmd_runYU3dOO 0x000000010bf4c066 core.thread.Thread[] core.thread.Thread.getAll() + 102 9 dmd_runYU3dOO 0x000000010bf6c354 void std.parallelism._sharedStaticDtor1358() + 24 10 dmd_runYU3dOO 0x000000010bf6c218 void std.parallelism.__modshareddtor() + 8 11 dmd_runYU3dOO 0x000000010bf5eaa6 void rt.minfo.__T17runModuleFuncsRevS442rt5minfo11ModuleGroup8runDtorsMFZ9__lambda1Z.runModuleFuncsRev(const(immutable(object.ModuleInfo)*)[]) + 86 12 dmd_runYU3dOO 0x000000010bf5e500 void rt.minfo.ModuleGroup.runDtors() + 16 13 dmd_runYU3dOO 0x000000010bf5e85b int rt.minfo.rt_moduleDtor().__foreachbody1(ref rt.sections_osx.SectionGroup) + 27 14 dmd_runYU3dOO 0x000000010bf5e83b rt_moduleDtor + 19 15 dmd_runYU3dOO 0x000000010bf5a933 rt_term + 107 16 dmd_runYU3dOO 0x000000010bf5ad51 void rt.dmain2._d_run_main(int, char**, extern (C) int function(char[][])*).runAll() + 61 17 dmd_runYU3dOO 0x000000010bf5acd3 void rt.dmain2._d_run_main(int, char**, extern (C) int function(char[][])*).tryExec(scope void delegate()) + 55 18 dmd_runYU3dOO 0x000000010bf5ac25 _d_run_main + 497 19 dmd_runYU3dOO 0x000000010bf3dd57 main + 15 20 libdyld.dylib 0x00007fff9a4d25c8 start + 0 21 ??? 0x0000000000000000 0x0 + 0

FWIW, this doesn't crash anymore with git HEAD dmd.

Indeed it is! I'm closing the issue, thanks!