RFR: 8325133: Missing MEMFLAGS parameter in parts of os API [v2]
Johan Sjölen
jsjolen at openjdk.org
Mon Feb 5 09:30:10 UTC 2024
On Thu, 1 Feb 2024 15:54:24 GMT, Johan Sjölen <jsjolen at openjdk.org> wrote:
>> Hi,
>>
>> Some parts of the `os` API calls NMT functions without taking a `MEMFLAGS` as a parameter. This PR extends the API of these functions to change this. No callers are changed.
>
> Johan Sjölen has updated the pull request incrementally with one additional commit since the last revision:
>
> Missing executable argument
Thank you for the reviews, I've answered your questions below.
> Just curious if the PR is compiled and tested on all platforms?
No, but it works for all of our GHA tested platforms, which is sufficient for me. No way for me to check AIX, for example :-).
> Looks good.
>
> I only have one feedback that frankly I myself am not 100% sure we need to address.
> This seems fine to me. I might be good to think about pushing this further and remove some of the introduced default values.
First off: Yes, we should not have default values for these functions. I am not going to fix for example `ReservedSpace` in this PR, that's for a future RFE. Second of all, introducing these default values is the same as what we do for `os::reserve_memory`, so this is established behavior, but not preferable behavior as mentioned.
-------------
PR Comment: https://git.openjdk.org/jdk/pull/17672#issuecomment-1926549238
More information about the hotspot-runtime-dev
mailing list