http://bugs.winehq.org/show_bug.cgi?id=58623 --- Comment #9 from Bernhard Übelacker <bernhardu@mailbox.org> --- Still visible with wine-11.9. This is the callstack of the RaiseException, with some symbols visible to gdb: (rr) bt #0 RaiseException () from dlls/kernelbase/x86_64-windows/kernelbase.dll #1 taskbarunpin () from msi78fd.tmp #2 taskbarunpin () from msi78fd.tmp #3 ?? () from msi78fd.tmp #4 ?? () from msi78fd.tmp #5 ?? () from msi78fd.tmp #6 SetStartMenuWindows10T () from msi78fd.tmp #7 custom_proc_wrapper at dlls/msi/custom.c:482 #8 wine_msi_call_dll_function at dlls/msi/custom.c:554 #9 custom_action_thread (arg=<optimized out>) at programs/msiexec/msiexec.c:407 #10 BaseThreadInitThunk () from dlls/kernel32/x86_64-windows/kernel32.dll If hacking PathFileExistsW to return success with an empty string, the installation also succeeds. But that is not native behaviour of PathFileExistsW. The paramter to the first succeeding PathFileExistsW is retrieved by two calls to MsiGetPropertyW, one to get the needed buffer size, and one with a big enough buffer. But I could not find out where the path of the second PathFileExistsW gets retrieved. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.