On Mon Jan 30 18:27:02 2023 +0000, Alexandros Frantzis wrote:
@nekotekina Thanks, I understand the issue better now (in my last paragraphs of my original message I mixed up `data` with `handles` when thinking about code paths leading to this, so please ignore those). However, I now wonder if what you describe is a problem at all. It seems to me that everything is working as expected, and the 0xffffffff GetLastError you see is in fact not relevant:
NtUserPeekMessage wait_message count=0 handles=NULL X11DRV_MsgWaitForMultipleObjectsEx count=0 handles=NULL => -1 (0xffffffff) => WAIT_FAILED (0xffffffff) => FALSE, since wait_message() returned != WAIT_TIMEOUT (indicating that we received driver events) the code calls peek_message() again, finds nothing and returns FALSE, due to: ret = wait_message( 0, NULL, 0, QS_ALLINPUT, 0 ); /* if we received driver events, check again for a pending message */ if (ret == WAIT_TIMEOUT || peek_message( &msg, hwnd, first, last, flags, 0 ) <= 0) return FALSE;So, this sequence seems that's just part of normal operation, and a red herring for your investigation?
Is PeekMessage supposed to SetLastError at all? Because that's what happens in the game, and the value 0xffffffff is read much later in an unrelated code, causing the crash. I think the regression was introduced relatively recently by commit ac9b63580018354c5a7c9164bb5c781b9e2b6cbf. MSDN doesn't seem to say anything about GetLastError in PeekMessageW.