Segfault in CompiledMethod::is_unloading()

[Erik Österlund]

Hi Andrew,

On 2018-11-16 18:07, Andrew Haley wrote:
> On 11/16/18 12:51 PM, Erik Österlund wrote:
>> By the look of it, the fact that you are crashing there means that it
>> was not been visited for your AoT method, and when the sweeper tries
>> calling it, it needs to compute the value and has no means of doing that
>> any longer (it should already have been computed). I guess I must have
>> messed up the assumption that all compiled methods are walked during
>> code cache unloading in the case of AoT. Sorry about that.
> 
> OK, thanks. Given that this is a crasher on AOT-compiled code and
> there's no workaround at present, I guess it has to be fixed. I'm
> not really clear how it should be fixed, though: do we really want
> to walk all AOT-compiled methods during code cache unloading?
> 

We have to walk all is_alive() AOT-compiled code during code cache 
unloading for two reasons:

1) To compute is_unloading
2) To clean inline caches referring to compiled methods that died due to 
unloading.

By moving the CompiledMethod::is_unloading() logic to 
nmethod::is_unloading() and have AotCompiledMethod::is_unloading() 
simply return false, the first reason would become invalid. But I think 
we still need to walk all is_alive() AotCompiledMethods to clean their 
inline caches. But if my theory is right about what is going on here, 
then it might at least solve this bug.

/Erik