On Nov 11, 07 15:37:14 +0300, Vladimir Grebenschikov wrote:

It starts, and then just exits without any error messages or cores even with the same config that works for previous version of driver.

Uuh. Shocking. Could you try again, this time with "-logverbose 7" and attach the logs? Maybe even building it with CFLAGS="-O0 -g3" and running it in gdb to check whether (and where) it cores? Removing the monitor sections shouldn't be necessary any more, as RandR ignores it anyway. CU Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Nov 12, 07 15:34:18 +0300, Vladimir Grebenschikov wrote:

Second attempt was much better (do not know what was changed from last time - I've just rebuilt driver).

Probably dependencies didn't get right, or something similar.

It shows screen with resolution of smaller panel (external LCD). And duplicating LVDS to external VGA.

I've tried to change screens duplication to two-screens layout but failed. (probably I just do something wrong).

# xrandr -q Screen 0: minimum 320 x 200, current 1280 x 1024, maximum 1400 x 1400 VGA_CRT1/DAC_A connected 1280x1024+0+0 376mm x 301mm 1280x1024 60.0*+ 75.0 59.9 1280x960 59.9 1152x864 75.0 74.8 1024x768 75.1 70.1 60.0 832x624 74.6 800x600 72.2 75.0 60.3 56.2 640x480 75.0 72.8 66.7 60.0 720x400 70.1 PANEL_LCD1/LVDS/TMDS unknown connection 1280x1024+0+0 305mm x 228mm 1400x1050 60.0 + 50.0 1280x1024 59.9* 1024x768 60.0 800x600 60.3 640x480 60.0 59.9 DVI-I_DFP1/TMDS_A disconnected

So this is the initial setup? Apparently RandR sets up for the lowest common denominator... This is something outside of our control, except for replacing the RandR algorithm, or RandR altogether. In order to do two-screens layout, you have to specify that in the monitor sections in xorg.conf (which are referenced by Option "Monitor-CONNECTOR" "<monitor>", e.g. for Monitor-PANEL_LCD1/LVDS/TMDS) by Option "LeftOf" "<other-monitor>". This is described in xorg.conf(5). Alternatively, you just increase the Virtual in the Screen section to the combined output resolution, (which sets the maximum screen size described above in the xrandr output), and set the second screen later on e.g. with xrandr --output PANEL_LCD1/LVDS/TMDS --right-of VGA_CRT1/DAC_A Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Nov 11, 07 13:57:11 +0100, Michael Gaber wrote:

well i tried the branch tody and was shocked: the display's totally garbled, i have funny green red and whatever colored vertical lines, it seems i have a top-left aligned 800x600 desktop which repeats...

The log indicates that it tries to run 1024x768 on your panel. What size is your panel actually? Hm - did you have to use a monitor section in order to get the panel running before? In that case you would have to add something like Option "Monitor-PANEL_LCD1/LVDS/TMDS" "mymonitor" to your Xorg.conf. RandR won't use your normal monitor section. Otherwise, please send me also a log with the old driver running - I'm not exactly sure how monitor detection works in an RandR environment, and it's using the general functions not our own (which work fine for my flat panel, but maybe not for you).

log is here: http://rafb.net/p/MNpYTg46.html

Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

ok curreent git works. but still: randr wont enable the second monitor if i started X without it connected. and if i connect the second one before starting X i get whiteouts if i try to restart X without the second monitor attached or switching to console. also i notice that there is no display-size 0x0mm in the randr output and no preferred mode of my lcd but also no lower-res modes Matthias Hopf schrieb:

On Nov 12, 07 11:43:58 +0100, Matthias Hopf wrote:

...

On Nov 11, 07 13:57:11 +0100, Michael Gaber wrote:

...

well i tried the branch tody and was shocked: the display's totally garbled, i have funny green red and whatever colored vertical lines, it seems i have a top-left aligned 800x600 desktop which repeats... The log indicates that it tries to run 1024x768 on your panel. What size is your panel actually?

I had some bits missing for panel detection. Current git should detect your panel correctly.

Matthias

Matthias Hopf schrieb:

On Nov 23, 07 23:08:41 +0100, Michael Gaber wrote:

...

ok curreent git works. but still: randr wont enable the second monitor if i started X without it connected.

There was a major bug lurking in the base Randr code. I have added a workaround in current git, which could result in the described behavior.

...

and if i connect the second one before starting X i get whiteouts if i try to restart X without the second monitor attached or switching to console.

This sounds weired. If it's framebuffer startadress based, Egbert has fixed that one in latest git as well.

seems like it wasn't, bug's still there but i noticed it's not simply fading to white but white with colored vertical lines

...

also i notice that there is no display-size 0x0mm in the randr output and no preferred mode of my lcd but also no lower-res modes

That was a side effect of the bug mentioned above :o]

still the same

CU

Matthias

Hi, On Nov 12, 2007 3:21 PM, Matthias Hopf wrote:

On Nov 12, 07 13:06:19 +0200, Felipe Contreras wrote:

...

On Nov 9, 2007 11:00 PM, Matthias Hopf wrote:

...

Heho,

I have just pushed initial RandR 1.2 support into the branch 'initial-randr' in the git repository. It's no way sure it will stay this way (might be completely rewritten), but at the moment everything looks good.

With the fixes to randr/rrcrtc.c and xrandr.c I just published on the xorg mailing list this implementation gets through the xrandr testcase I also published there today just fine.

Still, I would like to have it tested before we even think about moving this into the master branch, so if you want to try (assuming git 1.5 or higher, and *hopefully* correct...):

git-fetch origin initial-randr:initial-randr git-checkout origin/initial-randr ... configure & compile

If you want to work on the branch yourself, better do a git-checkout -b initial-randr origin/initial-randr

And then please tell me what is working what isn't.

I tried and I just get black screen, and I never get output again. I have to reboot.

Grmbl. Doesn't sound good. Please try:

- Set Option "NoRandr" temporary in the device section, just to verify that the driver + Xserver in *principle* works. This doesn't change anything to the driver in master.

I tried that, and it didn't work.

- Try to run Xorg -logverbose 7 and check whether you get some reasonable logfile. Post it. What card do you have?

I'm attaching the logs with dvi, vga, nothing connected and with the "NoRandr" option.

- Check whether you can still log into the machine remotely when it is dead. Check whether the Xserver is running (just no image) or not. If it is running, also post the output of 'xrandr'.

I can't try that right now but I think it is running. Best regards. -- Felipe Contreras

Hi, On Nov 13, 2007 8:00 PM, Matthias Hopf wrote:

Felipe, please read my remarks about your logs at the end of this mail first. It might save you some time and troubles :-]

On Nov 13, 07 19:37:31 +0200, Felipe Contreras wrote:

...

...

...

I tried and I just get black screen, and I never get output again. I have to reboot.

Grmbl. Doesn't sound good. Please try:

- Set Option "NoRandr" temporary in the device section, just to verify that the driver + Xserver in *principle* works. This doesn't change anything to the driver in master.

I tried that, and it didn't work.

In that case, please retry current git master. There should be no difference, except that master now has TMDSB support. If that doesn't work either, you're hitting a general modesetting bug, and Luc is probably the one you want to talk to.

I just tried 6f1800ca52531f5baf3df2e65a7bbd93e4b1e637 and that works perfectly, as it has been working since a lot of time.

...

...

- Try to run Xorg -logverbose 7 and check whether you get some reasonable logfile. Post it. What card do you have? I'm attaching the logs with dvi, vga, nothing connected and with the "NoRandr" option.

I'll look through that.

...

...

- Check whether you can still log into the machine remotely when it is dead. Check whether the Xserver is running (just no image) or not. If it is running, also post the output of 'xrandr'.

I can't try that right now but I think it is running.

Ok, you can try whether you get your console back by hitting Ctrl+Alt+Backspace.

That didn't work before...

Analyzing the log:

...

[...] (II) RADEONHD(0): FUNCTION: RHDConnectorsInit [...] (II) RADEONHD(0): FUNCTION: RHDHPDRestore

which is the last line of all logs. This indicates that the function RHDRandrPreInit() is just to be called. And the first thing this function does is printing out that it has been called. I don't read that, so it probably isn't.

This very much sounds like a compilation problem (dependencies?) to me. Can you do a 'make clean' and recompile?

Yeah, I did that before doing those tests since I had compilation issues (autogen and everything). I downloaded the latest code and tried again and now Ctrl+Alt+Backspace works, but I still have a black screen. I'm attaching the new log. It would be easier for me to stay updated if http access did work (http://lists.opensuse.org/radeonhd/2007-10/msg00370.html). Best regards. -- Felipe Contreras

Hi, On Nov 14, 2007 3:34 PM, Matthias Hopf wrote:

On Nov 14, 07 13:10:02 +0200, Felipe Contreras wrote:

...

Yeah, I did that before doing those tests since I had compilation issues (autogen and everything).

I downloaded the latest code and tried again and now Ctrl+Alt+Backspace works, but I still have a black screen.

I just pushed some missing bits for the noRandR case. Could you please try initial-randr again, with 'Option "noRandR"' set? The order of initialization in ScreenInit is still different, and I hope it doesn't make any difference...

I fetched dce9ffc402511d0820c0fcdd7228e1f2ce8571fc and now norandr works fine, and actually randr too. This is what I get: Screen 0: minimum 320 x 200, current 1400 x 1050, maximum 1600 x 1600 VGA_CRT1/DAC_A disconnected PANEL_LCD1/LVDS/TMDS unknown connection 1400x1050+0+0 287mm x 215mm 1400x1050 60.0*+ 50.0 1280x1024 59.9 1360x768 59.8 60.0 1280x800 60.0 1152x864 60.0 1280x768 60.0 1280x720 60.0 1024x768 60.0 800x600 60.3 640x480 60.0 59.9 DVI-I_DFP1/TMDS_A connected 1400x1050+0+0 408mm x 306mm 1600x1200 60.0 + 59.9 1400x1050 74.9* 60.0 1280x1024 75.0 59.9 1280x960 59.9 1152x864 75.0 74.8 1024x768 75.1 70.1 60.0 832x624 74.6 800x600 72.2 75.0 60.3 56.2 640x480 75.0 72.8 66.7 60.0 720x400 70.1 I tried: xrandr --output DVI-I_DFP1/TMDS_A -s 1600x1200 But nothing happens. At least there is improvement :) Best regards. -- Felipe Contreras -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

I was happy today to give my class using a Free Software driver. So to the extent that I was able to do it, the RandR is working. But the initial-randr branch seems to have more trouble switching consoles, and (more problematically) "xrandr --auto" after adding/removing a monitor causes the screens to go black. Is that a known limitation, or should I investigate some more? Stefan -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Nov 20, 07 12:41:28 +0100, Matthias Hopf wrote:

On Nov 19, 07 21:33:28 -0500, Stefan Monnier wrote:

...

I was happy today to give my class using a Free Software driver. So to the extent that I was able to do it, the RandR is working. But the initial-randr branch seems to have more trouble switching consoles, and (more problematically) "xrandr --auto" after adding/removing a monitor causes the screens to go black. Is that a known limitation, or should I investigate some more?

Which version are you using? The first few releases had a few issues, most of them should be gone now. If not, it would be great if you could dig into that.

I also just noticed that it might be the xrandr changes I pushed upstream a few days ago. This program still had several bugs lurking in the code (and still has), and I sincerely hope I didn't make it worse :-P Also the base RandR code in the server had a serious and several not so serious bugs. Also fixed upstream. CU Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Nov 14, 07 13:10:02 +0200, Felipe Contreras wrote:

...

This very much sounds like a compilation problem (dependencies?) to me. Can you do a 'make clean' and recompile?

I downloaded the latest code and tried again and now Ctrl+Alt+Backspace works, but I still have a black screen.

Can you check your version? If you have 7dc989a5dfc3 or fe834af6713 get the latest bits again (9031b8706), and try again, please. CU Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Nov 12, 07 21:13:21 -0500, GerbilSoft wrote:

I'm using a ThinkPad T60p (FireGL V5200, 0x71C4), and so far RandR 1.2 support seems to be working well. I've noticed a few glitches though.

Video Modes: VGA_CRT1/DAC_A: 1600x1200 @ 75 Hz PANEL_LCD1/LVDS/TMDS: 1600x1200 @ 60 Hz

- The VGA output seems to have some issues. The right side of the screen isn't fully displayed (it's around 32 pixels off). Also, there's some ~8px-wide columns on the screen that appear to be flickering. These columns don't appear in <1600x1200, though.

Does this also happen if you use the driver in the non-RandR case (Option "NoRandr")? Check that you're using exactly the same mode, with only the VGA active (will probably choose a 60Hz mode otherwise).

- I can't seem to get it so that the VGA monitor is at e.g. coordinates -1600x0, so it doesn't affect the contents of the LCD. My CRT is on the left side of my desk where the laptop is on the right, so it would be useful if I could somehow get the CRT to simply use negative coordinates instead of having everything move over to the CRT when switching to multiple monitor mode.

There are no negative coordinates. You can, however, move your panel to the right (xrandr --output PANEL_LCD1/LVDS/TMDS --right-of VGA_CRT1/DAC_A). Theoretically, --left-of should produce the same result, practically there are some bugs lurking in RandR.

[Actually, somehow I managed to get the LCD indicated as monitor 1 even though its coordinates are not 0,0, but I'm not sure how I managed this!]

Me don't either. RandR has no notion about monitor numbers.

- LVDS doesn't work at resolutions other than native, but I believe that's a known bug.

Maybe Luc can comment on that. Or your LVDS doesn't support anything else than the native resolution, and scaler support isn't implemented yet. Thanks for testing Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Nov 13, 07 18:00:50 -0500, GerbilSoft wrote:

...

...

- The VGA output seems to have some issues. The right side of the screen isn't fully displayed (it's around 32 pixels off). Also, there's some ~8px-wide columns on the screen that appear to be flickering. These columns don't appear in <1600x1200, though.

Does this also happen if you use the driver in the non-RandR case (Option "NoRandr")? Check that you're using exactly the same mode, with only the VGA active (will probably choose a 60Hz mode otherwise).

With "NoRandr", I get no video on either monitor. I've attached the X log as xorg.norandr.log with NoRandr selected.

You're hitting the same issue as Tigerchen on irc. I'm already debugging this, I apparently got the initialization order wrong. Please recheck when I'm done (I'll announce this).

Also, with RandR enabled, I can't seem to enable the second monitor unless it was plugged in when X was started. If I try, I get this error:

david@gs_laptop.gerbilsoft.gsft ~ $ xrandr --output VGA_CRT1/DAC_A --auto X Error of failed request: BadMatch (invalid parameter attributes) Major opcode of failed request: 155 (RANDR) Minor opcode of failed request: 21 () Serial number of failed request: 17 Current serial number in output stream: 17

Bugs in xrandr. It isn't able to unclone crtcs. I pushed some fixes upstream, and also one fix for RandR base system. CU Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Tuesday 13 November 2007 10:41:18 am you wrote:

...

- I can't seem to get it so that the VGA monitor is at e.g. coordinates -1600x0, so it doesn't affect the contents of the LCD. My CRT is on the left side of my desk where the laptop is on the right, so it would be useful if I could somehow get the CRT to simply use negative coordinates instead of having everything move over to the CRT when switching to multiple monitor mode.

There are no negative coordinates. You can, however, move your panel to the right (xrandr --output PANEL_LCD1/LVDS/TMDS --right-of VGA_CRT1/DAC_A). Theoretically, --left-of should produce the same result, practically there are some bugs lurking in RandR.

It seems that this is also an issue with the Intel RandR 1.2 driver. There's a thread about it on the X.org mailing list[1], and it seems that it has to do with the fact that LVDS is on CRTC 1 while VGA is on CRTC 0. The old non-RandR Radeon driver had an option "MergedXineramaCRT2IsScreen0" that allowed the Xinerama screen IDs to be flipped, but I'm not sure if this is applicable to RandR 1.2. [1] http://lists.freedesktop.org/archives/xorg/2007-August/027665.html

On Nov 15, 07 03:14:02 +0100, Sebastian Brocks wrote:

...

And then please tell me what is working what isn't.

I just did installed it, and it only connect my primary display, not the TFT attached via DVI on the secondary connector.

I think I understood you ;) Both DVI? I didn't merge the latest bits from master in yet, which add TMDS-B support. CU Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org