https://bugs.winehq.org/show_bug.cgi?id=55154
Bug ID: 55154 Summary: Sacred 2 crashes due to missing 'acos' handling Product: vkd3d Version: 1.8 Hardware: x86-64 OS: Linux Status: NEW Severity: normal Priority: P2 Component: hlsl Assignee: wine-bugs@winehq.org Reporter: gijsvrm@gmail.com Distribution: ---
Continuation of bug 55084.
0024:err:d3dcompiler:D3DCompile2 Failed to compile shader, vkd3d result -4. 0024:err:d3dcompiler:D3DCompile2 Shader log: 0024:err:d3dcompiler:D3DCompile2 <anonymous>:123:13: E5005: Function "acos" is not defined. 0024:err:d3dcompiler:D3DCompile2
https://bugs.winehq.org/show_bug.cgi?id=55154
--- Comment #1 from Nikolay Sivov bunglehead@gmail.com --- This function is implemented now, https://source.winehq.org/git/vkd3d.git/commit/69294c290bad83470e55046f02e2d..., needs a retest.
https://bugs.winehq.org/show_bug.cgi?id=55154
--- Comment #2 from Gijs Vermeulen gijsvrm@gmail.com --- The message is gone now, so I don't know if that is enough?
Just as an aside:
How does one know if these are "really" fixed? In the case of 'tan' (bug 54377) it seemed like it was fixed because the message was gone and 'tan' was at least being parsed, but it wasn't really fixed.
Can that be the case for this bug and bug 55453 as well for example?
https://bugs.winehq.org/show_bug.cgi?id=55154
--- Comment #3 from Nikolay Sivov bunglehead@gmail.com --- Those are different. For 'tan' we have now sin(x) / cos(x), where sin and cos are expected to be supported as separate instructions. 'acos' however uses approximation. If you don't get compilation errors, we can consider it working for now.
https://bugs.winehq.org/show_bug.cgi?id=55154
Gijs Vermeulen gijsvrm@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Fixed by SHA1| |69294c290bad83470e55046f02e | |2d1c6167150cb Resolution|--- |FIXED
--- Comment #4 from Gijs Vermeulen gijsvrm@gmail.com --- Ah I see. Judging from the bugs 'sin' and 'cos' are not yet available for older shader models, so that would make sense, thanks!
In that case, marking FIXED.
https://bugs.winehq.org/show_bug.cgi?id=55154
--- Comment #5 from Nikolay Sivov bunglehead@gmail.com --- Yes, those should use approximation as well, depending on profile. Sometimes even when corresponding instructions are available I believe.
https://bugs.winehq.org/show_bug.cgi?id=55154
Nikolay Sivov bunglehead@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #6 from Nikolay Sivov bunglehead@gmail.com --- Closing bugs fixed in 1.11.