Re: [MPlayer-dev-eng] [PATCH] Direct3D libvo driver
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I've been banging my head against the wall with this one. It still doesn't work in vista (-vo gl does, so it's defiantly the vo module causing the issue) I stripped it back to the bare minimum and the crash seems to happen on /* vo_w32_window is w32_common variable. It's a handle to the window. */ if (FAILED (IDirect3D9_CreateDevice(gpD3DHandle, D3DADAPTER_DEFAULT, D3DDEVTYPE_HAL, vo_w32_window, D3DCREATE_SOFTWARE_VERTEXPROCESSING, /* D3DDEVCAPS HWTRANSFORMANDLIGHT */ &PresentParams, &gpD3DDevice))) { mp_msg(MSGT_VO,MSGL_ERR, "<vo_direct3d><INFO>Could not create the D3D device\n"); return FALSE; } If I include this then crash, if I leave it out then I get the window + audio playing but obviously no video. Any ideas anyone? As a general comment, I've been doing some looking about. I think that D3DRenderFrame() is probably missing a IDirect3DDevice9_BeginScene(gpD3DDevice); IDirect3DDevice9_EndScene(gpD3DDevice); pair, for correctness. Make sure the exit points have the requisite IDirect3DDevice9_EndScene(gpD3DDevice); Jim -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkkiGv8ACgkQhrNWoHjgI1B4wQCbBPWRPWUepUVtq7hNZPvJxXEi 10wAoLnZvQi3Xv/awzsEuk2+k3Euje0m =6mXv -----END PGP SIGNATURE-----
On Tue, 18 Nov 2008 01:31:43 +0000, Jim Hauxwell wrote:
If I include this then crash, if I leave it out then I get the window + audio playing but obviously no video.
Any ideas anyone?
can you make a gdb ? instructions are in the docs under bugreports. what video card? i wonder if its just missing one of those variables? -compn
Very interesting. I'm investigating this issue. What's your video card btw?
Thanks Jim - I've added Begin/End scene in D3DRenderFrame as well as draw_slice. As I understand I should call Begin/End Scene before/after locking the surface and stretchrect-ing inside. Later when flip_page is called and Present is executed, I'll have already called EndScene, so Present will happen after it. This doesn't seem to slow down things as well, so I leave it like that. It would be great if you can test the new code. I'll submit a new patch very soon!
participants (3)
-
compn -
Georgi Petrov -
Jim Hauxwell