mbxfb: Fix a chip bug? resulting in wrong pixclock

This is a workaround for what I think is a bug in the 2700G chip.

The PLL output frequency is adustable using 3 values (M, N and P. See code for formula). The N value range is documented to be 1 to 7 but when it is set to 1, the output frequency is lower than it should be (divided by 2), giving unexpected results such as no sync on a CRT display.

This patch prevents N=1 when searching for the best value for the requested pixclock.

This is a workaround for what I think is a bug in the 2700G chip.

The PLL output frequency is adustable using 3 values (M, N and P. See code for formula). The N value range is documented to be 1 to 7 but when it is set to 1, the output frequency is lower than it should be (divided by 2), giving unexpected results such as no sync on a CRT display.

This patch prevents N=1 when searching for the best value for the requested pixclock.

@@ -118,8 +118,19 @@
 	/* convert pixclock to KHz */
 	pixclock = PICOS2KHZ(pixclock_ps);
 
+	/* PLL output freq = (ref_clk * M) / (N * 2^P) 
+	 *
+	 * M: 1 to 63
+	 * N: 1 to 7
+	 * P: 0 to 7
+	 */
+	
+	/* RAPH: When N==1, the resulting pixel clock appears to
+	 * get divided by 2. Preventing N=1 by starting the following
+	 * loop at 2 prevents this. Is this a bug with my chip 
+	 * revision or something I dont understand? */	
 	for (m = 1; m < 64; m++) {
-		for (n = 1; n < 8; n++) {
+		for (n = 2; n < 8; n++) {
 			for (p = 0; p < 8; p++) {
 				clk = (ref_clk * m) / (n * (1 << p));
 				err = (clk > pixclock) ? (clk - pixclock) :

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642

_______________________________________________
Linux-fbdev-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-fbdev-devel

I just thought it might be useful to post information about how to reproduce this:

Install fbset and link /etc/fb.modes to the fb.modes.ATI distributed with fbset.

Do 'fbset "800x600-70" -depth 16' and the generated pixel clock will result in a v-sync around 25 hz. The 800x600-60 and 800x600-72 modes will work correctly, with the expected v-sync values.

After applying the patch, the "800x600-70" mode works correctly.

Here are the modelines from my fb.modes file, just in case:
mode "800x600-70"

     # D: 44.90 MHz, H: 44.544 kHz, V: 70.04 Hz
     geometry 800 600 800 600 8
     timings 22272 40 24 15 9 144 12
     hsync high

endmode

mode "800x600-60"

     # D: 40.00 MHz, H: 37.879 kHz, V: 60.32 Hz
     geometry 800 600 800 600 8
     timings 25000 88 40 23 1 128 4
     hsync high
     vsync high

endmode

mode "800x600-72"

     # D: 50.00 MHz, H: 48.090 kHz, V: 72.19 Hz
     geometry 800 600 800 600 8
     timings 20000 64 56 23 37 120 6
     hsync high
     vsync high

endmode

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642