RFR (S) 8223770: code_size2 still too small in some compressed oops configurations

[Doerr, Martin]

Hi Ralf,

looks good. Seems like you're the only user who really stresses the stub code generation in debug build on a modern Windows machine.

Thanks and best regards,
Martin


-----Original Message-----
From: hotspot-compiler-dev <hotspot-compiler-dev-bounces at openjdk.java.net> On Behalf Of Schmelter, Ralf
Sent: Montag, 13. Mai 2019 17:06
To: hotspot-compiler-dev at openjdk.java.net
Subject: [CAUTION] RFR (S) 8223770: code_size2 still too small in some compressed oops configurations

Hi,

this is an addition to https://bugs.openjdk.java.net/browse/JDK-8223617. It turns out, that the additionally given space is not always enough to not trigger the assertion. This is caused by some of the stub routines needing more memory (93 bytes overall), when the compressed oops mode used a non-zero disjoint base. To be on the safe side I've added another 200 bytes to the code_size2 constant and ran the hotspot/jtreg/gc/arguments/TestUseCompressedOopsErgo.java jtreg test, which showed the problem consistently before the change.

webrev: http://cr.openjdk.java.net/~rschmelter/webrevs/8223770/webrev.0/ 
bugreport: https://bugs.openjdk.java.net/browse/JDK-8223770

Best regards,
Ralf