On Wed Oct 29 21:47:26 2025 +0000, Paul Gofman wrote:
Also WRT setting last error I think it could also use some tests for other cases besides invalid handle due to special handle. Like, invalid handle which is a valid handle but not waitable, when there is STATUS_ACCESS_DENIED for wait, when there is timeout (maybe we already have some for the latter, but probably not for other error statuses as those would have been fixed todos). IIRC `STATUS_ACCESS_DENIED` is already correctly handled (both with inproc and server-side waits) and depends on if the opened handle has `SYNCHRONIZE` access.
I believe `STATUS_INVALID_HANDLE` is also returned for valid handles that are not waitable, but this could indeed use a bit more testing too. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/9305#note_120044