8225429: Regression in disassembly quality

[Ioi Lam]

On 6/9/19 12:20 PM, Schmidt, Lutz wrote:
> Hey John,
>
> You just put another work item on my list. :-)
>
> I agree, you wouldn‘t want that in hs_err.
>
> Regards, Lutz
>
> Lutz Schmidt | P&I Technology HCP Core | +49 (6227) 7-42834
>
> On 7. Jun 2019, at 22:31, John Rose <john.r.rose at oracle.com> wrote:
>
>> On Jun 7, 2019, at 12:11 PM, Schmidt, Lutz <lutz.schmidt at sap.com> wrote:
>>
>> The "big" solution would be to print/disassemble all nmethods in the code heap at failure time. That functionality doesn't exist yet.
> Yes, that's a good vision, and it's worth an RFE or a JEP.
>
> Because of security concerns (noted on other threads today),
> that function should be opt-in and should place its gigantic
> output into a separate file from hs_err.
>
If the failure produces a core file, you can attach to it using the 
serviceablility agent. I wonder if SA allows you to disassemble the 
nmethods with comments. That seems better than dumping tons of stuff 
into a text file.