RFR (XS): 8214523: Fix nmethod asserts for concurrent nmethod unloading

[Erik Osterlund]

Hi Vladimir,

Thanks for the review!

/Erik

> On 30 Nov 2018, at 18:56, Vladimir Kozlov <vladimir.kozlov at oracle.com> wrote:
> 
> +1
> 
> Vladimir
> 
>> On 11/30/18 7:49 AM, coleen.phillimore at oracle.com wrote:
>> This looks good to me.
>> Coleen
>>> On 11/30/18 10:46 AM, Erik Österlund wrote:
>>> By popular demand, I decided to CC hotspot-dev. And Coleen.
>>> 
>>> /Erik
>>> 
>>>> On 2018-11-30 11:29, Erik Österlund wrote:
>>>> Hi,
>>>> 
>>>> There are some asserts in nmethod.cpp that do not play well with concurrent nmethod unloading. This patch adjusts said asserts and makes nmethod::maybe_invalidate_installed_code() return early if the nmethod is not JVMCI compiled, to dodge JVMCI requirements on locking, that is far from supported right now anyway for concurrent unloading.
>>>> 
>>>> Bug:
>>>> https://bugs.openjdk.java.net/browse/JDK-8214523
>>>> 
>>>> Webrev:
>>>> http://cr.openjdk.java.net/~eosterlund/8214523/webrev.00/
>>>> 
>>>> Thanks,
>>>> /Erik