[nwst-dev] New patch

[Jochen R. <jo...@pa...>]

OK, next installment:

Based on my testing with fbset and fbi the frame buffer now
supports the following modes:

640x480-60 8/16/32bpp
800x600-60 16bpp
1024x768-60 8/16bpp

I added some modes to /etc/fb.modes, for instance this 32bpp
mode here:

mode "640x480-60"
    # D: 25.175 MHz, H: 31.469 kHz, V: 59.94 Hz
    geometry 640 480 640 480 8
    timings 39722 48 16 33 10 96 2
endmode

mode "640x480-60-32"
    # D: 25.175 MHz, H: 31.469 kHz, V: 59.94 Hz
    geometry 640 480 640 480 32
    timings 39722 48 16 33 10 96 2
endmode

To change modes at the console, run fbset -a 640x480-60-32 . If this
does not work, then please email me the driver debug messages from
/var/log/messages. If possible, please change the #if 0 in line 1511
and 1515 of s3trio.c to #if 1 so I can see the register settings.

I had to patch fbi to get it to properly display in true color modes:

diff -urN -x config* -x Makefile* fbi-1.23/main.c fbi-1.23-j1/main.c
--- fbi-1.23/main.c     Mon Feb  4 04:34:09 2002
+++ fbi-1.23-j1/main.c  Wed Jun 11 03:45:24 2003
@@ -366,10 +366,16 @@
        case 15:
        case 16:
            for (out = 0; out < owidth; out++) {
+               unsigned short s;
                in = (out << 16) >> shift;
-               ptr2[out] = lut_red[buffer[in*3]] |
+               s = lut_red[buffer[in*3]] |
                    lut_green[buffer[in*3+1]] |
                    lut_blue[buffer[in*3+2]];
+#if 0
+               ptr2[out] = (s>>8)|(s<<8);
+#endif
+               ptr[2*out+0] = s;
+               ptr[2*out+1] = s>>8;
            }
            ptr2 += owidth;
            return (char*)ptr2;
@@ -385,9 +391,14 @@
        case 32:
            for (out = 0; out < owidth; out++) {
                in = (out << 16) >> shift;
+#if 0
                ptr4[out] = lut_red[buffer[in*3]] |
                    lut_green[buffer[in*3+1]] |
                    lut_blue[buffer[in*3+2]];
+#endif
+               ptr[4*out+2] = buffer[3*in+0];
+               ptr[4*out+1] = buffer[3*in+1];
+               ptr[4*out+0] = buffer[3*in+2];
            }
            ptr4 += owidth;
            return (char*)ptr4;

The X server seems to have a similar endianness problem. I am
investigating that at the moment.

When running X, make sure you set up the XFConfig-4 properly, see
linux/Documentation/fb/framebuffer.txt for instructions.

Patch is here:
http://networkstation.sourceforge.net/patch-2.4.19-ns1000-20030610.gz

Regards,
Jochen