https://bugs.winehq.org/show_bug.cgi?id=48322
--- Comment #19 from Luca Boccassi luca.boccassi@gmail.com --- (In reply to Sveinar Søpler from comment #18)
(In reply to Luca Boccassi from comment #17)
Rather than bisecting the patchsets, I'm thinking it might be easier to bisect the staging git tree itself, with 4.17 being "good" and 4.18 being "bad"? In _theory_, unless intermittent breakages were added and then removed, it should be doable as both versions work with WoW.
I am not sure it is "as easy as that", but there was some strangeness that started happening with 4.16 for some games aswell. I am sure this is not ONLY WoW with this problem tho.
This is a bug posted, now marked as solved: https://bugs.winehq.org/show_bug.cgi?id=47834
This is what i gather as something that caused problems aswell: https://github.com/wine-staging/wine-staging/commit/ 0c7512f5f53deb168d3508bdd1f1107ecee774c6#comments
And it was disabled pre 4.18 launch: https://github.com/wine-staging/wine-staging/commit/ 5b066d6aed7fd90c0be0a2a156b0e5c6cbb44bba
But i see no tests mentioning WoW in this particular context.
Maybe do some comparison of staging-4.17 vs. 4.17 with these patches enabled/disabled? I think there was changes to the wine tree between 4.17 and 4.18 that prompted the need to update this user32-rawinput patchset tho, but it is way beyond me to figure out the ins and outs of this.
We are playing the "blame-game" here: Is it wine, or is it staging :)
And the winner is... wine! :-)
I have bisected the wine-staging tree, and the first bad commit was the 4.19 release. That didn't update any patch, but it changed wine's upstream commit from ccec532879ec14b2e79da08288152a69221ec4d1 to 6f2ef158b78f7d0a950c9398c26278d6523c112e. Sure enough, while keeping the wine-staging patches identical, switching the base wine tree between those two commits makes the issue appear/go away.
So I bisected that too, and the culprit is:
https://source.winehq.org/git/wine.git/commit/413aad39135b0b0f8255500b85fcc0... https://github.com/wine-mirror/wine/commit/413aad39135b0b0f8255500b85fcc0533... https://bugs.winehq.org/show_bug.cgi?id=47821
With wine at wine-4.19 plus a revert of that commit, and wine-staging at v4.19, the bug is fixed. Without the revert, the bug is 100% reproducible.
So, what's next? I'm really not familiar with the codebase, so I can't see what the issue is.