RFR: 8254972: Fix pretouch chunk calculations

[Stefan Johansson]

On Tue, 27 Oct 2020 15:49:30 GMT, Thomas Schatzl <tschatzl at openjdk.org> wrote:

> Hi all,
> 
>   can I have reviews for this follow-up to JDK-8252221 where Kim noticed that the code in the pretouch task has some minor issues?
> 
> In particular, some code could be simplified and due to some inconsistencies, and used less than ideal number of workers. There is also a fix to avoid an overflow if you want to pretouch >= 16TB of heap with 4k pages (order of casts in calculation of ``num_workers``).
> 
> Testing: tier1
> 
> Thanks,
>   Thomas

Looks good, maybe just add an extra assert.

src/hotspot/share/gc/shared/pretouchTask.cpp line 37:

> 35:     _end_addr(end_address),
> 36:     _page_size(page_size),
> 37:     _chunk_size(chunk_size) { }

Add an assert here that chunk_size must be larger than or equal to page_size.

-------------

Marked as reviewed by sjohanss (Reviewer).

PR: https://git.openjdk.java.net/jdk/pull/884