Created attachment 24513 [details]
`xrandr -q` output with KMS
Created attachment 24514 [details]
Xorg log without KMS (working as expected)
Created attachment 24515 [details]
`xrandr -q` output without KMS (working as expected)
Well yeah, except for the fact that it appears for me even when the screen is *connected* during boot. Would that be the same bug? Cheers, Nico Let Jesse to decide. 21030 has a patch now that you could try... if it works it's a DUP. :) Tried the patch, but it didn't help unfortunately. The logs still say the same thing; in fact, nothing seems to have changed for me. Cheers, Nico Can you modprobe your drm module with debug=1, plug in your external display, then modprobe i915 with modeset=1 and attach the dmesg here? We could be failing to get EDID on your config or totally failing to detect... Created attachment 24663 [details]
dmesg output after manually loading the modules (with debug)
Did the following:
$ modprobe drm debug=1
External DVI screen plugged in.
$ modprobe i915 modeset=1
Result: External monitor *not* recognized, black LVDS screen, and a panic. :/
dmesg output attached.
Cheers,
Nico
Ah that helps, so it looks like we're failing to read the EDID from your DVI connected monitor. There have been kernel fixes to the SDVO code recently though; can you give 2.6.30-rc4 a try? Created attachment 25443 [details]
output of `xrandr -q` with KMS on kernel 2.6.30_rc4
Alright, I tested 2.6.30_rc4, unfortunately with no success.
I'll post the logs in a minute.
Created attachment 25444 [details]
Xorg log with KMS on kernel 2.6.30_rc4
Hm, the manual kernel loading from as above doesn't produce any output on dmesg anymore, not sure why. If there's anything else I can provide, let me know. looks like this is causing the problem... (II) intel(0): Output LVDS1 connected (II) intel(0): Output DVI1 connected (II) intel(0): Output TV1 connected What's the model of this machine? What Michael said in comment #15 is right. In the KMS mode the LVDS is also detected. In fact maybe its detection should be skipped. At the same time it seems that tv is also detected in KMS mode. But in UMS mode it is ignored. Will you please attach the xorg.conf used in UMS mode? Will you please attach the output of dmidecode, vbios dump, acpidump? Thanks. Alright, I'm now on kernel 2.6.30, with xorg-server-1.6.1.90,1 and the intel 2.7.1 driver. The problem persists. Appending the logs.. Created attachment 26670 [details]
acpidump
Created attachment 26671 [details]
dmesg output with KMS
Created attachment 26672 [details]
dmicode output
Created attachment 26673 [details]
Xorg log with KMS
Created attachment 26674 [details]
Xorg log without KMS (working as expected)
Created attachment 26675 [details]
`xrandr -q` output with KMS
Created attachment 26676 [details]
`xrandr -q` output without KMS (working as expected)
Oh, and I'm not using an explicit xorg.conf anymore (see Xorg logs). How do I actually generate a 'vbios dump'? Cheers, Nico From the dmesg log in comment #19 it seems that there exists the following error messages: >i2c-adapter i2c-3: unable to read EDID block. >i915 0000:00:02.0: DVI-D-1: no EDID data Is this caused by that the EDID can't be obtained for SDVO-DVI device? Hi, Nico there exists the inconsistence in the log between comment #2 and comment #24. For example: there is no LVDS device in comment #2. But there exists the LVDS device in comment #24. Will you please double check it? It will be great if you can add the "option ModeDebug on" in xorg.conf and attach the Xorg.log with KMS enabled/disabled. Thanks. (In reply to comment #25) > Oh, and I'm not using an explicit xorg.conf anymore (see Xorg logs). > to add ModeDebug option , you need an xorg.conf file. to generate a xorg.conf: If your distribution doesn't ship xorg.conf, you can use 'X --configure' to let X spit out the default configuration it uses and save it as /etc/X11/xorg.conf.) > How do I actually generate a 'vbios dump'? > # cd /sys/devices/pci0000\:00/0000\:00\:02.0/ # echo 1 > rom # cat rom > /tmp/rom.bin # echo 0 > rom > Cheers, > Nico > Hi, you were right ykzhao, you can forget about the `xrandr -q` of comment #3. My laptop's native screen is LVDS (@1280x800), the external one TMDS-1 (@1440x900), correctly recognized at no KMS. ModeDebug'ged Xorg logs following. Not sure if they help so much, though, as the phenomenon (blank external screen) appears right after KMS has been loaded. This made me think it might be a kernel driver issue.. Anyway, you guys are on top of it. Cheers, Nico Created attachment 26860 [details]
X.org log with KMS (ModeDebug)
Created attachment 26861 [details]
X.org log without KMS (working as expected) (ModeDebug)
Oh, and when I follow the recipe for vbios dumps (which is all I can do there b/c I totally don't know my way in /devices/), I only get an almost empty file (just some disclaimer about free software). In order to reduce complexity, please modify your xorg.conf as below to ignore non-existent TV output, then upload your xorg.conf with modedebug optoin on, thanks.
Section "Device"
...
Option "monitor-TV" "TV"
...
EndSection
...
Section "Monitor"
Identifier "TV"
Option "Ignore" "True"
EndSection
Created attachment 26923 [details]
X configuration with ignored TV
Hm, what am I doing wrong? The complexity isn't reduced, rather the line
"monitor-TV" is not used
in the logs. xorg.conf appended.
(In reply to comment #33) > Created an attachment (id=26923) [details] > X configuration with ignored TV > > Hm, what am I doing wrong? The complexity isn't reduced, rather the line > > "monitor-TV" is not used > > in the logs. xorg.conf appended. > yeah, in KMS mode, you need to name it like "monitor-TV1". the name was changed in KMS. :( Progress, progress! In KMS, I disabled the incorrectly recognized TV-out, and now actually got the DVI connector to display my desktop! Unfortunately, the modelines are still not correct. Logs appended. Created attachment 26952 [details]
X.org log without KMS (ModeDebug, TV-out ignored)
Created attachment 26953 [details]
X.org log with KMS (ModeDebug, TV-out ignored)
Created attachment 26954 [details]
`xrandr -q` output with KMS (TV-out ignored)
Same problem here since switching to KMS. EDID does not seem to be read off the external device. Also, Apple mini-DVI to DVI / mini-DVI to VGA adapter detection does not seem to be working right either -- meaning when VGA adapter is plugged-in both VGA1 and DVI1 show as "connected" with "xrandr -q". Running on: x11-base/xorg-server-1.6.1.901-r3 x11-drivers/xf86-video-intel-2.7.1 media-libs/mesa-7.4.2 x11-libs/libdrm-2.4.11 sys-kernel/gentoo-sources-2.6.30-r1 And a question.. are FB_MODE_HELPERS and FIRMWARE_EDID kernel options related to this problem or not? I tried both built-in without success.. Created attachment 26975 [details]
X.org log, KMS enebled, mini-dvi to vga adapter plugged-in after xorg server start.
Created attachment 26976 [details]
X.org log, KMS enebled, mini-dvi to vga adapter plugged-in before xorg server start.
Created attachment 26977 [details]
X.org config
Could you please confirm the patch commit id 619ac3b75a1e9b2df66857f6a0fb466f1da5fa9e, which will succeed in reading EDID in KMS. Meanwhile please ignore TV1 too. Thanks Ma Ling (In reply to comment #43) Mentioned patch is already included in gentoo-sources-2.6.30-r1. Is there another tool to read EDID off external device besides xrandr? Indeed, the patch went into mainline at 2.6.30-rc8 I think. I'm using 2.6.30 (and got the screens misbehave). Uros, Mac is probably a different bug. would you please open a new bug to track rahter than update to this one? thanks. Err -- I'm on Mac(book 3,1) here as well, so I guess it's the exact same issue. I see. in fact, we are looking for someone who could help us do some testing on Macbook. Bugs summary updated. Macbook 4,1 here. If there's anything else I can do, let me know. Nico, Uros, Does any of you have HW to help test the bug# 11211? It's about a Mini-DVI-to-SVideo for connect MacBook to a TV... If you can help, would you please kindly help to post your update to that bug instead of this one? thanks a lot! Hm, unfortunately I don't have an S-Video connector, just DVI here.. :/ (In reply to comment #36) > Created an attachment (id=26952) [details] > X.org log without KMS (ModeDebug, TV-out ignored) (II) intel(0): Using fuzzy aspect match for initial modes (II) intel(0): Output LVDS using initial mode 1024x768 (II) intel(0): Output TMDS-1 using initial mode 1024x768 In UMS , edid has been found correctly, but X chose 1024x768. However in KMS it seem not to read EDID correctly, I will send one patch later to check it. Thanks Ma Ling (In reply to comment #52) > (In reply to comment #36) > > Created an attachment (id=26952) [details] [details] > > X.org log without KMS (ModeDebug, TV-out ignored) > > (II) intel(0): Using fuzzy aspect match for initial modes > (II) intel(0): Output LVDS using initial mode 1024x768 > (II) intel(0): Output TMDS-1 using initial mode 1024x768 > > In UMS , edid has been found correctly, but X chose 1024x768. given the modelines from both monitor, I think X is doing right as initial configuration here. > However in KMS it > seem not to read EDID correctly, I will send one patch later to check it. > this should be the root cause. the hacky code for Mac is commented out in KMS, compared with UMS, in get_modes function. > Thanks > Ma Ling > (In reply to comment #39) > > Also, Apple mini-DVI to DVI / mini-DVI to VGA adapter detection does not seem > to be working right either -- meaning when VGA adapter is plugged-in both VGA1 > and DVI1 show as "connected" with "xrandr -q". > this ought to be another thing need to fix - I think due to special HW design of Mac, its TMDS sdvo device seems also think it's connected with a monitor, when any MiniDVI adapter is used. We need to do a special check for Mac in sdvo's detect call back func, similar as we did for DVI-I case in the current code. (In reply to comment #32) > In order to reduce complexity, please modify your xorg.conf as below to ignore > non-existent TV output, then upload your xorg.conf with modedebug optoin on, > thanks. > > Section "Device" > ... > Option "monitor-TV" "TV" > ... > EndSection > ... > Section "Monitor" > Identifier "TV" > Option "Ignore" "True" > EndSection > even though we quirk it using conf file, we should confirm if this has been fixed by your TV detection patches. Looks UMS works even without the ignore in xorg.conf. Would you pls attach your KMS version of TV detection patchs to verify? Or is it Keith's loaddetect fix (03d6069912babc07a3da20e715dd6a5dc8f0f867)? Wow, many fixes need in one bug... (In reply to comment #50) > Does any of you have HW to help test the bug# 11211? It's about a > Mini-DVI-to-SVideo for connect MacBook to a TV... Found one.. will post there. (In reply to comment #56) > (In reply to comment #50) > > Does any of you have HW to help test the bug# 11211? It's about a > > Mini-DVI-to-SVideo for connect MacBook to a TV... > Found one.. will post there. It's very kind of you! :) Created attachment 27295 [details]
please try the debug patch in KMS mode, thanks
Alright, so I applied the patch, and the kernel just hanged right before it would normally switch to KMS (that's about two, three seconds into boot). The last things printed to screen are ==================== *snip* ==================== i2c adapter i2c-3: unable to read EDID block. i915 0000:00:02.0: DVI-D-I no EDID data i2c adapter i2c-3: unable to read EDID block. i915 0000:00:02.0: DVI-D-I no EDID data ==================== *snap* ==================== (In reply to comment #56) > (In reply to comment #50) > > Does any of you have HW to help test the bug# 11211? It's about a > > Mini-DVI-to-SVideo for connect MacBook to a TV... > > Found one.. will post there. > any update, Uros? Hi Nico, Could you please confirm the KMS patch (commit 8ed9a5bc9c9425ef93a1b03b418300a5e18b2361) has been included your version, if yes, please remove TV1 ignore optoin in xorg.conf and upload log file with modedebug option on. Thanks Ma Ling Created attachment 27502 [details]
Xorg log with KMS on kernel 2.6.31_rc2 (ModeDebug)
Alright, so I didn't have the patch installed, but I'm now running on 2.6.31_rc2 which includes the thing.
Xorg log attached.
There's no change comparing to the situation before the kernel patch: DVI1 fine, Modelines on external screen incorrect.
Cheers,
Nico
(In reply to comment #62) > Created an attachment (id=27502) [details] > Xorg log with KMS on kernel 2.6.31_rc2 (ModeDebug) > Alright, so I didn't have the patch installed, but I'm now running on > 2.6.31_rc2 which includes the thing. > Xorg log attached. > There's no change comparing to the situation before the kernel patch: DVI1 > fine, Modelines on external screen incorrect. > Cheers, > Nico (II) intel(0): Output VGA1 disconnected (II) intel(0): Output LVDS1 connected (II) intel(0): Output DVI1 connected (II) intel(0): Output TV1 disconnected log tell us TV is disconnected, I will refine patch to fetch EDID for DVI output in KMS mode. Thanks Ma Ling Created attachment 27782 [details] [review] pleaset try the patch on your machine in KMS mode, thanks. (In reply to comment #54) > (In reply to comment #39) > > > > Also, Apple mini-DVI to DVI / mini-DVI to VGA adapter detection does not seem > > to be working right either -- meaning when VGA adapter is plugged-in both VGA1 > > and DVI1 show as "connected" with "xrandr -q". > > > this ought to be another thing need to fix - I think due to special HW design > of Mac, its TMDS sdvo device seems also think it's connected with a monitor, > when any MiniDVI adapter is used. We need to do a special check for Mac in > sdvo's detect call back func, similar as we did for DVI-I case in the current > code. The patch intends to fetch edid for Mac machine. thanks Ma Ling Hm, against which kernel version is the patch? It'd help me a lot if it was at least an RC, but rc-2 and rc-3 didn't work. (In reply to comment #65) > Hm, against which kernel version is the patch? It'd help me a lot if it was at > least an RC, but rc-2 and rc-3 didn't work. v2.6.29-rc1-26127-gdff33cf. Thanks Ma Ling Ding-ding-ding: Success! I applied the patch to 2.6.31_rc2 (with the necessary adaptations of course), and now the external screen is correctly identified (as DVI1) with the correct modelines. Nice! By default, the resolution is not the highest, but that might have other reasons. Logs appended. Thanks! Nico Created attachment 27804 [details]
xrandr -q output with the latest patch
Created attachment 27805 [details]
X.org log with KMS and latest patch
Created attachment 28421 [details] [review] Read the EDID for SDVO-DVI by using VGA DDC Will you please try the debug patch on the latest Eric's drm-intel-next tree and see whether the EDID can be obtained for the SDVO-DVI? Thanks. Alright, I applied the patch to Eric's kernel and things seem to work out pretty fine. Logs attached. Created attachment 28476 [details]
Xorg log with KMS on Eric Anholt's kernel
Created attachment 28477 [details]
dmesg on Eric Anholt's kernel
Created attachment 28478 [details]
`xrandr -q` on Eric Anholt's kernel
Errm, err, by the way, Eric's kernel still works alright. (In reply to comment #75) > Errm, err, by the way, Eric's kernel still works alright. Thanks for the test. Now it seems that the EDID can be obtained correctly on Mini-dvi after applying the patch in comment #70. As the following commit is already shipped in Eric's drm-intel-next tree, it will be marked as resolved. commit 57cdaf90f5f607eb029356074fefb66c9b1c0659 Author: Keith Packard <keithp@keithp.com> Date: Fri Sep 4 13:07:54 2009 +0800 drm/I915: Use the CRT DDC to get the EDID for DVI-connector on Mac Thanks. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.
Created attachment 24512 [details] Xorg log with KMS Hi, I just tried KMS with the 2.6.29.1 kernel and the intel 2.6.99.902 driver. Things work out well in the sense that my laptop display really gets a native 1280x800 framebuffer console in the beginning, and the xserver starts up smoothly. The only thing is that my external screen (1440x900) doesn't get recognized at all (no signal all the time). Logs appended. If you need any more info, drop a line. Cheers, Nico PS. In case this bug belongs to some other place (kernel?) please let me know.