RFR: 8139589: [JVMCI] throw exceptions in faulty code installation operations

[Christian Thalinger]

> On Nov 6, 2015, at 5:44 AM, Roland Schatz <roland.schatz at oracle.com> wrote:
> 
> New webrev, including tests:
> http://cr.openjdk.java.net/~rschatz/JDK-8139589/webrev.01/ <http://cr.openjdk.java.net/~rschatz/JDK-8139589/webrev.01/>
> 
> All of these unit tests crash the VM without these changes. With these changes, they simply throw an exception.
> 
> This also includes a few more error checks compared to the previous webrev. These are mainly null accesses that were missed previously because there was no explicit "assert" or "guarantee" check in the code.

Thanks.  I’ve made a required change to the tests because of a jtreg bug:

+ * @requires (os.simpleArch == "x64" | os.simpleArch == "sparcv9") & os.arch != “aarch64"

and adjusted closed sources.

This looks good and I will push this now.

> 
> - Roland
> 
> 
> On 11/05/2015 07:27 PM, Christian Thalinger wrote:
>> 
>>> On Nov 5, 2015, at 2:46 AM, Roland Schatz < <mailto:roland.schatz at oracle.com>roland.schatz at oracle.com <mailto:roland.schatz at oracle.com>> wrote:
>>> 
>>> Hi,
>>> 
>>> That's a followup change to the JVMCI integration. Please review.
>>> 
>>> JIRA:
>>> https://bugs.openjdk.java.net/browse/JDK-8139589 <https://bugs.openjdk.java.net/browse/JDK-8139589>
>>> 
>>> Webrev:
>>> http://cr.openjdk.java.net/~rschatz/JDK-8139589/webrev.00/ <http://cr.openjdk.java.net/%7Erschatz/JDK-8139589/webrev.00/>
>> 
>> I believe you wrote some tests for these changes.  Tom said some use Graal functionally and some don’t.  Could you add the ones that don’t to this change?
>> 
>>> 
>>> Thanks,
>>> Roland
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/attachments/20151106/fe61dec1/attachment.html>