RFR: 8195621: JTReg failure in gc/g1/TestLargePageUseForAuxMemory.java

[David Holmes]

On 24/01/2018 11:00 PM, Thomas Stüfe wrote:
> On Wed, Jan 17, 2018 at 11:43 PM, David Holmes <david.holmes at oracle.com 
> <mailto:david.holmes at oracle.com>> wrote:
> 
>     Hi Daniel,
> 
>     On 18/01/2018 3:41 AM, stewartd.qdt wrote:
> 
>         Please review this webrev [1] which attempts to fix a test error
>         in gc/g1/TestLargePageUseForAuxMemory when it is run on a
>         machine with 64K pages. See the bug report [2] for further
>         details of why the error occurs. Essentially the test tries to
>         allocate 512 times the huge page size and fails when the huge
>         page size is around ½ - 1 GB is size, because there simply isn't
>         enough physical memory in the system.
> 
> 
>         This fix adds a function to WhiteBox to get the physical memory
>         and then uses that to ensure we are not trying to allocate more
>         than the physical amount of memory for our heap. Please let me
>         know if this is a bad approach and I'll be happy to change as
>         required.
> 
> 
>     This seems an improvement though I would expect the test to fail
>     long before we get close to the amount of physical memory available.
>     Should the test be adjusting what it does based on the page size
>     being used, in conjunction with the amount of "physical memory"?
> 
> 
> I am not even sure os::physical_memory() would be meaningful on Linux 
> with huge pages.

That's why I said "physical memory" not os::physical_memory() :)

> Would the real limiting factor not be the size of the huge page pool 
> (vm.nr_hugepages) ? If that one is too small, or the user has no access 
> rights to the pool, you would get huge page allocation errors regardless 
> what physical memory is in the system, no?

I don't know. Thanks for raising the question.

Cheers,
David

> Kind Regards, Thomas
> 
>     Hopefully one of the GC folk will chime in.
> 
>     Thanks,
>     David
> 
> 
> 
> 
>         Thanks,
> 
>         Daniel Stewart
> 
> 
>         [1] - http://cr.openjdk.java.net/~dstewart/8195621/webrev.00/
>         <http://cr.openjdk.java.net/~dstewart/8195621/webrev.00/>
>         [2] - https://bugs.openjdk.java.net/browse/JDK-8195621
>         <https://bugs.openjdk.java.net/browse/JDK-8195621>
> 
>