https://bugs.winehq.org/show_bug.cgi?id=53682
--- Comment #24 from Martin Storsjö martin@martin.st --- (In reply to Martin Storsjö from comment #23)
(In reply to Martin Storsjö from comment #22)
(In reply to Zeb Figura from comment #21)
If I'm reading this right, Alexandre fixed this with the "build the whole frame in assembly" route in https://source.winehq.org/git/wine.git/commitdiff/ 4069a8b384c15a90a7af66636ec5650eeb53b391 and previous commits.
Yes, it seems so. However it seems like something is wrong with this commit
- in builds where I didn't run into this issue at all before, I'm now seeing
the "user_check_not_lock BUG: holding USER lock" message on startup. However it's less fatal than in the cases with GCC where it failed outright directly on startup; the wineprefix setup does seem succeed somewhat, but some process hits this.
So all is not well yet. I'll see when I have time to dig into it...
With https://gitlab.winehq.org/wine/wine/-/merge_requests/1318, the regression is fixed, and then it does indeed look like this issue is fixed as well!
With this merged now, it does look to me like this is issue is fixed. (I had a setup where I could reproduce the issue well, and it seems to be fixed for me there.) Kevin, can you reconfirm it from your point too?
(The main fix landed in 4069a8b384c15a90a7af66636ec5650eeb53b391 and the follow-up fix in 464d3c86dcb1bd057bde84765e4b2ca1aafaec23.)