Bug 15708 (ati+xrandr@IGP350M)

Created attachment 16176 [details]
My xorg.conf File

Unfortunately, we never got dualhead working on certain older IGP chips.  I think there is some GPIO magic involved in setting it up, but it hasn't been sorted out yet.  You might try booting with the VGA attached so that the bios comes up on the VGA screen.

Alex,

Thanks for the reply.
I tried to boot the laptop with external monitor attached and turned on.
The booting process is shown on both monitors. I then enter the BIOs of the 
laptop. The BIOs is shown on both monitors too. only when I started X, the 
external monitor is off even xrandr shows it is connected and on.

More stranger, if i turn it off by 'xrandr --output VGA-0 --off', the LVDS get 
resized. my VGA-0 is supposed to run in 1280x1024 and LVDS 1024x768.
Hope that gives you more information for judgement.

And can you briefly explain what is the GPIO magic? Can I do anything more? or just give up? I actually had put a lot of my time on it and it drove me ;(.
 

Charlie  

(In reply to comment #3)
> Alex,
> 
> Thanks for the reply.
> I tried to boot the laptop with external monitor attached and turned on.
> The booting process is shown on both monitors. I then enter the BIOs of the 
> laptop. The BIOs is shown on both monitors too. only when I started X, the 
> external monitor is off even xrandr shows it is connected and on.
> 
> More stranger, if i turn it off by 'xrandr --output VGA-0 --off', the LVDS get 
> resized. my VGA-0 is supposed to run in 1280x1024 and LVDS 1024x768.
> Hope that gives you more information for judgement.
> 
> And can you briefly explain what is the GPIO magic? Can I do anything more? or
> just give up? I actually had put a lot of my time on it and it drove me ;(.
> 
> 
> Charlie  
> 

just a little clearifying,

I mean when I started X, the VGA-0 became black even though xrandr says it is on.

(In reply to comment #3)
> More stranger, if i turn it off by 'xrandr --output VGA-0 --off', the LVDS get 
> resized. my VGA-0 is supposed to run in 1280x1024 and LVDS 1024x768.
> Hope that gives you more information for judgement.
> 

right.  the driver thinks the output is working.  we are just missing some bit in the dac init sequence to make the dac actually produce an output.

> And can you briefly explain what is the GPIO magic? Can I do anything more? or
> just give up? I actually had put a lot of my time on it and it drove me ;(.

I think you may have to toggle some gpio pins to switch the dac output between the S-video port and the VGA port, but as you are still having problems even with VGA connected at boot, I suspect the problem may be elsewhere.  I'll have to check the register specs and see if there are any strange quirks with certain old IGP chips.

(In reply to comment #5)
> (In reply to comment #3)
> > More stranger, if i turn it off by 'xrandr --output VGA-0 --off', the LVDS get 
> > resized. my VGA-0 is supposed to run in 1280x1024 and LVDS 1024x768.
> > Hope that gives you more information for judgement.
> > 
> 
> right.  the driver thinks the output is working.  we are just missing some bit
> in the dac init sequence to make the dac actually produce an output.
> 
> > And can you briefly explain what is the GPIO magic? Can I do anything more? or
> > just give up? I actually had put a lot of my time on it and it drove me ;(.
> 
> I think you may have to toggle some gpio pins to switch the dac output between
> the S-video port and the VGA port, but as you are still having problems even
> with VGA connected at boot, I suspect the problem may be elsewhere.  I'll have
> to check the register specs and see if there are any strange quirks with
> certain old IGP chips.
> 

I really expect that you can figure things out, if it is not too painful.
If I can be of any help to you just let me know. And if you come up with any patch or hack, keep me informed and I will test it ASAP.

Thanks 

This is for Charlie Xia the originator of this Bug.

I have the same card and experienced same problem on Ubuntu Hardy Heron. Moreover the LCD screen of my Laptop is broken and I am using the Laptop as a Desktop through external VGA monitor and desperately trying to make it work.

I configured through xorg.conf and used vesa driver. Now the display works on both LCD (I can see through some patches working) and the VGA monitor. The output of xrandr -q is as follows:
~$ xrandr -q
Screen 0: minimum 640 x 480, current 1024 x 768, maximum 1024 x 768
default connected 1024x768+0+0 0mm x 0mm
   1024x768       61.0* 
   800x600        73.0  
   640x480        86.0  

The Device, Monitor and Screen sections of xorg.conf are as follows:

Section "Device"
	Identifier	"ATI Technologies Inc Radeon IGP 330M/340M/350M"
	Driver		"vesa"
	BusID		"PCI:1:5:0"
	Option "Monitor-VGA-0" "VGA monitor"
EndSection

Section "Monitor"
	Identifier	"VGA monitor"
EndSection

Section "Screen"
	Identifier	"Default Screen"
	Device		"ATI Technologies Inc Radeon IGP 330M/340M/350M"
	Monitor		"VGA monitor"
	DefaultDepth	24
	SubSection "Display"
	Modes		"1024x768"
	EndSubSection
EndSection

Please note that I had no problem in Ubuntu Gutsy Gibbon, therefore, you can consider to downgrade the xserver in Debian and the ati driver works.

Hope it helps.

*** Bug 15888 has been marked as a duplicate of this bug. ***

Hi, I have an HP nc4010 (nearly the same as nc4000) also with a radeon IGP 350M and running Debian lenny (i386).
After upgrading to xserver-xorg_7.3.0+10, xserver-xorg-core_1.4.1~git20080131-4 and xserver-xorg-video-ati_6.8.0-1, I got the same issue that the external monitor did not work.

Now, I downgraded to xserver-xorg_7.1.0-19, xserver-xorg-core_1.3.0.0.dfsg-12lenny2 (still online) and xserver-xorg-video-ati_6.7.197-1 (I built it myself from the online available debian sources because I didn't find this version online and with version 6.6.3-2 I had some problems).
After this downgrade it works again and I can also specify normal xrandr 1.2 commands and they succeed!

If I can help you with something, please tell me.

I can confirm this also with an HP ZE5500 series laptop (ATI Radeon IGP 340M).

I am wondering if I should roll back to a previous version of X when it worked. I am willing to suffer along for a time using vesa as a work around, but if it is going to be a great deal of time before this bug is looked at, then I will roll back. 

Maybe an even better question would be if this bug is going to get any time at all. I know these cards are on the older end of the scale, so developers may not have any desire/time to give. 

It would be greatly appreciated if someone "in the know" would tell me whether this bug is going to actually get attention or if it is just going to fade into the sunset.

(In reply to comment #9)
> Hi, I have an HP nc4010 (nearly the same as nc4000) also with a radeon IGP 350M
> and running Debian lenny (i386).
> After upgrading to xserver-xorg_7.3.0+10, xserver-xorg-core_1.4.1~git20080131-4
> and xserver-xorg-video-ati_6.8.0-1, I got the same issue that the external
> monitor did not work.
> 
> Now, I downgraded to xserver-xorg_7.1.0-19,
> xserver-xorg-core_1.3.0.0.dfsg-12lenny2 (still online) and
> xserver-xorg-video-ati_6.7.197-1 (I built it myself from the online available
> debian sources because I didn't find this version online and with version
> 6.6.3-2 I had some problems).
> After this downgrade it works again and I can also specify normal xrandr 1.2
> commands and they succeed!

If 6.7.197 worked and 6.8.0 didn't, it would be great if you could use git to bisect down which commit caused it to break.

I would be happy to use "git" to help troubleshoot, but I don't know what to do.

Is there a tutorial somewhere on how to use git to isolate a commit?

(In reply to comment #12)
> I would be happy to use "git" to help troubleshoot, but I don't know what to
> do.
> 
> Is there a tutorial somewhere on how to use git to isolate a commit?
> 

Building from git:
1. install build dependencies (xorg-dev packages)
2. checkout the ati git tree
git clone git://anongit.freedesktop.org/git/xorg/driver/xf86-video-ati
3. cd xf86-video-ati
4. ./autogen.sh --prefix=/usr
5. make
6. make install (as root)

Bisecting with git:
1. git bisect start
2. git bisect bad
3. tell git 6.7.197 is good:
git bisect good eb99c3c5c9a2249cb84920f0f225e525fc3a4144
4. build and install (see above)
5. if it's broken:
git bisect bad
if it's working:
git bisect good
6. repeats steps 4 and 5 until you've found the bad commit.

"build and install" - does that mean repeat steps 4-6 from the git build ?

"if it works" - does that mean I need to restart X after each build?

I am a bit of n00b, but I want to get this info to you. Thanks for your patience.

(In reply to comment #14)
> "build and install" - does that mean repeat steps 4-6 from the git build ?
> 

yes.

> "if it works" - does that mean I need to restart X after each build?
> 

yes.  restart X and and try to enable the VGA port.

Ok, according to my Xorg.0.log I was using the 6.7.197 driver and I still couldn't get my VGA-0 to display. 

xrandr knew it was there. The LVDS blinked when I tried to switch to the VGA-0 as if it was resizing properly, but the monitor never came on. 

Should I start earlier in git to help narrow it down?

(In reply to comment #16)
> 
> Should I start earlier in git to help narrow it down?
> 

The first thing to do is to find a known good version. Once you have that, you can use bisect to find the commit that broke it.  However, a lot of old IGP chips never worked properly with multiple heads enabled so YMMV.

This is strange. What versions of other xorg packages do you use?

For me, the following combination works:

ii  xserver-xorg                         1:7.1.0-19
ii  xserver-xorg-core                    2:1.3.0.0.dfsg-12lenny2
ii  xserver-xorg-dev                     2:1.3.0.0.dfsg-12lenny2 
ii  xserver-xorg-input-evdev             1:1.1.2-6
ii  xserver-xorg-input-kbd               1:1.1.0-4
ii  xserver-xorg-input-mouse             1:1.1.1-3
ii  xserver-xorg-input-synaptics         0.14.6-1
ii  xserver-xorg-video-ati               1:6.7.197-1
ii  xserver-xorg-video-vesa              1:1.3.0-1

These are all older versions (dunno whether they are still in debian archive, if you need them, I can send them to you).

(In reply to comment #18)
> ii  xserver-xorg-video-ati               1:6.7.197-1

If you could use git to bisect the problem as per comment 13 it would be very helpful.

Created attachment 17220 [details] [review]
Fixes this problem, but it would probably break other radeon cards

Thanks for motivating me to find a solution and providing me with a howto to use the bisect tool, really nice tool btw. :)

Wow, took me quite some time tracking the problem down (although using git bisect). First, bisect took me to another revision: 30cab1dbebb7bdb925f2fe9f91b84183312bbbfd
Applying the reverse patch did not fix it for me in HEAD revision.

Thus, tracking down the bug a second time. This time using the reverse patch from this revision.
Then bisect took me to 000741e250e54122b0adc91694eb4bfa320a70fb which is a huge diff with much radeon clean up. One line caught my eye: "IsIGP". And with the appended little patch, my external monitor works straight ahead.

Probably removing these two lines is not optimal for all users (IMHO). Thus it would be great if you could find a workaround that does not break it for all users. I will test again ;)

(In reply to comment #20)
> Created an attachment (id=17220) [details]
> Fixes this problem, but it would probably break other radeon cards

Does VGA-0 work if you boot without a monitor attached and attempt to turn on VGA-0 later using xrandr?

(In reply to comment #21)
> (In reply to comment #20)
> > Created an attachment (id=17220) [details] [details]
> > Fixes this problem, but it would probably break other radeon cards
> 
> Does VGA-0 work if you boot without a monitor attached and attempt to turn on
> VGA-0 later using xrandr?
> 

Yes, I just tested it and it works.

(In reply to comment #17)
> (In reply to comment #16)
> > 
> > Should I start earlier in git to help narrow it down?
> > 
> 
> The first thing to do is to find a known good version. Once you have that, you
> can use bisect to find the commit that broke it.  However, a lot of old IGP
> chips never worked properly with multiple heads enabled so YMMV.
> 

Ok, I just got it working with an older live CD (Ubuntu - Gutsy Gibbon). When I looked at the Xorg.0.log with the live CD it showed the ati driver with version 6.7.195.

Can I start the git bisect there? 



P.S. Sorry to ask, but what is YMMV?

(In reply to comment #23)

> Ok, I just got it working with an older live CD (Ubuntu - Gutsy Gibbon). When I
> looked at the Xorg.0.log with the live CD it showed the ati driver with version
> 6.7.195.
> 
> Can I start the git bisect there? 
> 

Sounds like it's the same issue as Fabian.  I would suggest trying his patch to start with.  I'm trying to get this sorted out internally.  Stay tuned.

> 
> 
> P.S. Sorry to ask, but what is YMMV?
> 

Your Mileage May Vary.

(In reply to comment #24)
> (In reply to comment #23)
> 
> > Ok, I just got it working with an older live CD (Ubuntu - Gutsy Gibbon). When I
> > looked at the Xorg.0.log with the live CD it showed the ati driver with version
> > 6.7.195.
> > 
> > Can I start the git bisect there? 
> > 
> 
> Sounds like it's the same issue as Fabian.  I would suggest trying his patch to
> start with.  I'm trying to get this sorted out internally.  Stay tuned.
> 


OK, I can confirm Fabian's patch works for my hardware. Thank you Alex and Fabian for hanging with me. 

I've attempted to fix this in git:
e78e8a21b4040cd7f1983c241c860d9209398396

Let me know if it fixes things for you.

Latest git works for me.

Thanks Alex.

(In reply to comment #26)
> I've attempted to fix this in git:
> e78e8a21b4040cd7f1983c241c860d9209398396
> 
> Let me know if it fixes things for you.

Works for me, too. Thank you very much!

The updated ati driver is available in Debin Sid. I have downloaded the required drivers from Debian Sid repositories and they are working on Hardy:
xserver-xorg-video-ati_6.8.192-1_i386.deb
xserver-xorg-video-mach64_6.8.0-1_i386.deb
xserver-xorg-video-r128_6.8.0-1_i386
xserver-xorg-video-radeon_6.8.192-1_i386

I have deleted xorg.conf file, xrandrr -q is showing VGA-0 and LVDS outputs and I could turn off LVDS.

I used "git bisect" to find this. Since changeset e78e8a21b4040cd7f1983c241c860d9209398396 my VGA flickers a lot and is grainy/fuzzy. I can't see anything on that monitor.

If I use latest git-code and revert this changeset, everything will work fine - I never had problems with VGA (and xrandr) before that. It's only this change that cause my problems.

lspci -v
01:05.0 VGA compatible controller: ATI Technologies Inc RS485 [Radeon Xpress 1100 IGP] (prog-if 00 [VGA controller])
	Subsystem: Hewlett-Packard Company Unknown device 30b0
	Flags: bus master, fast Back2Back, 66MHz, medium devsel, latency 64, IRQ 17
	Memory at c0000000 (32-bit, prefetchable) [size=128M]
	I/O ports at 6000 [size=256]
	Memory at d4300000 (32-bit, non-prefetchable) [size=64K]
	[virtual] Expansion ROM at d4320000 [disabled] [size=128K]
	Capabilities: [50] Power Management version 2

Notebook HP nx6325

This should be fixed now:
cfaa23d925e3c062cf87ea844566ac11ea02d69e

Thanks, that fixed the problem. Really fast. :-)