Skip to content

Interpreter: Fix a number of issues around endianness handling, including #25791

Andreas Klebinger requested to merge wip/andreask/bytecode_endian into master

This MR does a number of things. For now:

  • Fix #25791 by reverting 4f02d3c1
  • Fix the order in which we put sub word literals on the stack on big endian targets. (Likely the root cause for #23387 (closed))

I plan to fix the rest of the issues laid out in #25791 (comment 616925) as well.

I do not assume this will fix all big endian isses in the interpreter but should be a good step forward.

Merge request reports