On Tue, 28 Apr 2026 19:58:46 GMT, Frederic Parain <[email protected]> wrote:
>> Marc Chevalier has updated the pull request with a new target base due to a >> merge or a rebase. The pull request now contains 31 commits: >> >> - Move the flag around >> - Merge remote-tracking branch lworld into JDK-8381472 >> - Printing is gone again >> - Remove the old way >> - Fix asserts >> - Let new version do the job >> - Flag + using _layout_info >> - Printing is back >> - cosmetics >> - Typo in acmp >> - ... and 21 more: >> https://git.openjdk.org/valhalla/compare/cfda30f3...88740943 > > src/hotspot/share/classfile/classFileParser.cpp line 5401: > >> 5399: mask <<= leading_zeroes; >> 5400: assert(count_leading_zeros(mask) == 0, "fast acmp mask can be >> moved further!"); >> 5401: int offset = _layout_info->_payload_offset - leading_zeroes / >> BitsPerByte; > > The fast_acmp_offset is not guaranteed to be aligned in any particular way > (not on word boundary for instance). Is it going to be OK for code using it? Yes, the access is known to be mismatch and unaligned. It is already tested by anything whose payload is too short to keep the whole long busy. For instance, `Integer`. ------------- PR Review Comment: https://git.openjdk.org/valhalla/pull/2353#discussion_r3157493745
