Bug 43333

I'd also like to add that the VGA2 is connected through Chrontel 7317A-TF chip.

Can you please provide the output of "xrandr --verbose" with two monitors attached? (when the monitor is turning on/off)

Does the 1600x1200 and 1280x1024 modes work on that monitor? And if they *do* work, I'll ask you to test the 1440x900, 1360x768 and 1280x800 modes: if they *don't* work, then I'd say this bug might be a duplicate of 42174.

I was able to fix both issues. The problem with EDID of VGA2 being a duplicate of VGA1 turned out to be a bad ddc guess in intel_sdvo_guess_ddc_bus(). sdvo->controlled_output has the value of 14 in the case of and it maps to sdvo->ddc_bus 1, which retrieved EDID from VGA1. Changing this to 2 fixed the problem with monitor detection.

Modes which didn't work had +vsync -hsync set. I was able to fix this patch applied to intel_sdvo_mode_set():

		if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) {
			sdvox |= SDVO_VSYNC_ACTIVE_HIGH;
		+	sdvox |= SDVO_HSYNC_ACTIVE_HIGH;
		}
		if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) {
			sdvox |= SDVO_HSYNC_ACTIVE_HIGH;
		+	sdvox |= SDVO_VSYNC_ACTIVE_HIGH;
		}

I adopted this solution from an old patch for Xorg. I don't understand why it is needed to set both flags, but it simply works. Are there any refs for this stuff?

(In reply to comment #3)
> I was able to fix both issues. The problem with EDID of VGA2 being a duplicate
> of VGA1 turned out to be a bad ddc guess in intel_sdvo_guess_ddc_bus().
> sdvo->controlled_output has the value of 14 in the case of and it maps to
> sdvo->ddc_bus 1, which retrieved EDID from VGA1. Changing this to 2 fixed the
> problem with monitor detection.
> 

Is this documented anywhere? Is there a bug report or patch on a list? If you found a fix for a problem, let's try to push it upstream :)

> Modes which didn't work had +vsync -hsync set. I was able to fix this patch
> applied to intel_sdvo_mode_set():
> 
>         if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) {
>             sdvox |= SDVO_VSYNC_ACTIVE_HIGH;
>         +    sdvox |= SDVO_HSYNC_ACTIVE_HIGH;
>         }
>         if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) {
>             sdvox |= SDVO_HSYNC_ACTIVE_HIGH;
>         +    sdvox |= SDVO_VSYNC_ACTIVE_HIGH;
>         }
> 
> I adopted this solution from an old patch for Xorg.

Where is this patch? Can you please provide a link? This might really help us.

> I don't understand why it
> is needed to set both flags, but it simply works. Are there any refs for this
> stuff?

With your patch, modes that have -VSync +HSync will become +VSync +HSync and modes that have +VSync -HSync will become +VSync +HSync. And this *really* makes me think this bug is a duplicate of #42174. I have a theory on the reason but I'll need you to help me:

With your current patch, I believe that modes with -VSync -HSync still won't work for you. Can you please confirm that? Use "dmesg | grep Modeline | grep 0xa$" or "xrandr --verbose" to find out which modes have -VSync and -HSync and then test them and tell me whether they work or not with (or without) your patch.

Thank you,
Paulo

(In reply to comment #4)
> (In reply to comment #3)
> > I was able to fix both issues. The problem with EDID of VGA2 being a duplicate
> > of VGA1 turned out to be a bad ddc guess in intel_sdvo_guess_ddc_bus().
> > sdvo->controlled_output has the value of 14 in the case of and it maps to
> > sdvo->ddc_bus 1, which retrieved EDID from VGA1. Changing this to 2 fixed the
> > problem with monitor detection.
> > 
> 
> Is this documented anywhere? Is there a bug report or patch on a list? If you
> found a fix for a problem, let's try to push it upstream :)

See the series of patches Adam Jackson sent to fix this: 1308256588-8026-1-git-send-email-ajax@redhat.com

> > Modes which didn't work had +vsync -hsync set. I was able to fix this patch
> > applied to intel_sdvo_mode_set():
> > 
> >         if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) {
> >             sdvox |= SDVO_VSYNC_ACTIVE_HIGH;
> >         +    sdvox |= SDVO_HSYNC_ACTIVE_HIGH;
> >         }
> >         if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) {
> >             sdvox |= SDVO_HSYNC_ACTIVE_HIGH;
> >         +    sdvox |= SDVO_VSYNC_ACTIVE_HIGH;
> >         }
> > 
> > I adopted this solution from an old patch for Xorg.
> 
> Where is this patch? Can you please provide a link? This might really help us.

That's not a patch, but a hack around a buggy EDID, In this case hacking around reading the wrong one.

Created attachment 54212 [details]
xrandr -q --verbose

The patch for the orignal Xorg driver can be found here: https://bugs.freedesktop.org/show_bug.cgi?id=15766

Paulo Zanoni: you were right that modes with -vsync -hsync still didn't work. I fixed it by unconditionally SDVO_VSYNC_ACTIVE_HIGH and SDVO_HSYNC_ACTIVE_HIGH.

*** This bug has been marked as a duplicate of bug 42174 ***

A patch referencing this bug report has been merged in Linux v3.3-rc2:

commit ba68e086223a5f149f37bf8692c8cdbf1b0ba3ef
Author: Paulo Zanoni <paulo.r.zanoni@intel.com>
Date:   Fri Jan 6 19:45:34 2012 -0200

    drm/i915/sdvo: always set positive sync polarity