On Wed, 6 Dec 2023 17:03:08 GMT, Maurizio Cimadamore <mcimadam...@openjdk.org> wrote:
>> This PR fixes a couple of aligned accesses when reading/writing strings. >> Such aligned accesses crept in when we optimized string read/write >> operations to work in bulk. As a result, depending on the maximum alignment >> constraints supported by the heap segment, some string operations might fail. >> >> I've added some tests to make sure that all operations work as expected with >> unaligned semantics. >> >> Note: I've considered inferring an alignment constraint from the provided >> charset, and then use aligned operations (and document that behavior), but I >> found that to be unsatisfactory: memory operations typically accept a >> layout, which allow clients to opt out from alignment checks if needed. But >> if we always infer an alignment constraint from the provided charset, >> clients would find themselves w/o an escape hatch. For this reason, I think >> the best way to fix this is to use unaligned operations when reading/writing >> the string. > > Maurizio Cimadamore has updated the pull request incrementally with two > additional commits since the last revision: > > - Make test more robust > - Simplify test LGTM. Testing looks solid. ------------- Marked as reviewed by pminborg (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/16999#pullrequestreview-1769404275