Skip to contentSkip to conversation

neofb: avoid resetting display config on unblank

PATCH New Christian Trefzer 3 participants 6 messages Feb 8, 2006 – Feb 11, 2006 Framebuffer layer

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 neofb: avoid resetting display config on unblank

This patch fixes issues with the NeoMagic framebuffer driver.

It nicely complements my previous fix already in linus' tree. The only thing missing now is that the external CRT will not be activated at neofb init when external-only is selected, either by register read or module/kernel parameter.

Testing was done on a Dell Latitude CPi-A/NM2200 chip.

Previous behaviour:
- before booting linux, set the preferred display config X via FN+F8

- boot linux, neofb stores the register values in a private variable

- change the display config to Y via keystroke

- leave the machine in peace until display is blanked

- touching any key will result in display config X being restored

- booting up, the BIOS will acknowledge config Y, though...

Current behaviour:
At the time of unblanking, config Y is honoured because we now read back register contents instead of just overwriting them with outdated values.

Signed-off by: Christian Trefzer <[email protected]>

---

drivers/video/neofb.c renamed from drivers/video/neofb.c.orig +4 -1
@@ -1334,6 +1334,9 @@
 	struct neofb_par *par = (struct neofb_par *)info->par;
 	int seqflags, lcdflags, dpmsflags, reg;
 
+	/* Reload the value stored in the register, might have been changed via FN keystroke */
+	par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	
 	switch (blank_mode) {
 	case FB_BLANK_POWERDOWN:	/* powerdown - both sync lines down */
 		seqflags = VGA_SR01_SCREEN_OFF; /* Disable sequencer */
@@ -1366,7 +1369,7 @@
 	case FB_BLANK_NORMAL:		/* just blank screen (backlight stays on) */
 		seqflags = VGA_SR01_SCREEN_OFF;	/* Disable sequencer */
 		lcdflags = par->PanelDispCntlReg1 & 0x02; /* LCD normal */
-		dpmsflags = 0;			/* no hsync/vsync suppression */
+		dpmsflags = 0x00;	/* no hsync/vsync suppression */
 		break;
 	case FB_BLANK_UNBLANK:		/* unblank */
 		seqflags = 0;			/* Enable sequencer */
#1 lore raw

PATCH 1/1 neofb: add more logic to determine sensibility of register readback

Hello everyone,

testing my latest patches through productive use I noticed that we require a bit of a logic for the register read to not cause trouble at the time of unblanking. Thing is, blanking is implemented through deactivation of the LCD activation bit, so if we read back the register, the LCD backlight won't come back on until the LID is shut and reopened, effectively pushing the work to the BIOS.

The following patch is on top of my previous one, adding a variable to struct neofb_par to remember if we want to read the respective register values next time we (un)blank.

Logic depends on the assumption that we won't change display config in a blanked state, as a keypress should already be answered by unblanking. Essentially, we should read back the register values to variable on blank, and ignore them on unblank. Unfortunately, only checking for !blank_mode won't avoid the backlight remaining off, hence the additional checks.

include/video/neomagic.h +1 -0
@@ -159,6 +159,7 @@
 	unsigned char PanelDispCntlReg1;
 	unsigned char PanelDispCntlReg2;
 	unsigned char PanelDispCntlReg3;
+	unsigned char PanelDispCntlRegRead;
 	unsigned char PanelVertCenterReg1;
 	unsigned char PanelVertCenterReg2;
 	unsigned char PanelVertCenterReg3;
drivers/video/neofb.c +10 -2
@@ -1334,8 +1334,11 @@
 	struct neofb_par *par = (struct neofb_par *)info->par;
 	int seqflags, lcdflags, dpmsflags, reg;
 
-	/* Reload the value stored in the register, might have been changed via FN keystroke */
-	par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	/* Reload the value stored in the register, if sensible. It might have been
+	 * changed via FN keystroke. */
+	if (par->PanelDispCntlRegRead && !blank_mode) {
+		par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	}
 	
 	switch (blank_mode) {
 	case FB_BLANK_POWERDOWN:	/* powerdown - both sync lines down */
@@ -1355,21 +1358,25 @@
 			tosh_smm(&regs);
 		}
 #endif
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_HSYNC_SUSPEND:		/* hsync off */
 		seqflags = VGA_SR01_SCREEN_OFF;	/* Disable sequencer */
 		lcdflags = 0;			/* LCD off */
 		dpmsflags = NEO_GR01_SUPPRESS_HSYNC;
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_VSYNC_SUSPEND:		/* vsync off */
 		seqflags = VGA_SR01_SCREEN_OFF;	/* Disable sequencer */
 		lcdflags = 0;			/* LCD off */
 		dpmsflags = NEO_GR01_SUPPRESS_VSYNC;
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_NORMAL:		/* just blank screen (backlight stays on) */
 		seqflags = VGA_SR01_SCREEN_OFF;	/* Disable sequencer */
 		lcdflags = par->PanelDispCntlReg1 & 0x02; /* LCD normal */
 		dpmsflags = 0x00;	/* no hsync/vsync suppression */
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_UNBLANK:		/* unblank */
 		seqflags = 0;			/* Enable sequencer */
@@ -1387,6 +1394,7 @@
 			tosh_smm(&regs);
 		}
 #endif
+		par->PanelDispCntlRegRead = 1;
 		break;
 	default:	/* Anything else we don't understand; return 1 to tell
 			 * fb_blank we didn't aactually do anything */
#2 lore raw

Re: [PATCH] neofb: add more logic to determine sensibility of register readback

Quoted 76 lines · Christian Trefzer

Hello everyone,

testing my latest patches through productive use I noticed that we require a bit of a logic for the register read to not cause trouble at the time of unblanking. Thing is, blanking is implemented through deactivation of the LCD activation bit, so if we read back the register, the LCD backlight won't come back on until the LID is shut and reopened, effectively pushing the work to the BIOS.

The following patch is on top of my previous one, adding a variable to struct neofb_par to remember if we want to read the respective register values next time we (un)blank.

Logic depends on the assumption that we won't change display config in a blanked state, as a keypress should already be answered by unblanking. Essentially, we should read back the register values to variable on blank, and ignore them on unblank. Unfortunately, only checking for !blank_mode won't avoid the backlight remaining off, hence the additional checks.

include/video/neomagic.h +1 -0
@@ -159,6 +159,7 @@
 	unsigned char PanelDispCntlReg1;
 	unsigned char PanelDispCntlReg2;
 	unsigned char PanelDispCntlReg3;
+	unsigned char PanelDispCntlRegRead;
 	unsigned char PanelVertCenterReg1;
 	unsigned char PanelVertCenterReg2;
 	unsigned char PanelVertCenterReg3;
drivers/video/neofb.c +9 -2
@@ -1334,8 +1334,11 @@
 	struct neofb_par *par = (struct neofb_par *)info->par;
 	int seqflags, lcdflags, dpmsflags, reg;
 
-	/* Reload the value stored in the register, might have been changed via FN keystroke */
-	par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	/* Reload the value stored in the register, if sensible. It might have been
+	 * changed via FN keystroke. */
+	if (par->PanelDispCntlRegRead && !blank_mode) {
+		par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	}
 
 	switch (blank_mode) {
 	case FB_BLANK_POWERDOWN:	/* powerdown - both sync lines down */
@@ -1355,21 +1358,25 @@
 			tosh_smm(&regs);
 		}
 #endif
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_HSYNC_SUSPEND:		/* hsync off */
 		seqflags = VGA_SR01_SCREEN_OFF;	/* Disable sequencer */
 		lcdflags = 0;			/* LCD off */
 		dpmsflags = NEO_GR01_SUPPRESS_HSYNC;
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_VSYNC_SUSPEND:		/* vsync off */
 		seqflags = VGA_SR01_SCREEN_OFF;	/* Disable sequencer */
 		lcdflags = 0;			/* LCD off */
 		dpmsflags = NEO_GR01_SUPPRESS_VSYNC;
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_NORMAL:		/* just blank screen (backlight stays on) */
 		seqflags = VGA_SR01_SCREEN_OFF;	/* Disable sequencer */
 		lcdflags = par->PanelDispCntlReg1 & 0x02; /* LCD normal */
 		dpmsflags = 0x00;	/* no hsync/vsync suppression */
+		par->PanelDispCntlRegRead = 0;
 		break;
 	case FB_BLANK_UNBLANK:		/* unblank */
 		seqflags = 0;			/* Enable sequencer */

drivers/video/neofb.c line 1400

Quoted 7 lines
drivers/video/neofb.c +1 -0
@@ -1387,6 +1394,7 @@
 			tosh_smm(&regs);
 		}
 #endif
+		par->PanelDispCntlRegRead = 1;
 		break;
 	default:	/* Anything else we don't understand; return 1 to tell
 			 * fb_blank we didn't aactually do anything */

You can save a few lines by

par->PanelDispCntlRegRead = (blank_mode) ? 0 : 1;

Tony

#3 lore raw

Re: [PATCH] neofb: add more logic to determine sensibility of register readback

Hi Tony,

Quoted 6 lines · Antonino A. Daplas

You can save a few lines by

par->PanelDispCntlRegRead = (blank_mode) ? 0 : 1;

Tony

That's just beautiful. I'm a bloody n00b when it comes to C, so I lack the feeling for such subtleties : ) Thanks a bunch! Reworked patch below, also adding proper initialization which got missing during back-and-forth testing.

include/video/neomagic.h +1 -0
@@ -159,6 +159,7 @@
 	unsigned char PanelDispCntlReg1;
 	unsigned char PanelDispCntlReg2;
 	unsigned char PanelDispCntlReg3;
+	unsigned char PanelDispCntlRegRead;
 	unsigned char PanelVertCenterReg1;
 	unsigned char PanelVertCenterReg2;
 	unsigned char PanelVertCenterReg3;
drivers/video/neofb.c +9 -2
@@ -843,6 +843,9 @@
 
 	par->SysIfaceCntl2 = 0xc0;	/* VESA Bios sets this to 0x80! */
 
+	/* Initialize: by default, we want display config register to be read */
+	par->PanelDispCntlRegRead = 1;
+
 	/* Enable any user specified display devices. */
 	par->PanelDispCntlReg1 = 0x00;
 	if (par->internal_display)
@@ -1334,8 +1337,12 @@
 	struct neofb_par *par = (struct neofb_par *)info->par;
 	int seqflags, lcdflags, dpmsflags, reg;
 
-	/* Reload the value stored in the register, might have been changed via FN keystroke */
-	par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	/* Reload the value stored in the register, if sensible. It might have been
+	 * changed via FN keystroke. */
+	if (par->PanelDispCntlRegRead && !blank_mode) {
+		par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	}
+	par->PanelDispCntlRegRead = (blank_mode) ? 0 : 1;
 	
 	switch (blank_mode) {
 	case FB_BLANK_POWERDOWN:	/* powerdown - both sync lines down */
#4 lore raw

Re: [PATCH] neofb: add more logic to determine sensibility of register readback

In reply to Antonino A. Daplas
Quoted 10 lines · Antonino A. Daplas

drivers/video/neofb.c

Quoted 5 lines · Christian Trefzer
drivers/video/neofb.c +5 -0
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 1;

You can save a few lines by

par->PanelDispCntlRegRead = (blank_mode) ? 0 : 1;

How about the really simple so-obvious-its-impossible-to-misread solution?

par->PanelDispCntlRegRead = !blank_mode;

Personally I tend to get ?: constructs confused a _lot_, especially mistaking the really short ones like x?0:1 and x?1:0. Those two are usually better represented as !x or !!x, respectively.

Cheers,
Kyle Moffett

--
I lost interest in "blade servers" when I found they didn't throw knives at people who weren't supposed to be in your machine room.
-- Anthony de Boer

#5 lore raw

Re: [PATCH] neofb: add more logic to determine sensibility of register readback

Quoted 10 lines · Kyle Moffett
Quoted 9 lines · Antonino A. Daplas

Review

Quoted 5 lines · Christian Trefzer
(quoted diff) +5 -0
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 0;
+		par->PanelDispCntlRegRead = 1;

You can save a few lines by

Quoted 6 lines
Quoted 1 line

par->PanelDispCntlRegRead = (blank_mode) ? 0 : 1;

How about the really simple so-obvious-its-impossible-to-misread solution?

par->PanelDispCntlRegRead = !blank_mode;

Where is my mind..?

Sure, why not. Guess I had my head too much in the smoke clouds of not knowing why a dead-sure-to-work implementation still would not do the right thing, I already thought I got the C representation of true and false wrong. Even Tony's hint would be clearer than my code, but that can in turn be condensed into something really obvious. It's just a negation, after all. Silly me.

Changed patch below. Should be the final one now, right?

include/video/neomagic.h +1 -0
@@ -159,6 +159,7 @@
 	unsigned char PanelDispCntlReg1;
 	unsigned char PanelDispCntlReg2;
 	unsigned char PanelDispCntlReg3;
+	unsigned char PanelDispCntlRegRead;
 	unsigned char PanelVertCenterReg1;
 	unsigned char PanelVertCenterReg2;
 	unsigned char PanelVertCenterReg3;
drivers/video/neofb.c +9 -2
@@ -843,6 +843,9 @@
 
 	par->SysIfaceCntl2 = 0xc0;	/* VESA Bios sets this to 0x80! */
 
+	/* Initialize: by default, we want display config register to be read */
+	par->PanelDispCntlRegRead = 1;
+
 	/* Enable any user specified display devices. */
 	par->PanelDispCntlReg1 = 0x00;
 	if (par->internal_display)
@@ -1334,8 +1337,12 @@
 	struct neofb_par *par = (struct neofb_par *)info->par;
 	int seqflags, lcdflags, dpmsflags, reg;
 
-	/* Reload the value stored in the register, might have been changed via FN keystroke */
-	par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	/* Reload the value stored in the register, if sensible. It might have been
+	 * changed via FN keystroke. */
+	if (par->PanelDispCntlRegRead && !blank_mode) {
+		par->PanelDispCntlReg1 = vga_rgfx(NULL, 0x20) & 0x03;
+	}
+	par->PanelDispCntlRegRead = !blank_mode;
 	
 	switch (blank_mode) {
 	case FB_BLANK_POWERDOWN:	/* powerdown - both sync lines down */