[14] RFR(S): 8225653: Provide more information when hitting SIGILL from HaltNode

[Christian Hagedorn]

Hi Vladimir

Thanks for the review.

On 09.09.19 13:36, Vladimir Ivanov wrote:
> 
>> I updated the webrev choosing the option to use MacroAssembler::stop() 
>> instead which requires less code changes:
>> http://cr.openjdk.java.net/~chagedorn/8225653/webrev.01/
> 
> I suggest to make assert->fatal changes as a separate RFE.
> 
> There are different places where it is assumed 
> MacroAssembler::debug32/64 may return. I'd like those to be cleaned up, 
> but it would pollute the patch.
> 
> IMO printing error message on the console is OK for now (hs_err file 
> will contain generic error info).
> 

Sounds good, I created a new RFE for it [1].

> Some nitpicking:
>    "Halt node: Never-taken loop exit reached"
> 
> I question "Halt node" prefix utility. It definitely makes the message 
> self-descriptive on error, but IMO it adds more noise in the source 
> code. So, I'd prefer to leave it out.
> 

I updated the webrev to exclude those prefixes together with the removal 
of the changes to MacroAssembler regarding the assert->fatal changes:
http://cr.openjdk.java.net/~chagedorn/8225653/webrev.02/

Best regards,
Christian


[1] https://bugs.openjdk.java.net/browse/JDK-8230762