[pypy-issue] [issue1045] Crash of the JIT

[Daid]

Daid <daid303 at gmail.com> added the comment:

Actually, I don't agree with your earlier statement "other.level ==
LEVEL_UNKNOWN". Because you cannot know that for sure unless that was logged
somewhere. At this point the only thing you know fore sure is that other.level
is not "LEVEL_CONSTANT", it could have any value, even an invalid value for that
variable. You also don't describe where the problem is, you describe where the
error is detected, but that's just semantics.

This might be my C/embedded background speaking, where these kinds of errors are
more common. But the fact that it's a non-deterministic error makes me think in
that direction (stack/heap corruption or uninitialized value). Because if it was
a general jit optimizer bug then it should happen always right? Or is the jit
throwing dice? (not what I would expect from a jit, but it's not my area of
expertise)

But feel free to point me to the error of my reasoning :) if you want.

________________________________________
PyPy bug tracker <tracker at bugs.pypy.org>
<https://bugs.pypy.org/issue1045>
________________________________________