In https://bugs.openjdk.org/browse/JDK-8344883 we removed the default parameter value for mem_tag and forced in all the code to use an explicit value, i.e. mtNone.
We now will have to try and do our best to find actual values to replace all of the remaining mtNones.
Stefan recommended in https://github.com/openjdk/jdk/pull/24282 to change all mtNone to mtTest (mostly in gtests, which on surface seems very reasoanble), but that actually caused the build to break, and if I remember correctly even if we get the build to work there will be bunch of tests that will fail on top of that, we will need to proceed with caution.
We now will have to try and do our best to find actual values to replace all of the remaining mtNones.
Stefan recommended in https://github.com/openjdk/jdk/pull/24282 to change all mtNone to mtTest (mostly in gtests, which on surface seems very reasoanble), but that actually caused the build to break, and if I remember correctly even if we get the build to work there will be bunch of tests that will fail on top of that, we will need to proceed with caution.
- relates to
-
JDK-8353854 NMT: change map_memory() signature to match other functions
-
- Open
-
-
JDK-8344883 Force clients to explicitly pass mem_tag value, even if it is mtNone
-
- Resolved
-