Skip to contentSkip to conversation

Intel 965 and Custom Resolutions

DISCUSSION Lucas 5 participants 14 messages Aug 30, 2010 – Aug 31, 2010

lore
Share & tools
lore

Lucas

author
#0 lore raw

Hi All,

Pardon me if this has been discussed before but I sure as heck can't find the answer. Perhaps my googling skills aren't good enough.

Anyway, I have an AOpen DE 965-HG mini PC that has - surprise! - Intel 965 graphics. It has an HDMI-out port which I have connected to my 42" plasma Panasonic TV. The TV is a little old (circa 2005) so it apparently reports the wrong EDID information. It also needs to be adjusted for overscan. When I try to run Linux (any kind, but currently Ubuntu 10.04) X server it picks up the wrong resolution. With a different computer (NVidia graphics chip) I was able to disable EDID checking to set my custom Modeline. I can't seem to be able to do that with the Intel driver. It ignores my custom resolution and gives me the first of two incorrect resolutions.

Could somebody please guide me in how to disable EDID so that I can view my TV in all its HD glory?

Thanks!

Lucas

PS. Here is my xorg.conf

Section "ServerLayout"

    Identifier     "Layout0"
    Screen      0  "PANASONIC-TV" 0 0
    InputDevice    "Keyboard0" "CoreKeyboard"
    InputDevice    "Mouse0" "CorePointer"

EndSection

Section "ServerFlags"
Option "Xinerama" "0"
Option "DontZap" "no"
EndSection

Section "Files"
EndSection

Section "InputDevice"

    # generated from default
    Identifier     "Mouse0"
    Driver         "mouse"
    Option         "Protocol" "auto"
    Option         "Device" "/dev/psaux"
    Option         "Emulate3Buttons" "no"
    Option         "ZAxisMapping" "4 5"
EndSection

Section "InputDevice"

    # generated from default
    Identifier     "Keyboard0"
    Driver         "kbd"
EndSection

Section "Monitor"

    # HorizSync source: edid, VertRefresh source: edid
    Identifier     "PANASONIC-TV"
    VendorName     "Unknown"
    ModelName      "PANASONIC-TV"
    Modeline "1920x1080" 74.25 1862 1972 2016 2200 1045 1071 1081 1124 +hsync +vsync Interlace
    HorizSync       15.0 - 45.0
    VertRefresh     59.0 - 61.0
    Option         "DPMS" "FALSE"
    Option "ExactModeTimingsDVI" "TRUE"
    Option "PreferredMode" "1920x1080"

EndSection

Section "Device"
Identifier "Device0"
Driver "intel"
Option "ModeValidation" "NoEdidModes"
Option "UseEDID" "FALSE"
Option "NoDDC" "TRUE"
Option "DDC" "FALSE"
EndSection

Section "Screen"

    Identifier     "PANASONIC-TV"
    Device         "Device0"
    Monitor        "PANASONIC-TV"
    DefaultDepth    24
    SubSection     "Display"
        Depth       24
                Modes     "1920x1080" "1600x900"
    EndSubSection

EndSection

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#1 lore raw
Quoted 20 lines · ----- "Lucas

Hi All,

Pardon me if this has been discussed before but I sure as heck can’t find the answer.  Perhaps my googling skills aren’t good enough.

Anyway, I have an AOpen DE 965-HG mini PC that has – surprise! – Intel 965 graphics.  It has an HDMI-out port which I have connected to my 42” plasma Panasonic TV.  The TV is a little old (circa 2005) so it apparently reports the wrong EDID information.   It also needs to be adjusted for overscan.  When I try to run Linux (any kind, but currently Ubuntu 10.04) X server it picks up the wrong resolution. With a different computer (NVidia graphics chip) I was able to disable EDID checking to set my custom Modeline.  I can’t seem to be able to do that with the Intel driver.  It ignores my custom resolution and gives me the first of two incorrect resolutions.

Quoted 6 lines

Could somebody please guide me in how to disable EDID so that I can view my TV in all its HD glory?

Funny thing,

I tried to find out the same info -- with no success. The only possible solution is to try and use xrandr to change the resolution after X is started. Use a combination of CVT (to create modelines) and xrandr to to change the resolutions.

Nasa

BTW: I'm still working on that for my monitor (check earlier thread started by me)
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#2 lore raw

On 2010/08/30 16:44 (GMT) Nasa composed:

Quoted 4 lines

I tried to find out the same info -- with no success. The only possible solution is to try and use xrandr to change the resolution after X is started. Use a combination of CVT (to create modelines) and xrandr to to change the resolutions.

Did you miss
http://lists.freedesktop.org/archives/intel-gfx/2010-August/007917.html or didn't it help any?

Signature · 6 lines
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409

Felix Miata  ***  http://fm.no-ip.com/
#3 lore raw

Thanks Guys!

I'll let you know how I make out with these suggestions when I get home.

Lucas

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Felix Miata
Sent: Monday, August 30, 2010 9:54 AM
To: [email protected]
Subject: Re: [Intel-gfx] Intel 965 and Custom Resolutions

On 2010/08/30 16:44 (GMT) Nasa composed:

Quoted 4 lines

I tried to find out the same info -- with no success. The only possible solution is to try and use xrandr to change the resolution after X is started. Use a combination of CVT (to create modelines) and xrandr to to change the resolutions.

Did you miss
http://lists.freedesktop.org/archives/intel-gfx/2010-August/007917.html or didn't it help any?
--
"The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation)

Team OS/2 ** Reg. Linux User #211409

Felix Miata *** http://fm.no-ip.com/
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw
In reply to Nasa
Quoted 3 lines · Nasa

The only possible solution is to try and use xrandr to change the resolution after X is started. Use a combination of CVT (to create modelines) and xrandr to to change the resolutions.

I used to have to do the same, and had this at the top of my .xinitrc:

test -x /usr/bin/xrandr \

 && /usr/bin/xrandr \
    --newmode "1280x720" 74.25 1280 1390 1430 1650 720 725 730 750 \

&& /usr/bin/xrandr --addmode HDMI1 1280x720 \
&& /usr/bin/xrandr --output HDMI1 --mode 1280x720

Eventually, support was added to feed specific modes into KMS at boot via kernel command line parameters and so I did that for a while (I no longer have info on what that looked like, sorry).

At some point all that ceased to be necessary for me, and the driver now figures out that it should use ATSC 720p timings with my HDTV even though it returns what appears to be an entirely malformed EDID block (I'm just booting with video=i915:modeset=1 now). Not sure why, but those timings appear in the list of "probed modes for output HDMI1" in my Xorg.0.log even though it's not in the output of 'sudo get-edid | parse-edid' (a fallback resolution of some sort from the kernel, maybe?).

Signature · 4 lines
{ IRL(Jeremy_Stanley); PGP(97AE496FC02DEC9FC353B2E748F9961143495829);
SMTP([email protected]); IRC([email protected]#ccl); ICQ(114362511);
AIM(dreadazathoth); YAHOO(crawlingchaoslabs); FINGER([email protected]);
MUD([email protected]:6669); WWW(http://fungi.yuggoth.org/); }
#5 lore raw
Quoted 8 lines · The Fungi

At some point all that ceased to be necessary for me, and the driver now figures out that it should use ATSC 720p timings with my HDTV even though it returns what appears to be an entirely malformed EDID block (I'm just booting with video=i915:modeset=1 now). Not sure why, but those timings appear in the list of "probed modes for output HDMI1" in my Xorg.0.log even though it's not in the output of 'sudo get-edid | parse-edid' (a fallback resolution of some sort from the kernel, maybe?).

Pretty sure parse-edid doesn't understand CEA extension blocks. The kernel does though.

- ajax

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#6 lore raw
Quoted 2 lines · Adam Jackson

Pretty sure parse-edid doesn't understand CEA extension blocks. The kernel does though.

Aha--that probably explains it. Are there any userspace alternatives to parse-edid which implement this? If not, I'll look at coming up with a patch.

Signature · 4 lines
{ IRL(Jeremy_Stanley); PGP(97AE496FC02DEC9FC353B2E748F9961143495829);
SMTP([email protected]); IRC([email protected]#ccl); ICQ(114362511);
AIM(dreadazathoth); YAHOO(crawlingchaoslabs); FINGER([email protected]);
MUD([email protected]:6669); WWW(http://fungi.yuggoth.org/); }
#7 lore raw
Quoted 7 lines · The Fungi
Quoted 2 lines · Adam Jackson

Pretty sure parse-edid doesn't understand CEA extension blocks. The kernel does though.

Aha--that probably explains it. Are there any userspace alternatives to parse-edid which implement this? If not, I'll look at coming up with a patch.

git clone git://anongit.freedesktop.org/xorg/app/edid-decode

I try to keep that reasonably up to date, though it's lagging a little. Do let me know if you have patches for it.

- ajax

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#8 lore raw
Quoted 4 lines · Adam Jackson

git clone git://anongit.freedesktop.org/xorg/app/edid-decode

I try to keep that reasonably up to date, though it's lagging a little. Do let me know if you have patches for it.

Unfortunately, that doesn't seem to explain where my system's getting 1280x720 from. Passing the EDID returned by either get-edid or xrandr --verbose returns only 720x480 and 1920x540(interlaced).

The EDID as returned by xrandr --verbose is:

00ffffffffffff004dd9f80101010101
000e0103800000780a0dc9a057479827
12484c20000001010101010101010101
010101010101011d8018711c1620582c
250010090000009e8c0ad08a20e02d10
103e9600040300000018000000fc0053
4f4e592054560a2020202020000000fd
003b3d0f2e08000a20202020202001df

Is it possible xrandr doesn't report the entirety of the EDID the kernel sees?

Signature · 4 lines
{ IRL(Jeremy_Stanley); PGP(97AE496FC02DEC9FC353B2E748F9961143495829);
SMTP([email protected]); IRC([email protected]#ccl); ICQ(114362511);
AIM(dreadazathoth); YAHOO(crawlingchaoslabs); FINGER([email protected]);
MUD([email protected]:6669); WWW(http://fungi.yuggoth.org/); }
#9 lore raw
Quoted 21 lines · The Fungi
Quoted 4 lines · Adam Jackson

git clone git://anongit.freedesktop.org/xorg/app/edid-decode

I try to keep that reasonably up to date, though it's lagging a little. Do let me know if you have patches for it.

Unfortunately, that doesn't seem to explain where my system's getting 1280x720 from. Passing the EDID returned by either get-edid or xrandr --verbose returns only 720x480 and 1920x540(interlaced).

The EDID as returned by xrandr --verbose is:

   00ffffffffffff004dd9f80101010101
   000e0103800000780a0dc9a057479827
   12484c20000001010101010101010101
   010101010101011d8018711c1620582c
   250010090000009e8c0ad08a20e02d10
   103e9600040300000018000000fc0053
   4f4e592054560a2020202020000000fd
   003b3d0f2e08000a20202020202001df
Quoted 2 lines

Is it possible xrandr doesn't report the entirety of the EDID the kernel sees?

It's possible but improbable. What kernel are you running anyway?

- ajax

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#10 lore raw
Quoted 1 line · Adam Jackson

It's possible but improbable. What kernel are you running anyway?

That system had 2.6.35-rc5 i686 with PAE at the last boot (needed some patches for Wacom BT/HID support which got integrated in rc4 or rc5). I'll most likely replace that with a 2.6.35 release version next time I get around to rebooting, unless 2.6.36 is out by then.

Signature · 4 lines
{ IRL(Jeremy_Stanley); PGP(97AE496FC02DEC9FC353B2E748F9961143495829);
SMTP([email protected]); IRC([email protected]#ccl); ICQ(114362511);
AIM(dreadazathoth); YAHOO(crawlingchaoslabs); FINGER([email protected]);
MUD([email protected]:6669); WWW(http://fungi.yuggoth.org/); }
#11 lore raw
Quoted 7 lines · The Fungi
Quoted 1 line · Adam Jackson

It's possible but improbable. What kernel are you running anyway?

That system had 2.6.35-rc5 i686 with PAE at the last boot (needed some patches for Wacom BT/HID support which got integrated in rc4 or rc5). I'll most likely replace that with a 2.6.35 release version next time I get around to rebooting, unless 2.6.36 is out by then.

Sigh, okay. You should be able to get the complete EDID block out of /sys/class/drm/card0/*/edid, at least if you have in your kernel:

commit 7466f4cc508878a8328dff1c328a2b4108888d2e

Author: Adam Jackson <[email protected]>
Date:   Mon Mar 29 21:43:23 2010 +0000

drm/edid: Remove arbitrary EDID extension limit

which should already be present in 2.6.35. But the X driver looks like it never sets MONITOR_EDID_COMPLETE_RAWDATA, which means the X core doesn't know that any sections beyond the first are present, so it won't ever hand back more than 128 bytes to clients. Boo.

This should help:

---

src/drmmode_display.c +8 -8
index 6f7e5f1..886c58f 100644
@@ -731,6 +731,7 @@ drmmode_output_get_modes(xf86OutputPtr output)
 	drmModePropertyPtr props;
 	struct fixed_panel_lvds *p_lvds;
 	drmModeModeInfo *mode_ptr;
+	xf86MonPtr mon = NULL;
 
 	/* look for an EDID property */
 	for (i = 0; i < koutput->count_props; i++) {
@@ -751,14 +752,13 @@ drmmode_output_get_modes(xf86OutputPtr output)
 		drmModeFreeProperty(props);
 	}
 
-	if (drmmode_output->edid_blob)
-		xf86OutputSetEDID(output,
-				  xf86InterpretEDID(output->scrn->scrnIndex,
-						    drmmode_output->edid_blob->data));
-	else
-		xf86OutputSetEDID(output,
-				  xf86InterpretEDID(output->scrn->scrnIndex,
-						    NULL));
+	if (drmmode_output->edid_blob) {
+		mon = xf86InterpretEDID(output->scrn->scrnIndex,
+					drmmode_output->edid_blob->data);
+		if (drmmode->edid_blob->length > 128)
+			mon->flags |= MONITOR_EDID_COMPLETE_RAWDATA;
+	}
+	xf86OutputSetEDID(output, mon);
 
 	/* modes should already be available */
 	for (i = 0; i < koutput->count_modes; i++) {

---

- ajax

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#12 lore raw
Quoted 2 lines · Adam Jackson

Sigh, okay. You should be able to get the complete EDID block out of /sys/class/drm/card0/*/edid

[...]

Aha! Mystery solved... so here's the diff in edid-decode output between feeding it the EDID reoprted by get-edid/xrandr --verbose and the contents of /sys/class/drm/card0/card0-HDMI Type A-1/edid:

@@ -36,9 +36,41 @@
Has 1 extension blocks
Checksum: 0xdf

-Unknown extension block
+CEA extension block
+Extension version: 3
+22 bytes of CEA data
+ Video data block

+    VIC 05 (native)
+    VIC 02
+    VIC 03
+    VIC 04
+    VIC 06
+    VIC 07
+  Audio data block
+  Speaker allocation data block
+  Vendor-specific data block, OUI 000c03 (HDMI)

+ Source physical address 1.0.0.0
+Basic audio support
+Supports YCbCr 4:4:4
+Supports YCbCr 4:2:2
+6 native detailed modes
+Detailed mode: Clock 74.250 MHz, 16 mm x 9 mm
+ 1280 1390 1430 1650 hborder 0
+ 720 725 730 750 vborder 0
+ +hsync +vsync
+Detailed mode: Clock 27.000 MHz, 4 mm x 3 mm
+ 1440 1478 1602 1716 hborder 0
+ 240 244 247 262 vborder 0
+ -hsync -vsync interlaced
+Detailed mode: Clock 27.000 MHz, 16 mm x 9 mm

+                720  736  798  858 hborder 0
+                480  489  495  525 vborder 0

+ -hsync -vsync
+Detailed mode: Clock 27.000 MHz, 16 mm x 9 mm
+ 1440 1478 1602 1716 hborder 0
+ 240 244 247 262 vborder 0
+ -hsync -vsync interlaced
+Checksum: 0x97

-EDID block does NOT conform to EDID 1.3!
- Digital display field contains garbage: 80
EDID block does not conform at all!
Bad year of manufacture

Signature · 4 lines
{ IRL(Jeremy_Stanley); PGP(97AE496FC02DEC9FC353B2E748F9961143495829);
SMTP([email protected]); IRC([email protected]#ccl); ICQ(114362511);
AIM(dreadazathoth); YAHOO(crawlingchaoslabs); FINGER([email protected]);
MUD([email protected]:6669); WWW(http://fungi.yuggoth.org/); }
author
#13 lore raw
In reply to Nasa

Okay,

So I used xrandr to add my modeline but X just shows half a screen and then restarts. When I use CVT to create Modelines they all fail.

The modeline from my xorg.conf file works perfectly on an NVidia machine, what would the difference be?

Lucas