RFR: Store cpu features in AOTCodeCache header
Vladimir Kozlov
kvn at openjdk.org
Tue Jul 29 22:23:15 UTC 2025
On Tue, 29 Jul 2025 13:55:27 GMT, Andrew Dinn <adinn at openjdk.org> wrote:
>> @adinn
>>> However it assumes a lot of things fromthe (current) fact that this must be Apple silicon so things do not match that closely.
>>
>> I didn't follow your comment. Did you mean there is code that applies to Apple silicon but not to aarch64?
>>
>>> Interesting that none of the macos gate tests failed even though you changed nothing in this file . . .
>>
>> Yeah I am surprised I didn't get any failures to indicate the missing changes.
>
> @ashu-mehra I was merely indicating that the changes you made in vm_version_linux_aarch64.cpp are not simply able to be reapplied to vm_version_bsd_aarch64.cpp because the latter file checks things very differently.
>
> In particular, it bypasses a lot of the hardware checks on the assumption that this must be apple hardware (perhaps a questionable thing to rely on).
> @adinn @vnkozlov does this look good to go?
May be we should wait when #26515 is finalized to use in these changes.
Meanwhile I can re-submit testing with latest changes and see if timeouts indeed gone.
-------------
PR Comment: https://git.openjdk.org/leyden/pull/84#issuecomment-3134235190
More information about the leyden-dev
mailing list