Review request 8008310 - Some adjustments needed to, minimal VM warnings and errors for unsupported command line options

David Holmes david.holmes at oracle.com
Wed Feb 27 20:36:47 PST 2013


On 28/02/2013 3:41 AM, JOSEPH PROVINO wrote:
>
> On 2/26/2013 8:45 PM, David Holmes wrote:
>> On 27/02/2013 2:59 AM, JOSEPH PROVINO wrote:
>>>
>>> On 2/25/2013 8:43 PM, David Holmes wrote:
>>>> On 26/02/2013 1:58 AM, JOSEPH PROVINO wrote:
>>>>> Latest webrev is here:
>>>>> http://cr.openjdk.java.net/~jprovino/8008310/webrev.02
>>>>>
>>>>> - excluded filemap.cpp if CDS is 0.
>>>>>
>>>>> - confined changes for CDS to filemap.hpp.
>>>>
>>>> These changes are good - thanks.
>>>>
>>>> For arguments.cpp:
>>>>
>>>> 1076     FLAG_SET_DEFAULT(RequireSharedSpaces, false);
>>>>
>>>> If we execute this line then the flag is already false.
>>>>
>>>> Also, as discussed in email -Xshare:dump should probably be an error
>>>> not a warning, but note that if left as a warning then this code:
>>>>
>>>> 2518     // -Xshare:dump
>>>> 2519     } else if (match_option(option, "-Xshare:dump", &tail)) {
>>>> 2520       FLAG_SET_CMDLINE(bool, DumpSharedSpaces, true);
>>>> 2521       set_mode_flags(_int);     // Prevent compilation, which
>>>> creates objects
>>>>
>>>> would also force us into intepreter mode, so you would still need to
>>>> check INCLUDE_CDS here.
>>>
>>> If DumpSharedSpaces is changed to return JNI_ERR do I still need a
>>> conditional here?
>>
>> ?? You would need the conditional to decide to return JNI_ERR.
>
> I meant do I need #if INCLUDE_CDS around set_mode_flags(_int);

So I was assuming this was the only place you needed to check this, but 
you have to check for -XX:+DumpSharedSpaces as well. So if you left the 
-Xshare:dump code alone and later checked for DumpSharedSpaces and 
returned with JNI_ERR then it would not matter that the switch to 
interpreter mode had already been made.

David

> joe
>
>>
>> David
>>
>>> joe
>>>
>>>>
>>>> David
>>>> -----
>>>>
>>>>> thanks.
>>>>>
>>>>> joe
>>>>>
>>>>>
>>>
>


More information about the hotspot-dev mailing list