Patch to fix fdo bug #21064

Commit Message

Zhenyu Wang Aug. 7, 2009, 10:08 a.m. UTC

On 2009.08.06 17:12:39 +0200, René Gabriëls wrote:
> The attached patch fixes my garbled screen issue on my Intel 855GM notebook. For
> some reason the VGA_2X_MODE bit in the VGACNTRL register is set to 1 on my
> machine, causing the video chip to scale up the framebuffer, which results in a
> funky screen.
> 
> In the patch I forced this bit to zero and the problem is gone.  I'm not sure if
> there's anybody who's actually using this 2X_MODE on an 855GM.  However, before
> this bug was introduced (git: db9f5915ce812144ffd9d2aa42e8ba856129c35e), this
> bit was forced to zero as well.
> 
> - René

> diff --git a/src/i830_display.c b/src/i830_display.c
> index 59ededc..d180b2b 100644
> --- a/src/i830_display.c
> +++ b/src/i830_display.c
> @@ -1096,6 +1096,10 @@ i830_disable_vga_plane (xf86CrtcPtr crtc)
>  
>      vgacntrl |= VGA_DISP_DISABLE;
>  
> +    /* disable "VGA/Pop-up 2X Centered Mode Scaling" on 855GM platform */
> +    if (IS_I855(pI830))
> +        vgacntrl &= ~VGA_2X_MODE;
> +    
>      OUTREG(VGACNTRL, vgacntrl);
>      i830WaitForVblank(pScrn);
>  

Looks we brought more problem with VGA mode disable than it has fixed...
and sr bit setting should only apply to affected chip, which is a G43 as we
originally saw this.

So how about following patch? Does it work for you? I would try to test this
on origin problematic G43 when I had hands no it, and current kms patch for this
on drm-intel-next needs rework too, and I'll test if sr bit affects on new chip
or not.

Comments

René Gabriëls Aug. 7, 2009, 1:22 p.m. UTC | #1

> So how about following patch? Does it work for you? I would try to test this
> on origin problematic G43 when I had hands no it, and current kms patch for this
> on drm-intel-next needs rework too, and I'll test if sr bit affects on new chip
> or not.

This patch works, because it reverts to the old behavior on my machine (i.e.
setting VGACNTRLS to all zeroes except the MSB).  This is fine with me, as long
as it works and doesn;t break anyone else's driver!

- René

Patch

diff --git a/src/i830_display.c b/src/i830_display.c
index 59ededc..b749252 100644
--- a/src/i830_display.c
+++ b/src/i830_display.c
@@ -1076,27 +1076,20 @@  i830_disable_vga_plane (xf86CrtcPtr crtc)
 {
     ScrnInfoPtr pScrn = crtc->scrn;
     I830Ptr pI830 = I830PTR(pScrn);
-    uint32_t vgacntrl = INREG(VGACNTRL);
     uint8_t sr01;
 
-    if (vgacntrl & VGA_DISP_DISABLE)
-	return;
-
     /*
        Set bit 5 of SR01;
        Wait 30us;
        */
-    OUTREG8(SRX, 1);
-    sr01 = INREG8(SRX + 1);
-    OUTREG8(SRX + 1, sr01 | (1 << 5));
-    usleep(30);
-    /* disable center mode on 965GM and G4X platform */
-    if (IS_I965GM(pI830) || IS_G4X(pI830))
-        vgacntrl &= ~(3 << 24);
-
-    vgacntrl |= VGA_DISP_DISABLE;
+    if (IS_G4X(pI830)) {
+	OUTREG8(SRX, 1);
+	sr01 = INREG8(SRX + 1);
+	OUTREG8(SRX + 1, sr01 | (1 << 5));
+	usleep(30);
+    }
 
-    OUTREG(VGACNTRL, vgacntrl);
+    OUTREG(VGACNTRL, VGA_DISP_DISABLE);
     i830WaitForVblank(pScrn);
 
 }