RFR: 8313306: More sensible memory reservation logging [v12]

[Thomas Stuefe]

On Wed, 13 Mar 2024 14:38:47 GMT, Sonia Zaldana Calles <szaldana at openjdk.org> wrote:

>> Do you need more than the System.err output?
>> 
>>  stdout: [[0.002s][debug][os,map] Reserved [0x00007fffb1960000 - 0x00007fffb1970000), (65536 bytes)
>> [0.002s][debug][os,map] Committed [0x00007fffb1960000 - 0x00007fffb1970000), (65536 bytes)
>> [0.004s][debug][os,map] Committed [0x00007fffaec10000 - 0x00007fffaec40000), (196608 bytes)
>> [0.005s][debug][os,map] Reserved [0x00007fff90a40000 - 0x00007fff9fa40000), (251658240 bytes)
>> [0.005s][debug][os,map] Committed [0x00007fff97e00000 - 0x00007fff98000000), (2097152 bytes)
>> [0.005s][debug][os,map] Reserved [0x00007fffae2c0000 - 0x00007fffae2e0000), (131072 bytes)
>> [0.005s][debug][os,map] Committed [0x00007fffae2c0000 - 0x00007fffae2d0000), (65536 bytes)
>> [0.005s][debug][os,map] Committed [0x00007fff90a40000 - 0x00007fff90c40000), (2097152 bytes)
>> [0.005s][debug][os,map] Reserved [0x00007fffae1d0000 - 0x00007fffae2c0000), (983040 bytes)
>> [0.005s][debug][os,map] Committed [0x00007fffae1d0000 - 0x00007fffae1e0000), (65536 bytes)
>> [0.006s][debug][os,map] Committed [0x00007fff98670000 - 0x00007fff98870000), (2097152 bytes)
>> [0.006s][debug][os,map] Reserved [0x00007fffae0e0000 - 0x00007fffae1d0000), (983040 bytes)
>> [0.006s][debug][os,map] Committed [0x00007fffae0e0000 - 0x00007fffae0f0000), (65536 bytes)
>> [0.019s][debug][os,map] Reserved [0x00000000fe000000 - 0x0000000100000000), (33554432 bytes)
>> [0.019s][debug][os,map] Reserved [0x00007fffaeba0000 - 0x00007fffaebb0000), (65536 bytes)
>> [0.019s][debug][os,map] Reserved [0x00007fffac0f0000 - 0x00007fffac100000), (65536 bytes)
>> [0.019s][debug][os,map] Reserved [0x00007fffac070000 - 0x00007fffac0f0000), (524288 bytes)
>> [0.019s][debug][os,map] Reserved [0x00007fffac060000 - 0x00007fffac070000), (65536 bytes)
>> [0.020s][debug][os,map] Reserved [0x00007fff8d640000 - 0x00007fff8f640000), (33554432 bytes)
>> [0.020s][debug][os,map] Committed [0x00007fff8d640000 - 0x00007fff8f640000), (33554432 bytes)
>> [0.035s][debug][os,map] Reserved [0x0000000800000000 - 0x0000000841000000), (1090519040 bytes)
>> [0.035s][debug][os,map] Released [0x0000000800000000 - 0x0000000841000000), (1090519040 bytes)
>> [0.035s][debug][os,map] reserve_between (range [0x0000000100000000-0x0001000000000000), size 0x41000000, alignment 0x100000000, randomize: 1)
>> [0.035s][trace][os,map] before reserve, 32 Points: 40 32120 1814 30964 2929 30001 3184 29643 4112 27899 5148 26897 6884 26456 7802 24834 8193 23633 9418 23061 10247 22165 11949 20547 12790 19711 14147 18647 14746 17802 15828 17...
>
> Hi @TheRealMDoerr, I think I've sorted out the errors you reported. Root cause for the linux ppc64le failure was in the test set up. 
> 
> Could you possibly run the tests again when you have a chance? 
> 
> Thank you for your help!

@SoniaZaldana About the AIX bug, this is the VMA accounting we do in AIX. It is highly likely that it complains about the sequence of invalid operations you throw as a test to the JVM. My proposal would be to disable this test for AIX and just let the AIX guys sort this out. After all, this is not a real issue. It may also be that in the future we don't need this bookkeeping anymore, there had been discussions around that with Joachim.

For testAttemptedReserveFailed, I would do:

- reserve anywere
- then attempt reserve at the same address

that will fail, but this should not assert, since the address in question is valid for attaching, its just occupied.

Other than that, to move this forward, I would reduce the tests for any remaining errors. You have done a lot more than your due diligence to get this running.

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

PR Comment: https://git.openjdk.org/jdk/pull/17356#issuecomment-1996838409