[i945] Macbook's Tv out doesn't work (i945)

System : Ubuntu 7.04 (2.6.20-16-generic)

It seems that the Tv Out doesn't work with intel-945 in the MacBook (mine is second generation, but it should be the same for others gen).

We have to use an adapter mini-DVI->S-Video but the only things I have on my tv is "crapy"... i can just see something move when I move a window for instance. It looks like the refresh rate is completely bad...
I had to put Option "MonitorLayout" "CRT,LFP" in xorg.conf in order to see something on the tv (with "TV" replacing CRT, i got a black tv).
I tried with the tv plugged at boot time, nothing change... and I tried several way of configure Xorg too...

If it can helps you, apple say that on its web site :
---------------------------------------------------------------
The cable detect function on pin 25 is implemented by connecting pin 25 to +5V in the adapters. The computer detects which adapter is present by reading its EDID (Extended Display Identification Data) via DDC. The EDID for video is in the adapter; the EDID for VGA and DVI is in the display.
---------------------------------------------------------------

---------------
--Other links--
---------------
Here is the bug report on ubuntu launchpad : https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/116207

Mail on freedesktop list : http://lists.freedesktop.org/archives/xorg/2006-October/018739.html

Post on ubuntu forum : http://ubuntuforums.org/showthread.php?t=449506

Gentoo wiki : http://gentoo-wiki.com/Talk:HARDWARE_Apple_MacBook#DVI_and_TV_Out

I can confirm this bug.
I tried xrandr to setup the Tv output in Ubuntu 7.10 on a first generation macbook but it keeps saying it is disconnected
I only get some noisy output when I boot with the cable connected, until the xserver starts.

maarten or J Lange, would you please post your x.org log file and xorg.conf to this bug. Sorry for no response for a long time...

Created attachment 13534
xorg.conf which display "something" on the tv

Created attachment 13535
Xorg.0.log (with the s-video adaptor plugged at boot time)

This still seems to be an issue in Ubuntu Hardy Heron 8.04 with Intel driver 2:2.2.1-1ubuntu5.

https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/116207

You should try with "intel" driver instead of "i810". And if that's not working, pls attach xorg log with ModeDebug option turned on.

Created attachment 15225
xorg log

It still doesn't work... I attached the corresponding log and the output of xrandr.

Created attachment 15226
xrandr output

Created attachment 16086
xrandr --prop with adapter attached

I'm experiencing this bug as well, attached there is a xrandr --prop output.

Apparently the adapter provides some special EDID data to announce itself as svideo/composite adapter, then the mini-dvi output analog pins can be sent the actual analog signals (my best guess).

some "more" information is provided here:

http://developer.apple.com/documentation/Hardware/Developer_Notes/Macintosh_CPUs-G4/12inchPowerBookG4/3_Input-Output/chapter_4_section_15.html#//apple_ref/doc/uid/TP40001763-CH207-TPXREF119|12-inch

many thanks for your excellent work!

As a side note, https://bugs.freedesktop.org/show_bug.cgi?id=17342 indicates DVI->TV-out basically works on MacMini.

Have you tried with newer ubuntu release for this bug? I think mini-DVI to VGA and DVI should work fine, we don't know if S-video adapter can work or not. Please attach new X log with ModeDebug option on.

bug# 17342 has prove TV works in 2.4.0, but not work in 2.5.0. A good enough hint for zhenyu to find out why...

It looks mini-DVI adapter with S-Video TV does work as in bug 17342, although that one has EDID with TV info through SDVO port, but this one is through VGA port. Because this one is MacBook (laptop with LVDS), but 17342 is Mac Mini (desktop). They have different display configuration.

I'll attach a patch which try to use VGA EDID in Macbook case.

Created attachment 22454
Use VGA edid info for TV detect on MacBook

Please help to test this patch, it's against current git master of xf86-video-intel. Thanks.

(In reply to comment #14)
> It looks mini-DVI adapter with S-Video TV does work as in bug 17342, although
> that one has EDID with TV info through SDVO port, but this one is through VGA
> port.

have you ruled out the hack code Eric added to i830_sdvo_get_ddc_modes? it steal DDC data from VGA for Mac Mini...which looks like SDVO is capable to get DDC data itself.

>Because this one is MacBook (laptop with LVDS), but 17342 is Mac Mini
> (desktop). They have different display configuration.
>

HW form factor looks different. Google shows that MacMini use a DVI-I connector, but I doubt the internal mechanism is still same...

(In reply to comment #15)
> Created an attachment (id=22454) [details]
> Use VGA edid info for TV detect on MacBook
>
> Please help to test this patch, it's against current git master of
> xf86-video-intel. Thanks.
>

I've tested this patch here and it did not solve the issue for me.

(In reply to comment #17)
> (In reply to comment #15)
> > Created an attachment (id=22454) [details] [details]
> > Use VGA edid info for TV detect on MacBook
> >
> > Please help to test this patch, it's against current git master of
> > xf86-video-intel. Thanks.
> >
>
> I've tested this patch here and it did not solve the issue for me.
>

Luis, would you please kindly post your log with ModeDebug turns on in xorg.conf? thanks.

Created attachment 26885
please try the debug patch on your machine, thanks.

the xrandr output in comments #8 indicates VGA, LVDS connected, TV disconnected. We have modified VGA and TV detections. So could you please try the debug patch against latest 2D driver , then upload log file with modedebug option on.

Thanks for your help
Ma Ling

Created attachment 26902
xrandr output with patch from 2009-06-17

Created attachment 26903
Xorg.0.log with the patch from 20090617

This is the Xorg.0.log with the latest patch attached. I also didn't get any output on the TV. If there are any more information or tests you guys need, please just tell me. I'm really interested in having this fixed. Thanks alot for the work so far.

(In reply to comment #21)
> Created an attachment (id=26903) [details]
> Xorg.0.log with the patch from 20090617
>
> This is the Xorg.0.log with the latest patch attached. I also didn't get any
> output on the TV. If there are any more information or tests you guys need,
> please just tell me. I'm really interested in having this fixed. Thanks alot
> for the work so far.
>

would you please ignore the CRT and TMDS-1 using xorg.conf, and have a try again ( with the patch )?

to ignore, do similar thing like this:

Section "Device"
 ...
        Option "monitor-VGA" "VGA"
        ...
EndSection
        ...
Section "Monitor"
        Identifier "VGA"
        Option "Ignore" "True"
EndSection

Created attachment 26904
xrandr output with patch from 2009-06-17 and ignoring CRT and TMDS-1

Created attachment 26905
Xorg.0.log with patch from 2009-06-17 and ignoring CRT and TMDS-1

Still no output whatsoever on the TV.

Created attachment 26908
clear dac sense mask

how about with this patch? Note: might need you do some hand-editing to resolve conflict... thanks.

(In reply to comment #25)
> Created an attachment (id=26908) [details]
> clear dac sense mask
> how about with this patch? Note: might need you do some hand-editing to resolve
> conflict... thanks.

This patch has been merged into our master tree, so the patch in comments #19 has include the same behavior.

Created attachment 26913
please try the debug patch, thanks

With comments #22 to ignore CRT and TMDS, this patch will directly return our expected result. I think if your hardware interface closely match our tv spec,
we should clearly see the tv image although I know you ever saw vague tv image under i810 driver.

Thanks
Ma Ling

(In reply to comment #26)
> (In reply to comment #25)
> > Created an attachment (id=26908) [details] [details]
> > clear dac sense mask
> > how about with this patch? Note: might need you do some hand-editing to resolve
> > conflict... thanks.
>
> This patch has been merged into our master tree, so the patch in comments #19
> has include the same behavior.
>

right.. but I noticed that Luis is using 2.7.1 in his log...not sure if he has the mask definition in i830_reg.h which wasn't included in your patch in comment# 19.. Hmm..

(In reply to comment #28)
> (In reply to comment #26)
> > (In reply to comment #25)
> > > Created an attachment (id=26908) [details] [details] [details]
> > > clear dac sense mask
> > > how about with this patch? Note: might need you do some hand-editing to resolve
> > > conflict... thanks.
> >
> > This patch has been merged into our master tree, so the patch in comments #19
> > has include the same behavior.
> >
>
> right.. but I noticed that Luis is using 2.7.1 in his log...not sure if he has
> the mask definition in i830_reg.h which wasn't included in your patch in
> comment# 19.. Hmm..
>

I added the definition of the masks to i830_reg.h by hand, got them from git. I'm using the X packages that are provided here https://edge.launchpad.net/~ubuntu-x-swat on the stable updates. I'm gonna see about changing to the more bleeding edge stuff here, but I can't do any more testing today, I'll probably have to do them on the weekend. Will update here as soon as I have more info.

Could you kindly please try the patch in comments #27?

Thanks
Ma Ling

(In reply to comment #30)
> Could you kindly please try the patch in comments #27?
>
> Thanks
> Ma Ling
>

I've tried this patch and it also did not show any output whatsoever. I've tested it with both the TV_TYPE_SVIDEO and TV_TYPE_COMPOSITE return values, since the mini-dvi to tv-out adaptor has both connections and the cable I'm using is an S-Video (connected on the TV) to Composite (connected on the Macbook) cable, and I wasn't sure exactly which type should actually be returned in this setup.

(In reply to comment #31)
> (In reply to comment #30)
> > Could you kindly please try the patch in comments #27?
> >
> > Thanks
> > Ma Ling
> >
>
> I've tried this patch and it also did not show any output whatsoever. I've
> tested it with both the TV_TYPE_SVIDEO and TV_TYPE_COMPOSITE return values,
> since the mini-dvi to tv-out adaptor has both connections and the cable I'm
> using is an S-Video (connected on the TV) to Composite (connected on the
> Macbook) cable, and I wasn't sure exactly which type should actually be
> returned in this setup.
>

Oh, no, please don't use it in that way. The Svideo-Composite cable is only for you to connect Svideo on the computer side and Composite on the TV side.

Do you have a SVideo-to-Svideo cable or Composite-to-Composite cable to try?

(In reply to comment #32)
> (In reply to comment #31)
> > (In reply to comment #30)
> > > Could you kindly please try the patch in comments #27?
> > >
> > > Thanks
> > > Ma Ling
> > >
> >
> > I've tried this patch and it also did not show any output whatsoever. I've
> > tested it with both the TV_TYPE_SVIDEO and TV_TYPE_COMPOSITE return values,
> > since the mini-dvi to tv-out adaptor has both connections and the cable I'm
> > using is an S-Video (connected on the TV) to Composite (connected on the
> > Macbook) cable, and I wasn't sure exactly which type should actually be
> > returned in this setup.
> >
>
> Oh, no, please don't use it in that way. The Svideo-Composite cable is only for
> you to connect Svideo on the computer side and Composite on the TV side.
>
> Do you have a SVideo-to-Svideo cable or Composite-to-Composite cable to try?
>

Just tried it again with a regular SVideo-to-SVideo cable and got the same results (which is, no image displayed on the tv).

(In reply to comment #33)
> (In reply to comment #32)
> > (In reply to comment #31)
> > > (In reply to comment #30)
> > > > Could you kindly please try the patch in comments #27?
> > > >
> > > > Thanks
> > > > Ma Ling
> > > >
> > >
> > > I've tried this patch and it also did not show any output whatsoever. I've
> > > tested it with both the TV_TYPE_SVIDEO and TV_TYPE_COMPOSITE return values,
> > > since the mini-dvi to tv-out adaptor has both connections and the cable I'm
> > > using is an S-Video (connected on the TV) to Composite (connected on the
> > > Macbook) cable, and I wasn't sure exactly which type should actually be
> > > returned in this setup.
> > >
> >
> > Oh, no, please don't use it in that way. The Svideo-Composite cable is only for
> > you to connect Svideo on the computer side and Composite on the TV side.
> >
> > Do you have a SVideo-to-Svideo cable or Composite-to-Composite cable to try?
> >
>
> Just tried it again with a regular SVideo-to-SVideo cable and got the same
> results (which is, no image displayed on the tv).
>

so I assume you've tried the HW on Mac OS to prove it can work, right?

Did you test above with the debug patch in comment# 27?

Also as we discussed in comment# 11 to comment# 16, would you please try to contact <email address hidden> in bug# 17342? At least he can get a working TV after start. With the hacky patch in comment# 27, it should at least give some signal on your TV...

thanks.

(In reply to comment #34)
>
> so I assume you've tried the HW on Mac OS to prove it can work, right?
>
> Did you test above with the debug patch in comment# 27?
>
> Also as we discussed in comment# 11 to comment# 16, would you please try to
> contact <email address hidden> in bug# 17342? At least he can get a working TV after
> start. With the hacky patch in comment# 27, it should at least give some signal
> on your TV...
>
> thanks.
>

I don't have MacOSX installed anymore to try, but when I did I used it normally, including with that SVideo-Composite cable I mentioned.
I tried again the patch in comment #27, and the only feedback I've on the TV is when I leave X (for example, when rebooting or something like that). But inside X, I get no signal. I'll try contacting that person and will post here if I get any new results.
It's worth mentioning that I'm getting an error message on xrandr all the time about VGA being connected (which it isn't) but having no modes available.

(In reply to comment #35)
> (In reply to comment #34)
> >
> It's worth mentioning that I'm getting an error message on xrandr all the time
> about VGA being connected (which it isn't) but having no modes available.
>

I see. would you please continue to ignore the bogus CRT and TMDS as you did in comment# 23 ( sorry, didn't clarify this to you before ) and post your xorg.log with modedebug turns on? thanks.

Created attachment 27020
Xorg.0.log from 22/06/2009 with the return SVideo hack

Created attachment 27021
Xorg.0.log from 22/06/2009 with the "Use VGA edid info for TV detect" patch

Also attaching the Xorg.0.log from this one, decided to give it another try just to make sure that the SVideo-Composite cable wasn't getting in the way.

(In reply to comment #38)
> Created an attachment (id=27021) [details]
> Xorg.0.log from 22/06/2009 with the "Use VGA edid info for TV detect" patch
>
> Also attaching the Xorg.0.log from this one, decided to give it another try
> just to make sure that the SVideo-Composite cable wasn't getting in the way.
>

so you said you can see the signal on TV until X start, right? but all the regs that controls TV are empty...Hmm..

this is the log from bug# 17342 comment# 32. pay attention to the reg dump before X start

(In reply to comment #39)
> (In reply to comment #38)
> > Created an attachment (id=27021) [details] [details]
> > Xorg.0.log from 22/06/2009 with the "Use VGA edid info for TV detect" patch
> >
> > Also attaching the Xorg.0.log from this one, decided to give it another try
> > just to make sure that the SVideo-Composite cable wasn't getting in the way.
> >
>
> so you said you can see the signal on TV until X start, right? but all the regs
> that controls TV are empty...Hmm..
>
> this is the log from bug# 17342 comment# 32. pay attention to the reg dump
> before X start
>

Sorry, just to be more clear, outside of X I get some random noise on the TV, like white dots and lines, not proper signal.

Could you please tell me which log file is from the scenario:
we use the patch in comments #27, SVideo-to-SVideo connected between Macbook and TV, xorg.conf to ignor VGA and TMDS1?

Thanks
Ma Ling

(In reply to comment #41)
> Could you please tell me which log file is from the scenario:
> we use the patch in comments #27, SVideo-to-SVideo connected between Macbook
> and TV, xorg.conf to ignor VGA and TMDS1?
>
> Thanks
> Ma Ling
>

The one called "Xorg.0.log from 22/06/2009 with the return SVideo hack".

(In reply to comment #40)
> (In reply to comment #39)
> > (In reply to comment #38)
>
> Sorry, just to be more clear, outside of X I get some random noise on the TV,
> like white dots and lines, not proper signal.
>

looks like we might have some problem on the Macbook. It might be really different as Mac Mini, though I think it shouldn't, in terms of TV connection..

Windows can't work as well. See http://www.mac-forums.com/forums/windows-classic-linux-other-os/138205-macbook-got-s-video-out-working-under-xp-bootcamp.html

From the xorg log in comment #3 it seems that it is misregarded as VGA when the mini-DVI to S-video is used.(VGA is detected as connected and EDID can be obtained). In such case the TV is not detected correctly.
And it is very interesting that the TMDS-1 is also detected as connected.
In fact there exists only one mini-DVI connector. And from http://en.wikipedia.org/wiki/Mini-DVI we can know that mini-DVI can carry the DVI-D, VGA, TV signal.
Maybe on mini mac there exists so much mystery about the hardware design that our driver can't handle this.
thanks.

(In reply to comment #44)
> From the xorg log in comment #3 it seems that it is misregarded as VGA when the
> mini-DVI to S-video is used.(VGA is detected as connected and EDID can be
> obtained). In such case the TV is not detected correctly.
> And it is very interesting that the TMDS-1 is also detected as connected.
> In fact there exists only one mini-DVI connector. And from
> http://en.wikipedia.org/wiki/Mini-DVI we can know that mini-DVI can carry the
> DVI-D, VGA, TV signal.

Not sure if it means anything, but the adaptor in question is mini-DVI to both S-VIDEO and RCA.
http://www.screencapturenews.com/wp-content/uploads/mini-dvi-to-tv-video.gif

Could this account for both outputs being turned on? (I really don't know much about this, sorry if the question is stupid)

> Maybe on mini mac there exists so much mystery about the hardware design that
> our driver can't handle this.

Just making sure, you're talking about the macbook here right? I think that the mini mac already works ok, but I don't have one to test here.

--lf

> thanks.
>

Hi,
as another data point this is what I'm seeing with the adapter connected:

$ xrandr --prop
xrandr: Output VGA1 is not disconnected but has no modes
Screen 0: minimum 320 x 200, current 1280 x 800, maximum 8192 x 8192
VGA1 connected (normal left inverted right x axis y axis)
 EDID:
  00ffffffffffff000610089d01010101
  070c0103100000963067a9a056479926
  11484c00006101010101010101010101
  01010101010101010101010101010101
  01010101010101010101010101010101
  01010101010101010101010101010101
  010101010101010101010101000000fc
  004e5453432f50414c0a202020200038
LVDS1 connected 1280x800+0+0 (normal left inverted right x axis y axis) 286mm x 179mm
 EDID:
  00ffffffffffff0006105f9c00000000
  08100103801d12780a2f309758538b29
  25505400000001010101010101010101
  010101010101bc1b00a0502017303020
  36001eb3100000180000000100061020
  00000000000000000a20000000fe004c
  503133335758312d544c4131000000fe
  00436f6c6f72204c43440a20202000c2
 scaling mode: Fullscreen
  supported: Non-GPU Fullscreen No scale Aspect
   1280x800 59.9*+
   1024x768 85.0 75.0 70.1 60.0
   832x624 74.6
   800x600 85.1 72.2 75.0 60.3 56.2
   640x480 85.0 72.8 75.0 59.9
   720x400 85.0
   640x400 85.1
   640x350 85.1
DVI1 disconnected (normal left inverted right x axis y axis)
TV1 disconnected (normal left inverted right x axis y axis)
 bottom margin: 37 (0x00000025) range: (0,100)
 right margin: 46 (0x0000002e) range: (0,100)
 top margin: 36 (0x00000024) range: (0,100)
 left margin: 54 (0x00000036) range: (0,100)
 mode: NTSC-M
  supported: NTSC-M NTSC-443 NTSC-J PAL-M
             PAL-N PAL 480p@59.94Hz 480p@60Hz
             576p 720p@60Hz 720p@59.94Hz 720p@50Hz
             1080i@50Hz 1080i@60Hz 1080i@59.94H

I'm currently running on debian with kernel 2.6.31-1-amd64 with
xserver-xorg 1:7.5+1 and xserver-xorg-video-intel 2:2.9.1-2 and using KMS

I guess the latest patch won't apply since it is all moved to the kernel driver?

thanks in advance,
filippo

Looks like a failure to retrieve the EDID over the dongle, we might need more permissive timings (as in #28843).

Is this issue still valid with latest version of Intel graphics stack or was it eventually fixed meanwhile? There were lots of updates and bug #28843 was closed a long time ago.

Testing now with Gentoo LiveDVD 11.2 which has 3.0.0 kernel, xorg-server 1.10.3 and xf86-video-intel-2.15.0, still only shows VGA1 as active, TV1 as disconnected. Only shows snow on the TV.

It's on a MacBook with 945GM/GMS and a MiniDVI to RCA and S-Video out, testing out the S-Video.

3.3-rc1 has a bunch of tv-out timing fixes. Can you please retest?

Timeout. Please do reopen if you can still reproduce the issue and help us diagnose the problem, thanks.