[RC410] detects AGP on a PCIE card

Please attach the Xorg.0.log that you get in Karmic without using any xorg.conf.

Here is the default Xorg.0.log file when started from a live CD

The trouble starts here:
(II) RADEON(0): AGP card detected

which of course causes failure later:
(WW) RADEON(0): [agp] AGP not available
(EE) RADEON(0): [agp] AGP failed to initialize. Disabling the DRI.

so desktop effects won't work in the default configuration.

That you have to specify "PCI" is a workaround for this faulty detection. This is the first thing that should be fixed. It is really a PCIE card.

However when you specify PCI it corrects it to PCIE. Running with PCIE causes black screen. This is a second issue.

Please also run "apport-collect 475466" so we get all information attached.

When should I run "apport-collect 475466"?
1. with no xorg.conf
2. or with the xorg.conf with that line added to it (if this then how to because I get a blank screen)

Also because you said that "PCIE" is correct option for me, I tried that instead of "PCI" in jaunty driver and I get a blank screen.

I have Radeon Xpress 200 which is onboard so how can it be PCIe?

In Windows(where driver is correct) with GPU-Z running I get my bus type as AGP 8x@PCI ( http://gpuz.techpowerup.com/09/11/07/gqn.png )

You can run apport-collect with the xorg.conf, it is basically to get hardware information.

I was thinking since the chipset is reported as "PCIE" it has some internal on-board PCIE interface, but I do not know much about this integrated graphics processors (IGP). The chipset has an PCIE interface to other units, but since the IGP is sitting inside the chipset I don't know how it is communicating. "AGP 8x@PCI" might indicate emulation of AGP?

But when I put the xorg.conf, I don't get any screen. So how should I run apport-collect?

 --
Absence of proof is not proof of absence
apparle

________________________________
From: Tormod Volden <email address hidden>
To: <email address hidden>
Sent: Sat, 7 November, 2009 2:20:33 PM
Subject: [Bug 475466] Re: [RC410] detects AGP on a PCIE card

You can run apport-collect with the xorg.conf, it is basically to get
hardware information.

I was thinking since the chipset is reported as "PCIE" it has some
internal on-board PCIE interface, but I do not know much about this
integrated graphics processors (IGP). The chipset has an PCIE interface
to other units, but since the IGP is sitting inside the chipset I don't
know how it is communicating. "AGP 8x@PCI" might indicate emulation of
AGP?

--
[RC410] detects AGP on a PCIE card
https://bugs.launchpad.net/bugs/475466
You received this bug notification because you are a direct subscriber
of the bug.

      The INTERNET now has a personality. YOURS! See your Yahoo! Homepage. http://in.yahoo.com/

Isn't there just a way to force the radeon driver to take the card as PCI and not PCIE because it works fine when forced to PCI in jaunty

Sorry I meant the other way, or I really meant that it does not matter. We just want the hardware information and that does not change whatever the xorg.conf you are using or not.

The right fix is that the driver detects the card for what it is and does the right thing with this information, without any workarounds.

The way to force the driver to take the card as PCI is the BusType option, but you already tried that.

I get this when I run 'apport-collect 475466'
Logging into Launchpad... You have to allow "Change anything" privileges.
Downloading bug information...
Error connecting to Launchpad: [Errno 111] Connection refused

I have proper settings to use the http proxy in terminal. (Maybe college firewall is blocking it)

Also I am attaching the logfile when I have added the line {Option "BusType" "PCI"} to the xorg.conf

As we can see in the log, I am trying to force the bus type of the card as PCI, but it is getting forced into PCI Express mode.

I don't mind workarounds, but why is the radeon driver forcing the bus as PCIE even when I have specified PCI explicitly

Actually the driver treats BusType option PCI and PCIE the same, then applies PCI or PCIE depending on which generation the card is. However there is some tweaking logic further up which treats PCIE IGP cards as PCI. This should probably have been the case for you. So if the initial detection would be PCIE (and not AGP) it would have ended up as PCI (which worked for you before).

This was changed in July, hence the breakage in Ubuntu 9.10: http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=57f2c83a22f27567506c555af431f89e6031204c

I forwarded the report upstream to https://bugs.freedesktop.org/show_bug.cgi?id=25002. If possible, please subscribe to the upstream bug report so that you can answer requests from the developers directly. Thanks in advance.

This has now been fixed upstream. I will mark the bug "fix committed" to signal that the fix will be included next time the -ati package is updated from upstream. You can test it out with packages from the https://launchpad.net/~xorg-edgers/+archive/drivers-only PPA as soon as they are updated later today. An official fix in Ubuntu 9.10 can be considered if this appears to concern many people.

Hello, thanks for this fix; I can't test it at the moment. Is the fix in future Ubuntu releases automatic included? Will my ati then work out of the box? Thank you!

The fix will be included in Ubuntu 10.04, but you have to verify that it actually works for you.

So should I update the package xserver-xorg-video-ati or the package xserver-xorg-video-radeon (I can't find this in the PPA) ?
And the bug was in -ati package or -radeon package?
Also will I need to use that BusType option or not?

You should install both, at the same time. In fact the driver is in the -radeon .deb, so only that is strictly needed, but keep both in sync to avoid any trouble. Both .deb packages come from the -ati source package, so look there.

No, you should not use any xorg.conf at all.

YESSSSSSSSSSSSSSSS!!!!!!!! :D :D :D :D :D

I installed the -ati and -radeon packages from the ppa and the compositing now runs fine without any xorg.conf
Ubuntu Community rocks.

Also will this problem be solved for other distributions like openSUSE or Fedora etc?

Good to hear. Yes, it is now solved for all Linux distros and *BSD and whatever uses Xorg, as soon as they update to the new upstream -ati source (will be version 6.12.5). For Ubuntu it will be 10.04 unless we do an SRU, for Fedora probably 12.

Is there a possibility to test it with a livecd (nightly built or something like that). I would like to verify it too, but I can't install ubuntu at the moment (next weeks). Do you know, if the fix is already included in the next linux mint (linux mint 8)?
Thank you for your work!

Pondo, yes you can boot a Ubuntu 9.10 live CD, and then install the packages from the PPA, just log out and in again for the new driver to be loaded.

Hello, I tried this yesterday. Then the system freezes after logging in again. I'll try it again!

Yes, great, it works; writing in Ubuntu Livemode with compiz. Thanks to you and your team. You're the best!!!

This bug was fixed in the package xserver-xorg-video-ati - 1:6.12.99+git20091125.0061c4db-0ubuntu1

---------------
xserver-xorg-video-ati (1:6.12.99+git20091125.0061c4db-0ubuntu1) lucid; urgency=low

  * New upstream git snapshot 20091125 (master) up to commit 0061c4db
    - Fix PCI chips that report as AGP to force PCI (LP: #475466)
 -- Bryce Harrington <email address hidden> Mon, 21 Dec 2009 14:14:55 -0800

Is it possible, that the fix isn't in ubuntu 10.04 Alpha2.

I tried this LiveCD, but i Couldn't start the desktop effects.

Pondo, I don't think so, but check the log file to see if it says "AGP card detected".

This bug does not seem to be fixed in 10.04 beta
I am running Kubuntu 10.04 Beta
I have xserver-xorg-video-ati (1:6.12.191-1ubuntu2) and xserver-xorg-video-radeon (1:6.12.191-1ubuntu2)

Still I got the "AGP card detected".
I have attached the Xorg.0.log

I just update the packages in the beta.
Now I have the version 1:6.12.192-2ubuntu2

I am still facing the same problem

I am attaching the dmesg and the Xorg.0.log?field.comment=I just update the packages in the beta.
Now I have the version 1:6.12.192-2ubuntu2

I am still facing the same problem

I am attaching the dmesg and the Xorg.0.log

And here is dmesg

Sorry, my mistake.
The driver is alright.
It detects the card and displays the message "AGP Card Detected". But after that it checks if it is a from RS400 family and corrects it to CARD_PCI, but does not display any message regarding this change.

The DRI was not starting due to the KMS.
By adding radeon.modeset=0 while booting, the DRI started up fine.

But I would suggest the developers to display a message "PCI Card forced for RS400 family" when the checking for RS400 is done.

Is there any disadvantage of setting radeon.modeset=0.

Due to this problem, the graphics are not working in Live CD. Anyway to fix this?

Apoorv, there is no big disadvantage in turning off KMS at this point. The boot splash screens will be different, and some combinations of 3D rendering and desktop effects will not work correctly. 3D speed might be better without KMS. However, most bug fixing is now done in the KMS path, so don't expect any non-KMS bugs to get fixed. Please file a new bug if you do not get DRI out of the box (and you have to turn off KMS).

Apporv, looking at your dmesg output, your issue is probably covered by bug 557611, which needs a kernel cherry-pick fix.

http://www.futuria.com.br/error/Image017.jpg/
http://www.futuria.com.br/error/Image018.jpg/
http://www.futuria.com.br/error/Image019.jpg/
http://www.futuria.com.br/error/Image020.jpg/
http://www.futuria.com.br/error/Image021.jpg/