Attached you’ll find a patch for the DirectFB driver of SDL CVS
to emulate the DVDmax feature of Matrox G450/G550 cards:
You see the output from the first head also on the TV.
-> This patch allows me to run Freevo / xmame / zsnes on my TV.
Why is this patch needed? Because the second head (CRTC2) got a
fixed resolution of 720x576 (PAL) or 720x480 (NTSC). Freevo f.e.
will only run in 800x600 or 640x480.
The new code is activated when you set the environment variable
SDL_DIRECTFB_MGA_CRTC=x.
The picture can also be stretched fit-to-screen with the environment
variable SDL_DIRECTFB_MGA_STRETCH=x.
Attached you’ll find a patch for the DirectFB driver of SDL CVS
to emulate the DVDmax feature of Matrox G450/G550 cards:
You see the output from the first head also on the TV.
-> This patch allows me to run Freevo / xmame / zsnes on my TV.
I’ve never used DirectFB, but the patch looks sane.
Can I get you to take out the printf() calls and “//” comments in the
patch, though? If DirectFB isn’t Linux-specific (or may work on other Unix
flavors in the future), then the “//” will confuse some not-C99 compiler
somewhere (this has happened with SDL before).
Attached you’ll find a patch for the DirectFB driver of SDL CVS
to emulate the DVDmax feature of Matrox G450/G550 cards:
You see the output from the first head also on the TV.
-> This patch allows me to run Freevo / xmame / zsnes on my TV.
I’ve never used DirectFB, but the patch looks sane.
One thing came to my mind during breakfast this morning:
Isn’t the output of the second head in DirectFB_FlipHWSurface
one frame behind? I first flip the “offscreen” memory and
then copy it. Shouldn’t I copy it first and the flip it?
Can I get you to take out the printf() calls and “//” comments in the
patch, though? If DirectFB isn’t Linux-specific (or may work on other Unix
flavors in the future), then the “//” will confuse some not-C99 compiler
somewhere (this has happened with SDL before).
Ok, I’ll change the “//” to “/* */”, but I would like
to keep the comments if possible
Also I’ll put the printf() calls in f.e. #ifdef DIRECTFB_CRTC2_DEBUG.
I’ve never used DirectFB, but the patch looks sane.
One thing came to my mind during breakfast this morning:
Isn’t the output of the second head in DirectFB_FlipHWSurface
one frame behind? I first flip the “offscreen” memory and
then copy it. Shouldn’t I copy it first and the flip it?
Ok, I wrote a test program and the code works the way it is.
-> First flip, then copy
Attached is the “final” version of the patch. I’ve fixed the comments, #ifdefed the printfs() and changed the indent of my code.
Should be ready for CVS
