Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 256874 - [bump] x11-drivers/ati-drivers-9.1 (8.573)
Summary: [bump] x11-drivers/ati-drivers-9.1 (8.573)
Status: RESOLVED FIXED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: Current packages (show other bugs)
Hardware: All Linux
: High enhancement (vote)
Assignee: Luca Barbato
URL: http://ati.amd.com/support/drivers/li...
Whiteboard:
Keywords:
: 257076 (view as bug list)
Depends on:
Blocks:
 
Reported: 2009-01-30 03:11 UTC by Peter Alfredsen (RETIRED)
Modified: 2009-03-27 19:54 UTC (History)
14 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments
ati-drivers-9.1.ebuild (ati-drivers-9.1.ebuild,14.20 KB, text/plain)
2009-01-30 23:34 UTC, Tomáš Chvátal (RETIRED)
Details
Cal libraries install and PCI_MSI check (ati-drivers-9.1-r1.ebuild,14.43 KB, text/plain)
2009-02-01 13:17 UTC, Nicola
Details
Ebuild to work on <xorg-server-1.5 (ati-drivers-9.1-forxorg_lt_1.5.diff,880 bytes, patch)
2009-02-01 20:23 UTC, Ville Aakko
Details | Diff
Output of emerge, while trying to install ati-drivers-9.1-r1 (ati-drivers-9.1-r1.ebuild.log,9.94 KB, text/plain)
2009-02-02 04:12 UTC, Navid Zamani
Details
ati-drivers-9.1.ebuild (ati-drivers-9.1.ebuild,14.62 KB, text/plain)
2009-02-02 20:25 UTC, Tomáš Chvátal (RETIRED)
Details
A working xorg.conf for the Radeon HD 4850. (xorg.conf.example-ati,3.98 KB, text/plain)
2009-02-02 20:46 UTC, Navid Zamani
Details
dmesg (dmesg.txt,923 bytes, text/plain)
2009-02-02 22:09 UTC, Marco Paolone
Details
xorg log (Xorg.0.log,798 bytes, text/plain)
2009-02-02 22:10 UTC, Marco Paolone
Details
Output of emerge, while trying to install ati-drivers-9.2, kernel = 2.6.29-rc7-zen1 (ati-9.2_fail_log,22.75 KB, text/plain)
2009-03-17 23:42 UTC, Tsang Ka Wing
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Peter Alfredsen (RETIRED) gentoo-dev 2009-01-30 03:11:49 UTC
I suggest you name them ati-drivers-9.1, so we once and for all can lay this crazy versioning scheme behind us.
Comment 1 George 2009-01-30 09:27:36 UTC
I wouldn't mind seeing this either. Since the files are stored on AMD's server that way, it would probably make more sense from an ebuild design standpoint too.
Comment 2 Balazs Nemeth 2009-01-30 09:35:49 UTC
Tada... here it is: http://ati.amd.com/support/drivers/linux/linux-radeon.html
9.1 released
Comment 3 Samuel Bustillos Cabrera 2009-01-30 16:12:17 UTC
(In reply to comment #0)
> I suggest you name them ati-drivers-9.1, so we once and for all can lay this
> crazy versioning scheme behind us.
> 

Yes, that naming should make things a bit more clear and nice.
Comment 4 Tomáš Chvátal (RETIRED) gentoo-dev 2009-01-30 23:34:26 UTC
Created attachment 180332 [details]
ati-drivers-9.1.ebuild

Just proof of concept for the renaming, nothing else changed. but i am lazy to run diff :]
Comment 5 Pierrot Rey 2009-01-31 07:06:46 UTC
(In reply to comment #3)
> (In reply to comment #0)
> > I suggest you name them ati-drivers-9.1, so we once and for all can lay this
> > crazy versioning scheme behind us.
> > 
> 
> Yes, that naming should make things a bit more clear and nice.
> 

yes it's good idea
Comment 6 Nicola 2009-01-31 12:39:39 UTC
Here it is the ebuild in the old fashion style, plus the fix for the AMD-Cal libraries.
The version you uploaded here, and the 8.561 in portage do not install them. (they are needed by the AMD Stream SDK, and other apps that make use of the GPU for GP-GPU )

http://bugs.gentoo.org/show_bug.cgi?id=257076
Comment 7 Tomáš Chvátal (RETIRED) gentoo-dev 2009-01-31 13:00:55 UTC
*** Bug 257076 has been marked as a duplicate of this bug. ***
Comment 8 Marco Schinkel 2009-02-01 01:46:29 UTC
(In reply to comment #4)
> Created an attachment (id=180332) [edit]
> ati-drivers-9.1.ebuild
> 
> Just proof of concept for the renaming, nothing else changed. but i am lazy to
> run diff :]
> 

Works for me! Thank you!
Comment 9 Pierrot Rey 2009-02-01 04:58:35 UTC
In amd64 does not work I have this error

[ 3012.444847] fglrx: Unknown symbol pci_disable_msi
[ 3012.444973] fglrx: Unknown symbol pci_enable_msi
Comment 10 kflirik 2009-02-01 08:21:55 UTC
(In reply to comment #9)
> In amd64 does not work I have this error
> 
> [ 3012.444847] fglrx: Unknown symbol pci_disable_msi
> [ 3012.444973] fglrx: Unknown symbol pci_enable_msi
> 

amd64, x1950pro (AGP)
I have that same error
Comment 11 Peter Alfredsen (RETIRED) gentoo-dev 2009-02-01 08:43:42 UTC
Per the phoronix thread, PCI_MSI needs to be enabled in the kernel this is compiled for.
Comment 12 kflirik 2009-02-01 09:29:33 UTC
(In reply to comment #11)
> Per the phoronix thread, PCI_MSI needs to be enabled in the kernel this is
> compiled for.
> 

Thank you. Works for me.
Comment 13 Marco Paolone 2009-02-01 09:33:16 UTC
Doesn't work for me too.. x86 and Ati Mobility Radeon 9700. Trying previous (unstable too) version got the same errors: modprobe tells there's a memory allocation error; Xorg this:
(WW) This ATI Proprietary Linux Driver does not guarantee support of video driver ABI higher than 2.0
(WW) Video driver ABI version of the X server is 4.1
(II) AMD ASIC control file status: R12 S1 B313 N313 T312
(EE) No devices detected.

Fatal server error:
no screens found

My card too old?
Comment 14 kflirik 2009-02-01 12:26:59 UTC
(In reply to comment #13)
> Doesn't work for me too.. x86 and Ati Mobility Radeon 9700. Trying previous
> (unstable too) version got the same errors: modprobe tells there's a memory
> allocation error; Xorg this:
> (WW) This ATI Proprietary Linux Driver does not guarantee support of video
> driver ABI higher than 2.0
> (WW) Video driver ABI version of the X server is 4.1
> (II) AMD ASIC control file status: R12 S1 B313 N313 T312
> (EE) No devices detected.
> 
> Fatal server error:
> no screens found
> 
> My card too old?

help me here it is:

eselect opengl set xorg-x11
emerge -C ati-drivers
emerge ati-drivers
dispatch-conf
eselect opengl set ati
Comment 15 Nicola 2009-02-01 13:17:57 UTC
Created attachment 180536 [details]
Cal libraries install and PCI_MSI check

ok sorry for the noise about the other thread, but nobody took care of the fact that these last 2 ebuilds don't install the AMD-Cal libraries.

So in this ebuild I added the check on the kernel config for PCI_MSI, and the portion that installs the Cal libraries.
Comment 16 Ville Aakko 2009-02-01 20:23:16 UTC
Created attachment 180606 [details, diff]
Ebuild to work on <xorg-server-1.5

Here is an ebuild diff that also works on other xorg-server than the newest one (well at least 1.3.0.0-r6, which is the latest stable and what I'm using). There is no reason it should not! Also see bug #247763.
Comment 17 Pierrot Rey 2009-02-01 20:44:08 UTC
(In reply to comment #13)
> Doesn't work for me too.. x86 and Ati Mobility Radeon 9700. Trying previous
> (unstable too) version got the same errors: modprobe tells there's a memory
> allocation error; Xorg this:
> (WW) This ATI Proprietary Linux Driver does not guarantee support of video
> driver ABI higher than 2.0
> (WW) Video driver ABI version of the X server is 4.1
> (II) AMD ASIC control file status: R12 S1 B313 N313 T312
> (EE) No devices detected.
> 
> Fatal server error:
> no screens found
> 
> My card too old?
> 
 No look

 	  Which ATI graphics cards can use this driver?
A2: 	

The ATI Proprietary Linux driver currently supports Radeon 8500 and later AGP or PCI Express graphics products, as well as ATI FireGL 8700 and later products. We do not currently plan to include support for any products earlier than this. Drivers for earlier products should already be available from the DRI Project or Utah-GLX project.
Comment 18 Marco Paolone 2009-02-01 20:51:52 UTC
Great, thanks! :) even if now I'm using radeon driver from xorg... works quite good. I'll try this ebuild anyway.
Comment 19 kflirik 2009-02-01 23:18:53 UTC
(In reply to comment #13)
> Doesn't work for me too.. x86 and Ati Mobility Radeon 9700. Trying previous
> (unstable too) version got the same errors: modprobe tells there's a memory
> allocation error; Xorg this:
> (WW) This ATI Proprietary Linux Driver does not guarantee support of video
> driver ABI higher than 2.0
> (WW) Video driver ABI version of the X server is 4.1
> (II) AMD ASIC control file status: R12 S1 B313 N313 T312
> (EE) No devices detected.
> 
> Fatal server error:
> no screens found
> 
> My card too old?

help me here it is:

eselect opengl set xorg-x11
emerge -C ati-drivers
emerge ati-drivers
dispatch-conf
eselect opengl set ati
Comment 20 Navid Zamani 2009-02-02 04:12:57 UTC
Created attachment 180643 [details]
Output of emerge, while trying to install ati-drivers-9.1-r1

I just tried to install “ati-drivers-9.1-r1.ebuild”, and this (mess) is what I got.

Translations:
• Fehler: »struct task_struct« hat kein Element namens »……………«
  Error: »struct task_struct« has no element called »……………«
• Fehler: Dereferenzierung eines Zeigers auf unvollständigen Typen
  Error: Dereferencing of a pointer to an incomplete type
• Warnung: return erzeugt Zeiger von Ganzzahl ohne Typkonvertierung
  Warning: return creates pointer from integer without type coversion
Comment 21 Navid Zamani 2009-02-02 04:16:13 UTC
Somehow creating an attachment does not make me a CC. Hence this change.
Comment 22 Navid Zamani 2009-02-02 04:58:51 UTC
(In reply to comment #20)

Ok, I just found out, that this only happens with the 2.6.29 kernel. With 2.6.28 it installs fine. It’s apparently because of symbols that were made private in 2.6.29.

I have no idea of what the reasons for this are, but I expect that the next version of the driver would work with 2.6.29 …and take a while until it gets released, because I read that this would mean a major rewrite.
Comment 23 George 2009-02-02 06:52:15 UTC
I'm having some major issues with this, I tried the 9.1 beta and 9.2 betas that were leaked and had the same issues... but even the final kills my Xorg, I can't even load Xorg with it. modprobe works and it compiles and emerges fine, but then when I go to start Xorg my screen goes black and never returns, can't even ctrl+alt+F1 back to my terminal. 

xorg.log has a bunch of entries saying "[mi] EQ overflowing. The server is probably stuck in an infinite loop." and "[mi] mieqEnequeue: out-of-order valuater event; dropping."

Is there something new with the 9.x drivers that the ebuild should check for?
Comment 24 Pierrot Rey 2009-02-02 07:18:07 UTC
And what there error in Xorg.0.log
Comment 25 George 2009-02-02 07:47:12 UTC
There is no error, it just keeps adding that until I force the PC off. Keyboard inputs are dead at this point but the power button still safely halts the system.
Comment 26 Jacek Trubłajewicz 2009-02-02 12:13:44 UTC
The same si(In reply to comment #25)
> There is no error, it just keeps adding that until I force the PC off. Keyboard
> inputs are dead at this point but the power button still safely halts the
> system.
> 

The same situation here, screen goes blank (black) with monitor still on, just after Xorg starts. Before that X also tries to change resolution a couple of times (I can hear monitor clicking). Keyboard seems to be still on because I can do crtl+alt+del, but when I try to switch to terminal there is no effect. No error in Xorg.log.

I solved this problem by running:
aticonfig --initial -f

I guess there was something messed up with my xorg.conf, so I suggest you try the same thing George.


9.1 seems to run stable now, tv-out is working, AIGLX performance is better now comparing to older drivers on my 9550 card, but scrolling in FF with compiz enabled is still slow. 
I've also noticed performance decrease in glxgears from something about 2000fps to ~500fps, I'm not sure about this but it seems there is fps decrease in Wolf Enemy Territory all well.

Comment 27 Balazs Nemeth 2009-02-02 17:34:15 UTC
There is an another bugreport in bugzilla. There is a missing dependency in the ebuild. Is there any way to include this dependency in this ebuild?

The dependency bug: #243172
Comment 28 Jacek Trubłajewicz 2009-02-02 18:36:52 UTC
I found that putting:

# === Video Overlay for the Xv extension ===

	Option	    "VideoOverlay" "on"

# === OpenGL Overlay ===

	Option	    "OpenGLOverlay" "off"

in xorg.conf, causes screen blanking problems mention by me earlier.


I also noticed that reverting form ati-drivers-9.1 to ati-drivers-8.522 breaks direct rendering due to missing: 
/usr/lib/xorg/modules/extensions/libdri.so. During 9.1 installation this file is overwritten by symbolic link to: /usr/lib/opengl/ati/extensions/libdri.so, which is also symbolic link to: /usr/lib/opengl/xorg-x11/extensions/libdri.so - during switching from 9.1 to 8.522 this file is deleted. 

Using eselect tool doesn't solve this problem, to bring back direct rendering I needed to emerge xorg-server again.  


Comment 29 Ville Aakko 2009-02-02 19:06:29 UTC
(In reply to comment #27)

I already commented bug #243172, but I'll say this here too to make sure the ebuild supportes notice. Maybe we should add this line to RDEPEND? :

x86? ( =virtual/libstdc++-3.3* )

Works for me.
Comment 30 Jeffrey Gardner (RETIRED) gentoo-dev 2009-02-02 19:34:13 UTC
/me jumps in to do some testing :)
Comment 31 Tomáš Chvátal (RETIRED) gentoo-dev 2009-02-02 20:25:52 UTC
Created attachment 180736 [details]
ati-drivers-9.1.ebuild

Updated ebuild with most of the mentioned fixes :]
I also ordered the deps so it looks nicer :P

For the error: this driver clearly supports only <=2.6.28 :]
Comment 32 Navid Zamani 2009-02-02 20:46:37 UTC
Created attachment 180737 [details]
A working xorg.conf for the Radeon HD 4850.

Hey, I solved the problems, and the driver is working very fine and stable here. No Compiz tough.

I did not have to use any tweaks. I only had to disable the composite extension. I have DRI enabled, sporting a very nice 3D and video acceleration.

I coded up a nice example xorg.conf from my configuration. With comments and information in it. I recommend, that we add this file to the ebuild as an example to put into /etc/X11/. That way people will have much less problems and can start running a stable desktop or decide what to tweak quickly.
Comment 33 Marco Paolone 2009-02-02 22:09:29 UTC
Created attachment 180751 [details]
dmesg
Comment 34 Marco Paolone 2009-02-02 22:10:09 UTC
Created attachment 180753 [details]
xorg log
Comment 35 Marco Paolone 2009-02-02 22:11:42 UTC
Sorry for the separated files :( anyway, module refuses to load properly again. Someone of you used a particular kernel configuration?
Comment 36 Pierrot Rey 2009-02-03 04:40:25 UTC
(In reply to comment #31)
> Created an attachment (id=180736) [edit]
> ati-drivers-9.1.ebuild
> 
> Updated ebuild with most of the mentioned fixes :]
> I also ordered the deps so it looks nicer :P
> 
> For the error: this driver clearly supports only <=2.6.28 :]
> 

In the new ebuild there is a dependency for libstc-v3 and it does not compile. I find it strange because before this dependency it was not there and everything was working perfectly. Accuracy I amd64

Comment 37 Jeffrey Gardner (RETIRED) gentoo-dev 2009-02-03 05:09:12 UTC
ati-drivers-8.573 is in the tree...
I've kept the driver version scheme,
I can't test with xorg-server-1.3,
I cannot test with libstdc++-3.3 either, so I omitted these features from the ebuild.
Thanks for the help guys!
Comment 38 Navid Zamani 2009-02-03 15:07:23 UTC
(In reply to comment #26 and comment #25)
> > There is no error, it just keeps adding that until I force the PC off.
> > Keyboard inputs are dead at this point but the power button still
> > safely halts the system.
> 
> Keyboard seems to be still on because I
> can do crtl+alt+del, but when I try to switch to terminal there is no effect.
> No error in Xorg.log.

Hey, I found out some important things.
First, when you get that black screen, the keyboard and power button only work for the first seconds. If you are quick, you can always change to the console.
BUT, you might still have a black screen, while being on the console. Press some keys (not sure if it’s relevant which), and you will see the console.

Then do not try to stop xdm, or you lose your control again. Also if you shut down, init will wait for xdm to terminate too, before starting to shut down the services. But at least your application data will be saved, if your app saves it before closing.

And I found a possibly related error in the logs, making it clear, why the system hangs after some seconds. The “⋮” is a vertical “…” aka “...”, and means that there is flooding of the lines above in the logs.
--------------------------------------------------------------------------------
[mi] EQ overflowing. The server is probably stuck in an infinite loop.
⋮
Backtrace:
0: /usr/bin/X(xorg_backtrace+0x26) [0x4e4c36]
1: /usr/bin/X(mieqEnqueue+0x289) [0x4c6259]
2: /usr/bin/X(xf86PostMotionEventP+0xc4) [0x472174]
3: /usr/bin/X(xf86PostMotionEvent+0xaf) [0x47233f]
4: /usr/lib64/xorg/modules/input//evdev_drv.so [0x7ff0e53fd01a]
5: /usr/bin/X [0x484b45]
6: /usr/bin/X [0x468fae]
7: /lib/libpthread.so.0 [0x3ff8c0ea00]
8: /usr/lib/libpixman-1.so.0 [0x3001c2851b]
9: /usr/lib64/xorg/modules//libfb.so(fbCopyNtoN+0x273) [0x7ff0f9fe0573]
10: /usr/lib64/xorg/modules//libfb.so(fbCopyRegion+0x200) [0x7ff0f9fdf4a0]
11: /usr/lib64/xorg/modules//libfb.so(fbDoCopy+0x443) [0x7ff0f9fdf9f3]
12: /usr/lib64/xorg/modules//libfb.so(fbCopyArea+0x4c) [0x7ff0f9fdfb8c]
13: /usr/lib64/xorg/modules//libxaa.so [0x7ff0f9d798da]
14: /usr/lib64/xorg/modules//libxaa.so [0x7ff0f9dc07e8]
15: /usr/bin/X [0x526d52]
16: /usr/bin/X [0x4f4e5d]
17: /usr/bin/X(compAllocPixmap+0x35) [0x4f50c5]
18: /usr/bin/X(compRealizeWindow+0x46) [0x4f48e6]
19: /usr/bin/X [0x430755]
20: /usr/bin/X(MapWindow+0x14e) [0x43266e]
21: /usr/bin/X(ProcMapWindow+0x46) [0x445dc6]
22: /usr/bin/X(Dispatch+0x364) [0x4465f4]
23: /usr/bin/X(main+0x45d) [0x42d48d]
24: /lib/libc.so.6(__libc_start_main+0xe6) [0x3ff801e5c6]
25: /usr/bin/X [0x42c879]

[mi] mieqEnequeue: out-of-order valuator event; dropping.
[mi] EQ overflowing. The server is probably stuck in an infinite loop.
⋮
--------------------------------------------------------------------------------

This error might not make it to your logs, if you shut down the wrong way. I don’t know...

It seems to be an event queue of some kind. And I also found out who is supposed to get the events:
/etc/init.d/atieventsd

I found a forum post, stating that adding this to the *boot* runlevel, should solve this problem.

I, for one, will not test it for now, because I rather work with the onboard nvidia and compiz enabled, that working without it, and crashing at least once a day. I work on too important files to risk this.
Comment 39 Navid Zamani 2009-02-03 15:08:35 UTC
We should really really add atieventsd to the boot runlevel on the installation of ati-drivers.
Comment 40 Navid Zamani 2009-02-03 15:13:59 UTC
(In reply to comment #32)
> Created an attachment (id=180737) [edit]
> A working xorg.conf for the Radeon HD 4850.

I noticed, that video playback still is a bit buggy with this one. I can’t move the player window, because the overlay does not move with it. This is with compiz disabled (of course). But apart from that, it works fine.

Of course, the driver still crashes in the middle of doing nothing, from time to time. (≈ once a day)
Comment 41 Marco Paolone 2009-02-03 15:51:56 UTC
Sorry guys, I've some questions for you. I would know which is your current version of gcc, kernel, xorg, and if you're on amd64 or x86.

Previous (and unstable too) versions of ati-drivers doesn't work for me. Currently I have a Mobility Radeon 9700 on x86.

Thanks in advance and sorry again!
Comment 42 Navid Zamani 2009-02-03 16:19:31 UTC
(In reply to comment #41)
> Sorry guys, I've some questions for you. I would know which is your current
> version of gcc, kernel, xorg, and if you're on amd64 or x86.

As you may know, this is no support forum. But I guess you knew that, so here it comes, straight from “emerge --info”:

Portage 2.1.6.7 (default/linux/amd64/2008.0/desktop, gcc-4.3.3, glibc-2.9_p20081201-r1, 2.6.28-zen7 x86_64)
System uname: Linux-2.6.28-zen7-x86_64-AMD_Athlon-tm-_64_X2_Dual_Core_Processor_4400+-with-glibc2.2.5
Comment 43 Navid Zamani 2009-02-03 16:23:08 UTC
Can someone with the rights to do it, re-open this bug, so it can be fixed *before* it gets to the portage tree, so *there is not even more non-working crap* in the tree, with no bugs attached, which then gets marked stable, without actually being so!

Or create bugs for it, and put a link in here.

Seriously guys, this sloppiness is why we get heat from the users, that the Portage tree gets less and less clean.
Comment 44 Navid Zamani 2009-02-03 16:25:38 UTC
(In reply to comment #42)
> (In reply to comment #41)

I forgot xorg: 7.4
Comment 45 Navid Zamani 2009-02-03 16:29:45 UTC
I created a ticket at ATi/“AMD customer care“, and told them, that the developer could join in here. I hope he does. :)
Together, we can solve the biggest problems quicker, cheaper, and for all users.
Comment 46 Jeffrey Gardner (RETIRED) gentoo-dev 2009-02-03 21:41:08 UTC
reopening
Comment 47 Keith Harrison 2009-02-03 22:08:16 UTC
Not sure if this helps but ati-drivers seems broken on HD3200 (780G) (I'm guessing it's ati's fault and not the ebuild)

8.561 doesn't work at all

8.573 locks up shortly after starting xorg if i do an "eselect opengl set ati"

If i do an "eselect opengl set xorg-x11" then it works but video playback with mythtv is extremely laggy.

dmesg shows nothing out of the ordinary

I tried my own personal Xorg.conf and the one posted by earlier by Navid Zamani

Xorg.0.log gives the following errors (which are not present with 8.552-r2):

(EE) fglrx(0): [pcie] Failed to gather memory of size 0Kb for PCIe. Error (-1)

...

(EE) fglrx(0): atiddxDriScreenInit failed, GPS not been initialized.
(WW) fglrx(0): ***********************************************
(WW) fglrx(0): * DRI initialization failed!                  *
(WW) fglrx(0): * (maybe driver kernel module missing or bad) *
(WW) fglrx(0): * 2D acceleraton available (MMIO)             *
(WW) fglrx(0): * no 3D acceleration available                *
(WW) fglrx(0): ********************************************* *

...

(EE) fglrx(0): XMM failed to open CMMQS connection.
(II) fglrx(0): XMM failed to initialize!

...

(II) AIGLX: Screen 0 is not DRI capable
[glesx] __glESXExtensionInit: No GL ES2.0 capable screen found!
(II) fglrx(0): Restoring recent mode: 1280x720@60Hz
(EE) fglrx(0): PPLIB: PPLIB is not initialized!.
(EE) fglrx(0): PPLIB: swlPPLibNotifyEventToPPLib() failed!
(EE) fglrx(0):        ulEventType = 0000000c, ulEventData = 00000001
Comment 48 Keith Harrison 2009-02-03 22:18:47 UTC
I also own a Mobility Radeon 9600 (or it could be a 9700) and the past 3 or 4 ati drivers have not worked for me on this card and it just givs a blank screen and Xorg.0.log ends by saying:

Fatal server error:
no screens found

Strangely enough the windows drivers stopped working about the same time.  I wonder if these cards are no longer being supported, they are still listed as supported on the release notes.  Thanks for all your hard work guys.
Comment 49 Peter Alfredsen (RETIRED) gentoo-dev 2009-02-03 22:22:58 UTC
Please open a new bug. This driver is in the tree, this bug is dead.
Comment 50 Navid Zamani 2009-02-03 23:27:28 UTC
(In reply to comment #49)
> Please open a new bug. This driver is in the tree, this bug is dead.

Then let us link so those bugs from here, because there certainly are more bugs than features in this driver.


(In reply to comment #47)
Have you tried my configuration? You absolutely *have* to disable the composite extension, and there are some other strange things too. Try my config, and open a new bug (or use an existing one that is more related to your problem). Good luck. If you still can, I recommend getting your money back by returning the card, and buying an nVidia one. :)
Comment 51 Navid Zamani 2009-02-03 23:33:43 UTC
(In reply to comment #48)
> I also own a Mobility Radeon 9600 (or it could be a 9700) and the past 3 or 4
> ati drivers have not worked for me on this card […]

The open source drivers for that card should be pretty good by now.
See here, for a nice chart of available features: http://xorg.freedesktop.org/wiki/RadeonFeature
In fact, the features for your card (RV350. The mobile 9600 and 9700 are the same chip) are nearly complete.
Comment 52 George 2009-02-04 01:45:12 UTC
Meh, still no luck. I've tried Jasek's xorg.conf (with changes for my input devices and PCI bus) and it just does the same thing. Could the new driver require different kernel settings than the older ones that I'm not using?
Comment 53 Ville Aakko 2009-02-04 17:27:35 UTC
I agree that this bug should be closed.  As someone already said, the driver is already in the tree. This is a binary blob, so it can't be really fixed by the community. A Suggestion: to iron out the bugs in the _ebuild_, maybe a stabilization request metabug could be made, and depend the bugs (that are fixable via the ebuild) there.

Also, everyone, just a friendly (shouting) reminder: AFAIK THIS IS NOT A SUPPORT FORUM!

1) This bug is only about getting 9.1 driver in the tree, 

2) any comments having nothing to do with the bug in question make it difficult to track the bug

3) It is impossible to track everyones individual problems with the driver in this bug! So posting here servers no purpose in that case.

4) Make separate bug reports (but remember this is a binary blob, so probably only problems in the ebuild make sense to be reported?)

5) If you just need help, really go to the forums, please! You'll get more readers there and the forums are a lot better tool for troubleshooting discussion with the community anyways.

If you think you have a better view of what a bugzilla is there for, please feel free to diasgree and comment =).
Comment 54 Navid Zamani 2009-02-05 07:14:04 UTC
(In reply to comment #53)
> Also, everyone, just a friendly (shouting) reminder: AFAIK THIS IS NOT A
> SUPPORT FORUM!

Either friendly, OR SHOUTING. You can’t have both. :)
Please behave youself.

> If you think you have a better view of what a bugzilla is there for, please
> feel free to diasgree and comment =).

I feel free, as I have exactly one point:
There should be a forum post for every bug. Linking forward and back between them. With clear statements what is for what.
So, who‘s the guy in charge? Because both is web development (in a cron job), and if it’s PHP, Python of Java, I can do it. I just need my script to get called, with a list of bugs to create forum threads and bug links for, and a place to put it.

Comment 55 Ville Aakko 2009-02-06 17:59:23 UTC
(In reply to comment #54)
> (In reply to comment #53)
> > Also, everyone, just a friendly (shouting) reminder: AFAIK THIS IS NOT A
> > SUPPORT FORUM!
> 
> Either friendly, OR SHOUTING. You can’t have both. :)
> Please behave youself.

Oops, you're right of course. I'll try to behave in the future =)

> There should be a forum post for every bug. Linking forward and back between
> them. With clear statements what is for what.
> So, who‘s the guy in charge? Because both is web development (in a cron job),
> and if it’s PHP, Python of Java, I can do it. I just need my script to get
> called, with a list of bugs to create forum threads and bug links for, and a
> place to put it.

Well, just from a general impression of things, I'd say that in 98% of the bugs a discussion thread in the forums is not really needed (since there wouldn't be enough general interest in the bug, the bug is so simple there is nothing to discuss etc... ). So creating a thread automatically for every bug might create too much overhead...  but a button to automate this when someone sees a thread useful, might be nice - maybe a link in the bugzilla to create a thread and link to it from the bug and vice versa?

This is getting terribly OT by the way. And I was te one talking about misusing bugzilla.... and sorry for shouting to everyone.
Comment 56 ProTech 2009-02-07 21:17:31 UTC
After installing the latest ebuild in portage 8.573-r1 with the xorg server 1.3 patch. I got the following error:

(EE) Failed to load module "glx" (module does not exist, 0)

I was able to solve this by linking /usr/lib64/opengl/xorg-x11/extensions/libglx.so to /usr/lib64/xorg/modules/extensions.

It working so far, but my guess is that this file should be linked from the opengl/ati directory not from the opengl/xorg-x11. But there is no such file in the ati directory. Can someone with more knowledge tell me where this file should come from?
Comment 57 Navid Zamani 2009-02-08 20:09:55 UTC
I agree that we close this one, as suggested, and add links to the bugs with this driver here. Or else everyone is going to post his problems here.

(In reply to comment #56)
Please ask in the forums. I think you forgot the “eselect opengl set ati“.
Comment 58 Tsang Ka Wing 2009-03-17 23:42:32 UTC
Created attachment 185359 [details]
Output of emerge, while trying to install ati-drivers-9.2, kernel = 2.6.29-rc7-zen1

I think this bug cannot be closed yet because 9.x still does not exist in portage.
I've also tested with the 9.2, that cannot build with 2.6.29-rc
Comment 59 Tsang Ka Wing 2009-03-17 23:47:53 UTC
(In reply to comment #58)
> Created an attachment (id=185359) [edit]
> Output of emerge, while trying to install ati-drivers-9.2, kernel =
> 2.6.29-rc7-zen1
> 
> I think this bug cannot be closed yet because 9.x still does not exist in
> portage.
> I've also tested with the 9.2, that cannot build with 2.6.29-rc
> 
er... I noted that 9.2 = 8.582 sorry for my misunderstanding.
Comment 60 Wilson M. Michaels 2009-03-27 18:31:01 UTC
After applying the suggestions I found in this Bug report I discovered a solution that may not be obvious. There is a file named /etc/ati/amdpcsdb (on Gentoo) that had some bad state information in it. It caused fglrx to crash soon after the Xorg.log file reported: 

fglrx(0): Restoring recent mode: 

I fixed my problem by copying "amdpcsdb.default" to "amdpcsdb" and rebooting. Resetting "amdpcsdb" to a minimal state allowed the fglrx driver to properly configure itself.
Comment 61 Navid Zamani 2009-03-27 19:54:04 UTC
I just want to note, that version 8.600 is out, and it’s much better than this buggy version. You still can forget Xinerama, and it still has problems with the  VirtualBox kernel modules, video playback still looks horrible, it still is not compatible with the most recent kernels, but at least Composite now works pretty well. Yay.

If you read this, consider this "bug" here dead and buried.