On Thu, 21 May 2026 13:40:22 GMT, Thomas Stuefe <[email protected]> wrote:
>There is also a question of interface cutting: seems weird to have an API for >resident region iteration in os:: instead of just an API that, given a range, >returns size of residency. After all, this is what the sole user in NMT does. >Might as well do that iteration down in os scope. I agree that would feel much nicer. >But thats for a future RFE. Question, what happens if we have multiple >disjunct resident regions in a stack? I think that if there are multiple disjunct resident parts we eventually account for all of them. `SnapshotThreadStackWalker` ends up calling `first_resident_in_range` multiple times in a loop until it has processed the entire `VirtualMemoryRegion`. In a future improvement it probably makes sense to wrap this all in a function like you suggest: `resident_in_range` which walks an entire given region summing up all the potentially disjunct resident portions. ------------- PR Comment: https://git.openjdk.org/jdk/pull/31124#issuecomment-4510918101
