Request for review 7122219: Passed StringTableSize value not verified

[harold seigel]

Hi David,

I did not address this in my previous mail:

    Was the option of renaming StringTableSize to something reflecting
    that it is actually the bucket count rejected? I know there is a
    follow on CR in this area but giving this variable a more accurate
    name doesn't seem unreasonable.

I would prefer that this be dealt with in the follow on CR in order to 
reduce the number of times the flag gets renamed and its meaning gets 
changed.

Thanks, Harold

On 11/10/2012 7:43 PM, David Holmes wrote:
> On 10/11/2012 5:11 AM, harold seigel wrote:
>> Please review the following change to fix bug 7122219.
>>
>> Summary:  This change prevents the SIGFPE and SIGSEGV problems that can
>> occur when using this option. It checks that the specified value is >=
>> the current default (1009) and <= the maximum requestable heap size, and
>> issues an error if it is not.
>>
>> This webrev also deletes an unused create_table() method and adds a
>> check for integer rollover to method os::malloc().
>
> Minor nit: we generally refer to overflow rather than rollover.
>
>> With this change, if users request an invalid value, they will get an
>> error message like this:
>>
>>     % $JAVA_HOME/bin/java -XX:StringTableSize=1g -cp .. 
>> <java-main-class>
>>     StringTable size of 1073741824 is invalid; must be between 1009 and
>
> Shouldn't that be "StringTableSize of ..."
>
> Was the option of renaming StringTableSize to something reflecting 
> that it is actually the bucket count rejected? I know there is a 
> follow on CR in this area but giving this variable a more accurate 
> name doesn't seem unreasonable.
>
> That said I'm okay if this is pushed as-is.
>
> David
>
>>     536870911
>>     Error: Could not create the Java Virtual Machine.
>>     Error: A fatal exception has occurred. Program will exit.
>
>
>
>> Open webrev at http://cr.openjdk.java.net/~hseigel/bug_7122219
>> <http://cr.openjdk.java.net/%7Ehseigel/bug_7122219>
>> Bug link at http://bugs.sun.com/view_bug.do?bug_id=7122219
>>
>> The changes were tested with JCK, JPRT, JTREG, and UTE tests, and with
>> hand-run tests on 32-bit Linux and 64-bit Windows.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/attachments/20121112/c4fe9580/attachment.html