On Fri, 21 Nov 2025 15:52:48 GMT, Paul Hübner <[email protected]> wrote:
>> Hi all, >> >> This patch fixes several `int` overflows relating to flat array element >> loads/stores. >> I've included some gtests that address a subset of these changes. We should >> test flat arrays larger than 16GB in size, which is not something we can >> feasibly do in our tier1 jtreg tests. I've filed >> [JDK-8372261](https://bugs.openjdk.org/browse/JDK-8372261) as a follow-up >> RFE to investigate this. >> >> Testing: tiers 1-4, >> [`LocalityCheck.java`](https://bugs.openjdk.org/secure/attachment/117136/LocalityCheck.java) >> run locally 100 times. > > Paul Hübner has updated the pull request incrementally with one additional > commit since the last revision: > > C1 changes This looks good. Did you have a jvmci change still? src/hotspot/share/oops/flatArrayOop.inline.hpp line 42: > 40: > 41: address addr = (address) base(); > 42: addr += (index << Klass::layout_helper_log2_element_size(lh)); Because index is an int, this could go negative right? ------------- Marked as reviewed by coleenp (Committer). PR Review: https://git.openjdk.org/valhalla/pull/1752#pullrequestreview-3493388550 PR Review Comment: https://git.openjdk.org/valhalla/pull/1752#discussion_r2550229139
