RFR: Fix SCCache stats logged on exit [v3]

[Ashutosh Mehra]

On Thu, 20 Feb 2025 22:38:27 GMT, Ashutosh Mehra <asmehra at openjdk.org> wrote:

>> The count of various SCCEntry kinds is not correctly computed. This patch fixes it.
>> This code was introduced as part of i2c2i adapters and was picked up from Andrew Dinn's work on storing stub and blobs. Clearly I messed it up.
>> The patch also computes and shows the counts for old entries and new entries separately. Though not useful for now, but in incremental workflow it may be helpful to distinguish between the old and new entries.
>
> Ashutosh Mehra has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Fix compile failures
>   
>   Signed-off-by: Ashutosh Mehra <asmehra at redhat.com>

I have pushed another change that refactors the way stats are collected and do it in a more organized way rather than in a ad-hoc manner. I have separated the stats into two categories - stats for the code as stored in the AOT Cache, and stats reflecting runtime characteristic of the code.
Note that it does not change the stats that are printed on `-Xlog:init`, but it gives a slightly better breakdown on exit (`-Xlog:scc+exit`) with stats for each compilation tier. So now on exit it shows following:


[7.544s][info   ][scc,exit] Wrote 2209 SCCEntry entries(181748 max size) to AOT Code Cache
[7.544s][info   ][scc,exit]   Adapter: total=545(old=0+new=545)
[7.544s][info   ][scc,exit]   Stub: total=3(old=0+new=3)
[7.544s][info   ][scc,exit]   Blob: total=1(old=0+new=1)
[7.544s][info   ][scc,exit]   Code: total=1660(old=0+new=1660)
[7.544s][info   ][scc,exit]     Tier 1: total=103(old=0+new=103)
[7.544s][info   ][scc,exit]     Tier 2: total=1176(old=0+new=1176)
[7.544s][info   ][scc,exit]     Tier 4: total=191(old=0+new=191)
[7.544s][info   ][scc,exit]     Tier 5: total=190(old=0+new=190)
[7.544s][info   ][scc,exit] Total=2209(old=0+new=2209)

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

PR Comment: https://git.openjdk.org/leyden/pull/42#issuecomment-2672860290