Fwd: Request for review: 7173044: Memory monitor demo hangs if MemoryUsage obj returns -1 in getMax

David Holmes david.holmes at oracle.com
Sun Jun 3 20:26:30 PDT 2012


Hi Sean,

This seems reasonable. Perhaps add a couple of comments:

125  private float usedMemMax[]; // used when max pool size is undefined

202  if (totalMemory < 0) {  // max is undefined for this pool

David
-----


On 4/06/2012 12:12 PM, Sean Chou wrote:
> Hello,
>
>      Would any one like to take one more look ? It will help much !
>
> ---------- Forwarded message ----------
> From: *Sean Chou* <zhouyx at linux.vnet.ibm.com
> <mailto:zhouyx at linux.vnet.ibm.com>>
> Date: Thu, May 31, 2012 at 3:17 PM
> Subject: Request for review: 7173044: Memory monitor demo hangs if
> MemoryUsage obj returns -1 in getMax
> To: serviceability-dev at openjdk.java.net
> <mailto:serviceability-dev at openjdk.java.net>
>
>
> Hi all,
>
>      MemoryMonitor demo will hang if MemoryUsage obj returns -1
> in getMax method.
> The demo obtains totalMemory from getMax method, and it is used in
> following loop:
>
> "    // .. Memory Free ..
>       big.setColor(mfColor);
>       int MemUsage = (int) (((totalMemory - usedMemory) / totalMemory) *
> 10);
>       int i = 0;
>       for ( ; i < MemUsage ; i++) {
>            mfRect.setRect(x1+5,(float) y1+ssH+i*blockHeight,
>                      blockWidth, blockHeight-1);
>            big.fill(mfRect);
>       }
> "
>     If -1 is returned, MemUsage becomes a very large positive value and
> the loop hangs
> the demo.
>
>     The bug is not seen in current openjdk because the beans returned
> all has a Max
> value returned. However, -1 is allowed by spec when Max is not set for
> the bean, and
> other jdk might return.
>
>     In the patch, max is set to 1m or the max memory usage in history
> for the bean
> if there is no max specified. Please take a look.
>
> Bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7173044
> Webrev: http://cr.openjdk.java.net/~zhouyx/7173044/webrev.01/  .
>
>
> --
> Best Regards,
> Sean Chou
>
>
>
>
> --
> Best Regards,
> Sean Chou
>


More information about the serviceability-dev mailing list