RFR (XXS): 8024396: VM crashing with assert(!UseLargePages || UseParallelOldGC || use_large_pages) failed: Wrong alignment to use large pages

Thomas Schatzl thomas.schatzl at oracle.com
Thu Sep 12 08:08:18 PDT 2013


Hi,

On Wed, 2013-09-11 at 18:11 +0200, Stefan Karlsson wrote:
> On 09/11/2013 05:50 PM, Thomas Schatzl wrote:
> > hi,
> >
> > On Wed, 2013-09-11 at 17:29 +0200, Stefan Karlsson wrote:
> >> On 09/11/2013 04:54 PM, Thomas Schatzl wrote:
> >>> Hi all,
> >>>
> >>>     can I have reviews for the following tiny issue: while initializing
> >>> the heap a failing assertion is hit.
> >>>
> >>> Instead of checking whether any parallel gc combination is allowed to
> >>> have a non-large page aligned heap when large pages are used, it only
> >>> verifies that ParallelOldGC is set.
> >>> Since -XX:+ParallelGC -XX:-ParallelOldGC is a valid combination, the
> >>> assert fails in this case.
> >>>
> >>> I added a test case that checks successful startup for all gcs with and
> >>> without UseLargePages. Not sure if it is needed to do that for all gcs,
> >>> as this bug is really specific to parallel gc, and I'm somewhat
> >>> reluctant due to the added number of gc invocations - any comments?
> >>>
> >>> Webrev:
> >>> http://cr.openjdk.java.net/~tschatzl/8024396
> >> The fix looks good.
> >>
> >> Most of our test names are prefixed with "Test". Maybe rename the new
> >> test to TestAlignmentToUseLargePages?
> >>
> > Done. (Same webrev).
> 
> Looks good.


Thanks Stefan and Bengt!

Thomas





More information about the hotspot-gc-dev mailing list