[Mlir-commits] [mlir] [MLIR] Split ExecutionEngine Initialization out of ctor into an explicit method call (PR #153524)

[Mehdi Amini]

joker-eph wrote:

> > I went back to the original PR and the bot report: the only sanitizer bot that failed seems to be the Aarch64 one. So isn't the bug just about the aarch64 issue which can't initialize global ctor here? Why do we need to disable the test with sanitizers on other platforms?
> 
> right, maybe only disable it specifically for aarch64 makes more sense. It was following the precedent of `NativeMemRefJit` which has `#if __has_feature(memory_sanitizer)` regardless arch

It is covering a different issue though, and it only restricts on msan: your additions don't seem necessary.

Also wouldn't the test fail on aarch64 even without any sanitizers? Right now we're not initiating at all here:

```
  if (!jit->getTargetTriple().isAArch64())
    cantFail(jit->initialize(jit->getMainJITDylib()));
```

https://github.com/llvm/llvm-project/pull/153524