RFR: JDK-8304954: SegmentedCodeCache fails when using large pages [v4]

[Damon Fenacci]

On Mon, 24 Jul 2023 12:51:35 GMT, Thomas Stuefe <stuefe at openjdk.org> wrote:

>> Damon Fenacci has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   JDK-8304954: merge ifs checking when to print warning
>
> src/hotspot/share/code/codeCache.cpp line 321:
> 
>> 319:                    "Reverting to smaller page size (" SIZE_FORMAT "%s).",
>> 320:                    byte_size_in_exact_unit(LargePageSizeInBytes), exact_unit_for_byte_size(LargePageSizeInBytes),
>> 321:                    byte_size_in_exact_unit(ps), exact_unit_for_byte_size(ps));
> 
> You could use EXACTFMT and EXACTFMTARGS to shorten this, see globalDefinitions.hpp

Changed to make use of PROPERFMT and PROPERFMTARGS.

> test/hotspot/jtreg/compiler/codecache/CheckLargePages.java line 27:
> 
>> 25:  * @test
>> 26:  * @bug 8304954
>> 27:  * @summary Test checks that if using large pages and code cache gets above the limit it tries to revert to smaller pages instead of failing
> 
> Proposal: "Code cache reservation should gracefully downgrade to using smaller pages if the code cache size is too small to host the requested page size."

Changed the summary. Thanks!

> test/hotspot/jtreg/compiler/codecache/CheckLargePages.java line 54:
> 
>> 52:                     "-XX:ReservedCodeCacheSize=2g",
>> 53:                     "-XX:LargePageSizeInBytes=1g",
>> 54:                     "-Xlog:pagesize*=debug",
> 
> If all you scan for is the "Failed to reserve" (please change the text :-), then you should *not* specify Xlog, since its a warning, and we expect to see this warning unconditionally (UL: everything above "log_info" is printed unconditionally if no Log options are present).

`-Xlog:pagesize*=debug` removed.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/14903#discussion_r1273459776
PR Review Comment: https://git.openjdk.org/jdk/pull/14903#discussion_r1273459553
PR Review Comment: https://git.openjdk.org/jdk/pull/14903#discussion_r1273459132