RFR: 8377507: Store cpu features in AOTCodeCache [v3]

[Ashutosh Mehra]

On Wed, 11 Feb 2026 17:30:15 GMT, Vladimir Kozlov <kvn at openjdk.org> wrote:

>>> I don't think output needed. As I noticed in my testing, it is usually duplicate of recorded features you print next.
>> 
>> okay, I will remove that.
>> 
>>> May be in addition to report when features are not supported print which features does not match but we can still use AOT code.
>> 
>> @vnkozlov Did you mean print the extra features present in the the runtime env?
>
> @ashu-mehra yes, instead of "Available CPU features:" print "Additional available CPU features:" after you print recorded features.  Available CPU features printing is good but it is hard to see what is difference from what was recorded.
> Printing additional features allow to see difference clearly.

I added a log message to print additional runtime features. Now it prints a message like this:


[0.034s][debug][aot,codecache,init] CPU features recorded in AOTCodeCache: cx8, cmov, fxsr, mmx, 3dnowpref, sse, sse2, popcnt, lzcnt, tsc, tscinvbit, aes, erms, clmul, bmi1, bmi2, adx, sha, fma, clflush, clflushopt, clwb, serialize, rdtscp, rdpid, fsrm, gfni, pku, ospke, cet_ibt, cet_ss, hybrid
[0.034s][debug][aot,codecache,init] Additional runtime CPU features: sse3, ssse3, sse4.1, sse4.2, avx, avx2, vzeroupper, f16c


Does that look good?

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

PR Review Comment: https://git.openjdk.org/jdk/pull/29655#discussion_r2800861993