Description
Tomasz Grabiec
2007-12-11 11:20:45 UTC
Please verify with the latest git, and provide a log with -logverbose 7, as an attachment. Tomasz: any feedback on this one? Could you please retest with latest git? Created attachment 13466 [details]
Xorg.0.log
I have similar problems with latest git and RS690 (Asus M2A-VM HDMI), the X starts with black screen and approximately each 10 seconds the right picture blinks for a half of second. Monitor is connected through DVI, clean boot without fglrx module, I've updated from some pre-1.0.0 versions.
Petr, is this new or did it happen before. If it is new - can you find out when it was introduced? Created attachment 13496 [details]
more logs
Using bisect I've found out that it was introduced with commit 47f8a6f784acc69da001b318ba3d80a9fa4c883d (Add support for HDMI connectors).
Since that the DVI output doesn't work neither with VGA output connected nor DVI only (this never worked for me). I'm attaching some more logs and config files. Card is RS690 bases (ASUS M2A-VM HDMI with some fresh BIOS).
Bugzilla Upgrade Mass Bug Change NEEDSINFO state was removed in Bugzilla 3.x, reopening any bugs previously listed as NEEDSINFO. - benjsc fd.o Wrangler Petr, just as an update: I've bought exactly this board and today I even finally managed to buy a case for it. So I will be able to look into this problem in the near future. For reference: Original problem: LVTMA on M76 with LVDS panel output. Tomasz, Please verify whether this bug still exists against the current driver, there were some further fixes pushed in that might influence this. When there is no response within 7 days, this bug will be closed. No feedback for over a month. closing. I'm having the same (or similar) problem with my Radon HD 2600 with latest from git. I have an HP 8510p laptop + docking station + HP LP2465 external dvi monitor. Symptoms: A) If my laptop is un-docked everything is ok. B) If my laptop is docked (but with no DVI monitor) everything is ok. C) If my laptop is docked with an external DVI monitor plugged in then both the laptop panel and the external DVI go black on login to desktop. Work around for option C: I dock my laptop, ensure DVI monitor is unplugged, login to my desktop, connect the DVI monitor, run xrandr commands (see below), success. XRandR commands to turn on DVI: xrandr --newmode "1920x1200_60.00" 162 1920 2048 2256 2592 1200 1201 1204 1242 -HSync +Vsync xrandr --addmode DVI-D_1 1920x1200_60.00 xrandr --output DVI-D_1 --mode 1920x1200_60.00 xrandr --output DVI-D_1 --right-of PANEL System info: $ cat /etc/redhat-release Fedora release 9 (Sulphur) $ uname -a Linux devlap4-king 2.6.25.11-97.fc9.i686 #1 SMP Mon Jul 21 01:31:09 EDT 2008 i686 i686 i386 GNU/Linux $ lspci | grep Radeon 01:00.0 VGA compatible controller: ATI Technologies Inc M76 [Radeon Mobility HD 2600 Series] 01:00.1 Audio device: ATI Technologies Inc RV630/M76 audio device [Radeon HD 2600 Series] I will attach the xorg log, xorg conf, and xrandr output as attachments next. Created attachment 17959 [details]
my xorg log
Created attachment 17960 [details]
xorg when external DVI is *not* connected
Created attachment 17961 [details]
xorg when external is connected
Created attachment 17962 [details]
output from the command "xrandr -q; xrandr -q --verbose"
Thanks for the logs. I can see one problem in one of the logs for which I've just pushed a fix. I'm not sure if this will address your problem already. Still it would be worthwhile to test. Created attachment 17976 [details]
RADEONHD: version 1.2.1, built from git branch master, commit 996d309f
Woops, should have mentioned in the last attachment that the problem still persists. Could you please try the following: with no panel attached start X, then attach the panel and run 'randr -q'. This should detect the newly attached panel. You should see a 1920x1200 in the list for DVI-D_1. Now do: xrandr --output DVI-D_1 --mode 1920x1200. Does this light up your screen, too? What I don't understand in your log in comment #12: RandR seems to have been called several times. The last state that was set was: RRCrtc #0 [rhd CRTC 1]: active 1 [1] mode unnamed (1920x1200) +0+0 RRCrtc #1 [rhd CRTC 2]: active 0 [0] mode unnamed (0x0) +0+0 RROut VGA_1 [Out DAC A Conn VGA 1] Crtc null [null] [inactive] disconnected RROut PANEL [Out LVDS Conn PANEL] Crtc null [null] [inactive] connected RROut DVI-D_1 [Out TMDS A Conn DVI-D 1] Crtc CRTC 1 [CRTC 1] [active] connected RROut TV_7PIN_DIN [Out DAC B Conn TV 7PIN_DIN] Crtc null [null] [inactive] disconnected This however shows that the PANEL was set inactive and only the DVI is on. Did you run xrandr by hand or could it be that some desktop helper was interfering here? (Why the DVI isn't lit is a different issue of course). Could you please start a bare X from a text console and see what the outcome there is? You should be able to do this logged in as a normal user just running 'X' on the command line. > Could you please try the following: with no panel attached start X, then attach > the panel and run 'randr -q'. This should detect the newly attached panel. > You should see a 1920x1200 in the list for DVI-D_1. Now do: > xrandr --output DVI-D_1 --mode 1920x1200. Does this light up your screen, too? No, the screen doesn't come on after running that command. That's why I use the xrandr commands to add a custom mode as seen in comment #11. I'm not sure why the detected modes like "1920x1200" don't work though. > This however shows that the PANEL was set inactive and only the DVI is on. Did > you run xrandr by hand or could it be that some desktop helper was interfering > here? Not that I know of. > Could you please start a bare X from a text console and see what the outcome there is? Coming right up. Created attachment 18002 [details]
bare X from a text console
(In reply to comment #20) > > Could you please try the following: with no panel attached start X, then attach > > the panel and run 'randr -q'. This should detect the newly attached panel. > > You should see a 1920x1200 in the list for DVI-D_1. Now do: > > xrandr --output DVI-D_1 --mode 1920x1200. Does this light up your screen, too? With panel I was referring to the DVI monitor, but I think you understood this already :) > > No, the screen doesn't come on after running that command. That's why I use the > xrandr commands to add a custom mode as seen in comment #11. I'm not sure why > the detected modes like "1920x1200" don't work though. Right. That's what surprises me, too. > Coming right up. > Also the external LCD screen or just the internal laptop panel? Because the external panel is driven with exactly the same mode as the one that is set when you do xrandr --output DVI-D_1 --mode 1920x1200. But the fact that things are coming right up and the data from the log file indicate that there is something running on your desktop which shuts down the internal panel at least. Of course what you can do is to try out the atombios_support branch on git. You need to set the option 'Option "UseAtomBIOS" "True"'. I'd be curious if this LCD mode will work there. >>> Could you please start a bare X from a text console and see what the outcome there is? >> Coming right up. >Also the external LCD screen or just the internal laptop panel? Sorry for the confusion. Silly english language. By "coming right up" what I meant was that I would be attaching the log file, which is in comment #12. (In reply to comment #23) > Sorry for the confusion. Silly english language. By "coming right up" what I > meant was that I would be attaching the log file, which is in comment #12. > Ah, OK. so what do you get, when you run a bare server? Does the laptop panel light up? Because in the previous log files there definitely is indication that something turned off your panel as is visible from the snippet of log file I put into comment #19. I see several device probes in those log files which were triggered by some Xrandr interaction. The log file in comment #21 doesn't contain those interactions. There the state is that both screens are on. I don't expect the external DVI display to come up as there seems to be a problem with the mode line, but since the panel is on in the single head situation I would thing that it is on when you start a bare Xserver with the external DVI attached, too. > Ah, OK. so what do you get, when you run a bare server? Does the laptop panel > light up? Ok so when I run 'X' the laptop panel lights up, with the DVI connected or disconnected. This behaviour is different from when i run 'startx'. One thing to note though is that when the DVI *is* connected the laptop gets a non native resolution of 1920x1200Scaled that looks blurry. > Because in the previous log files there definitely is indication that > something turned off your panel as is visible from the snippet of log file I > put into comment #19. The Xorg log in comment #21 was produced by running "X -logverbose". The Xorg log in comment #14 was produced by running "startx -- -logverbose 7". This is starting to sound like a gnome desktop thing then? Something is calling that xrandr command to turn off my PANEL. (In reply to comment #25) > > Ah, OK. so what do you get, when you run a bare server? Does the laptop panel > > light up? > > Ok so when I run 'X' the laptop panel lights up, with the DVI connected or > disconnected. This behaviour is different from when i run 'startx'. One thing > to note though is that when the DVI *is* connected the laptop gets a non native > resolution of 1920x1200Scaled that looks blurry. Yes, that's the scaled mode. This way you get the same view on the laptop as on the external screen. At least I know now that the laptop panel lights up when there is no application trying to control it. We now have to figure out why your monitor doesnt light up with the mode it advertises itself. The problem that both heads are dark seems to be induced by some desktop helper that tries to improve your user experience but helps only to turn everything dark. > > > Because in the previous log files there definitely is indication that > > something turned off your panel as is visible from the snippet of log file I > > put into comment #19. > > The Xorg log in comment #21 was produced by running "X -logverbose". The Xorg > log in comment #14 was produced by running "startx -- -logverbose 7". This is > starting to sound like a gnome desktop thing then? Something is calling that > xrandr command to turn off my PANEL. > Yes, definitely. Life was definitely easier before such thingies started to appear. At least I know that I have to tell people to run a 'bare' Xserver when they claim some heads are not lighting up. Created attachment 18037 [details] xorg log when using atom bios branch (In reply to comment #26) > We now have to figure out why your monitor doesnt light up with the mode it > advertises itself. Ok I tried out the atom bios branch. Unfortunately it crashes on startup for me. $ X -logverbose 7 X.Org X Server 1.4.99.905 (1.5.0 RC 5) Release Date: 5 September 2007 X Protocol Version 11, Revision 0 Build Operating System: Linux 2.6.18-92.1.6.el5xen i686 Current Operating System: Linux devlap4-king 2.6.25.11-97.fc9.i686 #1 SMP Mon Ju l 21 01:31:09 EDT 2008 i686 Build Date: 02 July 2008 01:36:37PM Build ID: xorg-x11-server 1.4.99.905-2.20080702.fc9 Before reporting problems, check http://wiki.x.org to make sure that you have the latest version. Module Loader present Markers: (--) probed, (**) from config file, (==) default setting, (++) from command line, (!!) notice, (II) informational, (WW) warning, (EE) error, (NI) not implemented, (??) unknown. (==) Log file: "/var/log/Xorg.0.log", Time: Thu Jul 31 09:53:21 2008 (==) Using config file: "/etc/X11/xorg.conf" rhd_biosscratch.c:636: rhdBIOSScratchUpdateBIOSScratchForOutput: Assertion 'Devi ce != atomNone' failed. Backtrace: 0: X(xf86SigHandler+0x79) [0x80bc189] 1: [0x110400] 2: [0x110416] 3: /lib/libc.so.6(kill+0x16) [0x451976] 4: /usr/lib/xorg/modules/drivers//radeonhd_drv.so(RhdAssertFailed+0x51) [0x3a1b5 1] 5: /usr/lib/xorg/modules/drivers//radeonhd_drv.so [0x3bd69f] 6: /usr/lib/xorg/modules/drivers//radeonhd_drv.so [0x3bd791] 7: /usr/lib/xorg/modules/drivers//radeonhd_drv.so [0x3b496d] 8: X(xf86SetDesiredModes+0x45) [0x80e51d5] 9: /usr/lib/xorg/modules/drivers//radeonhd_drv.so(RHDRandrModeInit+0x8f) [0x3b41 af] 10: /usr/lib/xorg/modules/drivers//radeonhd_drv.so [0x39e565] 11: X(AddScreen+0x19f) [0x806acdf] 12: X(InitOutput+0x206) [0x80a3886] 13: X(main+0x279) [0x806b479] 14: /lib/libc.so.6(__libc_start_main+0xe6) [0x43d5d6] 15: X(FontFileCompleteXLFD+0x221) [0x806aa61] Fatal server error: Caught signal 11. Server aborting Aborted Yeah, I've added this assert as there should not be any device/connector combo without a device id. There are two possible reasons why there isn't: 1. I still have a bug someplace (which is very conceivable) 2. AtomBIOS connector tables are broken. (which I also wouldn't rule out). Could you please provide a logverbose 7 log file (as usual ;) )? (In reply to comment #28) > Could you please provide a logverbose 7 log file (as usual ;) )? The xorg log from running the atom bios branch (when it crashed) is at the top of comment #27. Leif, could you please dump me your BIOS? You can use rhd_conntest 1:0.0 -d to do this. Thanks! Leif, could you please refetch the atombios_support branch? I've just committed a patch that should fix the crash you are seeing. Created attachment 18065 [details]
rhd_conntest 1:0.0 -d
rhd_conntest: v1.2.1, git branch master, commit 05f91165
Checking connectors on 0x9581, 0x103C, 0x30C5 (@01:00:00):
Load Detection: RHD_OUTPUT_TMDSA
HotPlug: RHD_HPD_0
DDC: RHD_DDC_1
Created attachment 18066 [details]
xorg log from latest atom bios branch
This is interesting,the atom bios branch causes my laptop panel to light up now on login to gnome desktop (yet it doesn't with radeonhd master from git)
However my external DVI monitor still doesn't light up when X starts (until I run xrandr commands to add custom modeline)
Also, the external DVI monitor is listed as disconnected by "xrandr -q" when it's not.
Created attachment 18067 [details]
"xrandr -q | xrandr -q --verbose" with atom bios git branch
Notice it says the DVI-D_1 is disconnected, but it's really not.
The log file from the atombios_support branch shows the last state of both the PANEL and DVI-I_1 active: RROut VGA_1 [Out DAC A Conn VGA 1] Crtc null [null] [inactive] disconnected RROut PANEL [Out LVDS Conn PANEL] Crtc CRTC 1 [CRTC 1] [active] connected RROut DVI-D_1 [Out TMDS A Conn DVI-D 1] Crtc CRTC 2 [CRTC 2] [active] connected while the log file in attachment #17976 [details] shows the DVI-I_1 output inactive: RROut VGA_1 [Out DAC A Conn VGA 1] Crtc null [null] [inactive] disconnected RROut PANEL [Out LVDS Conn PANEL] Crtc null [null] [inactive] connected RROut DVI-D_1 [Out TMDS A Conn DVI-D 1] Crtc CRTC 1 [CRTC 1] [active] connected RROut TV_7PIN_DIN [Out DAC B Conn TV 7PIN_DIN] Crtc null [null] [inactive] disconnected So the desktop helper that turned off your panel hasn't done this for some reason when you ran the atombios_support branch based driver. In the log file in attachment #17976 [details] I can see many RandR queries and indication that something disabled the PANEL after the Xserver had come up. None of this is present in the log in attachment #18067 [details]. According to the Xserver log the external DVI isn't disconnected. When you used the atombios_support branch you forgot to set the option: Option "UseAtomBIOS" "True" Thus you are using the same mode setting as the other driver, therefore I expect the same behavior regarding the external DVI. Maybe you can retest with the option set. I would like to see if this makes a difference for the native 1900x1200 display mode. Leif, did you have a chance to test this with the option set? Created attachment 18139 [details]
xorg log using latest atom bios branch (with Option "UseAtomBIOS" "True")
(In reply to comment #35) > When you used the atombios_support branch you forgot to set the option: > Option "UseAtomBIOS" "True" > Thus you are using the same mode setting as the other driver, therefore I > expect the same behavior regarding the external DVI. > Maybe you can retest with the option set. I would like to see if this makes a > difference for the native 1900x1200 display mode. > Sorry for the delay, long weekend in Canada :) I tried out the UseAtomBIOS option. When I run plain X from the command line I see both my laptop *and* external DVI light up! The resolutions are whacky, especially on the external DVI. See comment #37 for the xorg log. However when I log into my gnome desktop using atom bios...it's a good news and bad news thing. Good News: My external DVI monitor lights up with 1920x1200 resolution this time! Bad News: 1) My laptop panel doesn't light up anymore. I tried turning it on manually with xrandr and it says: xrandr --output PANEL --auto xrandr: Configure crtc 1 failed 2) The external DVI display has some horizontal flickering problems. It's not unusable, but annoying. I will attach the xorg log from logging into the desktop. It's quite different at quick glance. Not sure why again. Created attachment 18140 [details]
xorg log ( with atom bios branch ) logging into gnome
Comment on attachment 18140 [details]
xorg log ( with atom bios branch ) logging into gnome
Sorry I just realized this log was not run with -logverbose 7. So sort of useless.
Leif, could you please test with the latest GIT version using the option: Option "AtomBIOS" "output=on" and see if the native 1920x1200 mode is still working? Hi Egbert, yes i will, however it will have to wait until next week. Possibly Tuesday. Hello all, I have the same issue with my HD 2600 (MSI RX2600PRO-T2D256EZ/D2) using latest git driver. The screen is back, no response from the X, I am not able to shutdown them. I will attach the config and log. Created attachment 18637 [details]
xorg.conf
Created attachment 18638 [details]
Xorg.0.log
(In reply to comment #41) > Leif, could you please test with the latest GIT version using the option: > Option "AtomBIOS" "output=on" and see if the native 1920x1200 mode is still > working? > Running just plain X causes my laptop to light up and the 1920x1200 display gets a really weird resolution of 840x1050 (determined from the display OSD). As before, logging into my fedora 9 gnome desktop causes my laptop display to turn off and the external to light up with native 1920x1200 to light up by default. But the 1920x1200 screen has problems and goes black for a few seconds every few minutes. Created attachment 18640 [details]
(X -logverbose 7) with Option "AtomBIOS" "output=on"
(In reply to comment #46) > (In reply to comment #41) > > Leif, could you please test with the latest GIT version using the option: > > Option "AtomBIOS" "output=on" and see if the native 1920x1200 mode is still > > working? > > > > Running just plain X causes my laptop to light up and the 1920x1200 display > gets a really weird resolution of 840x1050 (determined from the display OSD). > The log shows a 1920x1200 mode. I wonder if this is the display detection the resolution wrong and scaling the output in a strange way. In comment #38 you explained that running a bare Xserver with AtomBIOS produced a wroking output on the external monitor. This was with Option "UseAtomBIOS" "yes" Now you say the size is strange with Option "AtomBIOS" "output=on" Is this one change responsible for the breakage? > As before, logging into my fedora 9 gnome desktop causes my laptop display to > turn off and the external to light up with native 1920x1200 to light up by > default. But the 1920x1200 screen has problems and goes black for a few seconds > every few minutes. > The former is a gnome-desktop/fedora problem (I think we talked about it). The latter one is due to the fact that the gnome desktop probes the monitor connection all the time. Leif, we need to find out how the behavior of comment #38 can be restored. Then it needs to be determined which subsystem is responsible for it working or failing. (In reply to comment #49) > Leif, we need to find out how the behavior of comment #38 can be restored. > Then it needs to be determined which subsystem is responsible for it working or > failing. > I wish I remembered what I meant in comment #38 by "The resolutions are whacky, especially on the external DVI". I see in the log of comment #37 that I was using this version of the driver: (II) RADEONHD: version 1.2.1, built from git branch atom, commit b4fe9f13 How do I get that from git to test again? Created attachment 18680 [details] (cell phone camera photo) bad scaling on the external 1920x1200 display (In reply to comment #46) > The log shows a 1920x1200 mode. > I wonder if this is the display detection the resolution wrong and scaling the > output in a strange way. I took this photo with a cellphone camera to show how badly the external DVI-D_1 display looks like. The laptop PANEL however (not pictured) looks good. As you noted the xorg log suggests DVI-D_1 is trying to display 1920x1200 but the monitor OSD says it's displaying 840x1050. (In reply to comment #51) > I took this photo with a cellphone camera to show how badly the external > DVI-D_1 display looks like. The laptop PANEL however (not pictured) looks good. > As you noted the xorg log suggests DVI-D_1 is trying to display 1920x1200 but > the monitor OSD says it's displaying 840x1050. > Yes, I'm not sure why this is the case. The display seems to think the mode is only 840x1050, it cuts of the edges and does a weired scaling job. I'm not sure why this is the case as I've never seen something like this myself. It could be a sync polarity problem. Have you tried randr to change the resolution on the external panel? This bug started out as something different but it seems it has turned into a duplicate of bug #14500. *** This bug has been marked as a duplicate of bug 14500 *** |
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.