https://bugs.winehq.org/show_bug.cgi?id=55710
--- Comment #31 from Dmitry Timoshkov dmitry@baikal.ru --- (In reply to Paul Gofman from comment #30)
Well, from the present info it is hard to guess what Wine or Staging change could cause it (if any). The patch in question catches syscalls based on address separation since long ago, nothing changed. It was disabled for some time (due to need for upstream rebase) and re-enabled not so long ago, maybe around the time of this breakage, so maybe system change which affected it happened earlier.
And yes, the way it separates syscalls is not a full proof solution by any means, it may be broken by certain non-standard (these days) setups, like, e. g., those two mentioned in the patch, there not much of sane options of fixing this specific problem (only if maybe additionally check for .elf loader address to disable patch effect if it is low, but that looks like rather arbitrary check very specific to this particular setup; it already checks for libc address which catches more common legacy VM configs in general).
Paul, do you have any suggestions or an idea how to make the patch work on systems like the one I have here? Or the only solution is to disable the patch?