how to get XV, DRI working on HD 3300 (rv620) ??

[schwarzygesetzlos]

The last 2 days I unsuccessfully tried to get XV working on my ASRock AOD790GX. Graphics card is an onboard Radeon HD3300, which seems to be the same rv620 chipset as the Radeon HD3450.

After doing research I ended up using the r6xx-r7xx-branch from the radeonhd git repository. I followed the instructions on http://xorg.freedesktop.org/wiki/radeonhd:r6xx_r7xx_branch but still no success...

One problem seems that the drm-module doesn't recognize my Radeon HD 3300. I used the r6xx_r7xx_branch from the drm git too. Or is it possible to run XV without relying on DRM?

The drm-failure from my Xorg.0.log:
(WW) RADEONHD(0): Direct rendering for R600 and up forced on - This is NOT officially supported yet and may cause instability or lockups
(II) RADEONHD(0): Found libdri 5.4.0.
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenByBusid: Searching for BusID pci:0000:01:05.0
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenByBusid: drmOpenMinor returns -19
drmOpenDevice: node name is /dev/dri/card1
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed

[...]

drmOpenByBusid: drmOpenMinor returns -19
drmOpenDevice: node name is /dev/dri/card14
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenByBusid: drmOpenMinor returns -19
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: node name is /dev/dri/card1
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed

[...]

drmOpenDevice: node name is /dev/dri/card14
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
(EE) RADEONHD(0): RHDDRIVersionCheck: drmOpen("radeon", "pci:0000:01:05.0") failed.
(WW) RADEONHD(0): RHDDRIPreInit: Version check failed. Disabling DRI.

The more successful part oy my Xorg.0.log:
(II) Loading /usr/lib64/xorg/modules//libexa.so
(II) Module exa: vendor="X.Org Foundation"
compiled for 1.5.2, module version = 2.4.0
ABI class: X.Org Video Driver, version 4.1
(II) RADEONHD(0): FB: Allocated Offscreen Buffer at offset 0x006F4000 (size = 0x00CCC000)
(--) Depth 24 pixmap format is 32 bpp
(II) do I need RAC? No, I don't.
(II) resource ranges after preInit:
[0] -1 0 0xffffffff - 0xffffffff (0x1) MX[B]
[1] -1 0 0x000f0000 - 0x000fffff (0x10000) MX[B]
[2] -1 0 0x000c0000 - 0x000effff (0x30000) MX[B]
[3] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX[B]
[4] -1 0 0xffffffff - 0xffffffff (0x1) MX[B]
[5] -1 0 0x000f0000 - 0x000fffff (0x10000) MX[B]
[6] -1 0 0x000c0000 - 0x000effff (0x30000) MX[B]
[7] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX[B]
[8] -1 0 0xffffffff - 0xffffffff (0x1) MX[B]
[9] -1 0 0x000f0000 - 0x000fffff (0x10000) MX[B]
[10] -1 0 0x000c0000 - 0x000effff (0x30000) MX[B]
[11] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX[B]
[12] -1 0 0xffffffff - 0xffffffff (0x1) MX[B]
[13] -1 0 0x000f0000 - 0x000fffff (0x10000) MX[B]
[14] -1 0 0x000c0000 - 0x000effff (0x30000) MX[B]
[15] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX[B]
[16] -1 0 0xffffffff - 0xffffffff (0x1) MX[B]
[17] -1 0 0x000f0000 - 0x000fffff (0x10000) MX[B]
[18] -1 0 0x000c0000 - 0x000effff (0x30000) MX[B]
[19] -1 0 0x00000000 - 0x0009ffff (0xa0000) MX[B]
[20] 0 0 0x000a0000 - 0x000affff (0x10000) MS[B]
[21] 0 0 0x000b0000 - 0x000b7fff (0x8000) MS[B]
[22] 0 0 0x000b8000 - 0x000bffff (0x8000) MS[B]
[23] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
[24] -1 0 0x00000000 - 0x00000000 (0x1) IX[B]
[25] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
[26] -1 0 0x00000000 - 0x00000000 (0x1) IX[B]
[27] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
[28] -1 0 0x00000000 - 0x00000000 (0x1) IX[B]
[29] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
[30] -1 0 0x00000000 - 0x00000000 (0x1) IX[B]
[31] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
[32] -1 0 0x00000000 - 0x00000000 (0x1) IX[B]
[33] 0 0 0x000003b0 - 0x000003bb (0xc) IS[B]
[34] 0 0 0x000003c0 - 0x000003df (0x20) IS[B]
(II) RADEONHD(0): Mapped IO @ 0xfeaf0000 to 0x7f68e0b30000 (size 0x00010000)
(II) RADEONHD(0): Mapped FB @ 0xf0000000 to 0x7f68d4aa6000 (size 0x08000000)
(WW) RADEONHD(0): RHDCSInit: No CS for R600 and up yet.
(==) RADEONHD(0): Backing store disabled
(==) RADEONHD(0): Silken mouse enabled
(II) RADEONHD(0): RandR 1.2 enabled, ignore the following RandR disabled message.
(II) RADEONHD(0): Mapping DIG2 encoder to KLDSKP_LVTMA
(II) RADEONHD(0): On Crtc 0 Setting 59.9 Hz Mode: Modeline "1680x1050" 119.00 1680 1728 1760 1840 1050 1053 1059 1080 +hsync +vsync
None
(II) RADEONHD(0): RHDAudioSetClock: using UNIPHY_KLDSKP_LVTMA as clock source with 119000 khz
(II) RADEONHD(0): Using ACR timing N=4096 CTS=119000 for frequency 32000
(II) RADEONHD(0): Using ACR timing N=6272 CTS=132222 for frequency 44100
(II) RADEONHD(0): Using ACR timing N=6144 CTS=119000 for frequency 48000
(II) RADEONHD(0): RHDAudioSetSupported: config 0x60040 codec 0x1
(II) RADEONHD(0): DPMS enabled
(--) RandR disabled
(II) Setting vga for screen 0.
(II) Initializing built-in extension MIT-SHM
(II) Initializing built-in extension XInputExtension
(II) Initializing built-in extension XTEST
(II) Initializing built-in extension XKEYBOARD
(II) Initializing built-in extension XINERAMA
(II) Initializing built-in extension XFIXES
(II) Initializing built-in extension RENDER
(II) Initializing built-in extension RANDR
(II) Initializing built-in extension COMPOSITE
(II) Initializing built-in extension DAMAGE
(II) Initializing built-in extension XEVIE
(II) AIGLX: Loaded and initialized /usr/lib64/dri/swrast_dri.so
(II) GLX: Initialized DRISWRAST GL provider for screen 0
(II) RADEONHD(0): Setting screen physical size to 473 x 296

I am running a gentoo system with xorg 7.4 (xorg-server-1.5.2) and gentoo-kernel 2.6.27.

Any ideas what to do next?

[bridgman]

Can you pastebin the kernel messages (dmesg) ? The most common problem I see is people building the 6xx-7xx drm branch but not getting the new drm successfully installed... so they're still running the old drm which does not support their hardware.

You also might want to hop onto the #radeonhd IRC channel; there's usually a few people online there who just finished installing the 6xx-7xx code so probably have gone through some of the same issues.

[schwarzygesetzlos]

Can you pastebin the kernel messages (dmesg) ? The most common problem I see is people building the 6xx-7xx drm branch but not getting the new drm successfully installed... so they're still running the old drm which does not support their hardware.

You also might want to hop onto the #radeonhd IRC channel; there's usually a few people online there who just finished installing the 6xx-7xx code so probably have gone through some of the same issues.

Thanks for the fast reply maen!

Here's the dmesg output:
http://pastebin.ch/961

And the complete Xorg.0.log:
http://pastebin.ch/962

[rahman.duran]

Hi as I have some similar problem I want to ask here.

I have hd3200 igp. I am running OpenSuse 11.1 and Kde 4.2. I followed the same wiki as @schwarzygesetzlos. I get both drm module and radeonhd driver compiled and working as there is no error messages in xorg.conf log. But my problem is my desktop completely corrupted. Icons, colors, texts, menus, even desktop wallpaper is rendered wrong and full of garbage. I use EXA and DRI options enabled in xorg.conf. And It is beyond some corruptions, it is very very slow. Windows painted in 3-5 seconds. Scroling is like slide show :)

Is this the r6xx-r7xx branch real state or I do something wrong. All I want is fast 2d and XV.

[bridgman]

That sounds like the branch's real state but (a) that state is MUCH better than it was a week ago (much more reliable on hd3200, more functions accelerated, less corruption) (b) there are things you can do to improve your experience with the current code and (c) the amount of corruption varies from user to user and your corruption sounds worse than what some folks are seeing (but probably not the worst).

The corruption problems seem to happen in the EXA Composite function (aka EXA Render) and the slowness of scrolling and moving is because we're using an extremely basic algorithm for blitting with overlapped source and destination, and don't want to get into optimizing until all the sources of corruption have been fixed.

So...you have two different options :

1. Turn off the acceleration of EXA composite to remove the corruption, by adding Option "OXANoComposite" "true" to your xorg.conf. That should get rid of the corruption but leave the slow. If EXANoComposite option is not enough you can apparently switch acceleration to XAA which (I'm told by a user) removes 2D acceleration but keeps Xv.

2. Keep EXA Composite turned on, but get some benefit from it by enabling desktop composition in KWin. This will still have the corruption but should improve performance quite a bit. There's an option in KWin, possibly under Desktop Effectts, where you can choose the kind of acceleration used by the compositor - you want to pick "XRender" *NOT* OpenGL. The last user to try this mentioned that there were places you needed to pick "All"; let us know what you find.

Anyways, either of these approaches should improve things noticeably. Before you ask, no I don't think you can use the two options together ;)

Let us know how it works.

[rahman.duran]

That sounds like the branch's real state but (a) that state is MUCH better than it was a week ago (much more reliable on hd3200, more functions accelerated, less corruption) (b) there are things you can do to improve your experience with the current code and (c) the amount of corruption varies from user to user and your corruption sounds worse than what some folks are seeing (but probably not the worst).

The corruption problems seem to happen in the EXA Composite function (aka EXA Render) and the slowness of scrolling and moving is because we're using an extremely basic algorithm for blitting with overlapped source and destination, and don't want to get into optimizing until all the sources of corruption have been fixed.

So...you have two different options :

1. Turn off the acceleration of EXA composite to remove the corruption, by adding Option "OXANoComposite" "true" to your xorg.conf. That should get rid of the corruption but leave the slow. If EXANoComposite option is not enough you can apparently switch acceleration to XAA which (I'm told by a user) removes 2D acceleration but keeps Xv.

2. Keep EXA Composite turned on, but get some benefit from it by enabling desktop composition in KWin. This will still have the corruption but should improve performance quite a bit. There's an option in KWin, possibly under Desktop Effectts, where you can choose the kind of acceleration used by the compositor - you want to pick "XRender" *NOT* OpenGL. The last user to try this mentioned that there were places you needed to pick "All"; let us know what you find.

Anyways, either of these approaches should improve things noticeably. Before you ask, no I don't think you can use the two options together ;)

Let us know how it works.

Thanks for your advice. For your 1. approach, I have no time now to test it but I will test and feedback soon.

For the second one: No, no, no :) kwin can't enable Desktop Effects neither in Opengl nor Xrender options. I don't know why it can't. I should add this that glxgears run smooth even it runs teribly slow at 200 fps. I should add this too: I don't think there should be worse situation :) Because poor KWIN can't even renders the wallpaper and just renders its naked white&grey boxed bacground :)

I can Attach xorg.conf log if it is necessary.


Anyways thanks for your advice.

[PWMx]

Because poor KWIN can't even renders the wallpaper and just renders its naked white&grey boxed bacground

I get the very same with the current git - So it's the state of branch just now.
But still this is much better than a while a go. Then it just froze the box :p

Maybe we'll try again after some new commits ..

[rahman.duran]

I get the very same with the current git - So it's the state of branch just now.
But still this is much better than a while a go. Then it just froze the box :p

Maybe we'll try again after some new commits ..

Yep, me to :) I also wish that I have the in-depth knowlage of C language
and driver developing skill. Then I should just rollover my sleeves and get my hand dirty:) Waiting is just boring an annoying. Unfortunately, I am a Java developer without these knowlage:(

[schwarzygesetzlos]

Can you pastebin the kernel messages (dmesg) ? The most common problem I see is people building the 6xx-7xx drm branch but not getting the new drm successfully installed... so they're still running the old drm which does not support their hardware.

You also might want to hop onto the #radeonhd IRC channel; there's usually a few people online there who just finished installing the 6xx-7xx code so probably have gone through some of the same issues.

Hmm, seems my first posting with the pastebinned kernel output got lost... So here we try again

kernel dmesg output:
http://pastebin.ch/961

full xorg.0.log:
http://pastebin.ch/962


You are right, there definetly is a problem with the kernel drm module on my machine. Maybe it's my fault but I am not aware where I made the error.

I did following procedure to install it:
git clone git://anongit.freedesktop.org/mesa/drm
git checkout -b r6xx-r7xx-support origin/r6xx-r7xx-support
cd drm/linux-core
make
cp drm.ko radeon.ko /lib/modules/2.6.27-gentoo-r8/kernel/drivers/gpu/
depmod

In tne kernel-config DRM and the radeon module are deselected. I did a make clean within the kernel dir before compiling the git drm module.

Hmm, I just had a thought that my errors have something to do with an outdated mesa? I am using mesa-7.2.


@rahman.duran:
Good to hear that accelleration support basically works on your machine! :) I too experience the slowness of scrolling but without screen corruptions. Well, there's at least no mess on my screen without the working drm module ;)

Could you please have a look which versions of: xorg-server, mesa, libdrm you are running and from what kernel sources you did the compile?

[bridgman]

I don't understand the comment about "KWin can't enable desktop effects". This is already working for a couple of KDE 4.x users. There is an "Advanced" button you need to click in order to reach the dialog where you can choose XRender.

[rahman.duran]

I don't understand the comment about "KWin can't enable desktop effects". This is already working for a couple of KDE 4.x users. There is an "Advanced" button you need to click in order to reach the dialog where you can choose XRender.

I know that. As I said None of the options in the "Advanced" tab make any difference. Choosing XRender and clicking Apply just gives "Can't enable desktop effects" error. It is the same With OpenGL mode. But if I check the "Disable functionality check" and choose Opengl, it enables Desktop effects but as it is in software rendering mode that I understand, none of the windows can repaint, the whole desktop turns grey. But it does't lock up becouse KWin reverts changes in about 10sn, because I cant click the keep changes button as it didn't painted on screen :)

@rahman.duran:
Good to hear that accelleration support basically works on your machine! I too experience the slowness of scrolling but without screen corruptions. Well, there's at least no mess on my screen without the working drm module


No there is no working Accelleration for me here, at least I can't call this messiness Acceleration :) Maybe I can upload a video to video.google.com then you can see how it works on my machine :)

Edit: I took some screenshots about the corruption.

This is what happens when you move windows around. This should explain how terribly slow window paintings are.
http://imagebin.ca/view/eYZCYb.html

these are screenshots:

http://imagebin.ca/view/Q2634Xn.html

http://imagebin.ca/view/xrqTtS.html

http://imagebin.ca/view/w5iREh3.html

http://imagebin.ca/view/xIPNJoN.html

And this is my xorg.0.log
http://pastebin.com/d34a48183

[tettamanti]

Hmm, seems my first posting with the pastebinned kernel output got lost... So here we try again

kernel dmesg output:
http://pastebin.ch/961

I don't see the radeon driver being loaded, it should say something like:

[drm] Initialized drm 1.1.0 20060810
pci 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
pci 0000:01:00.0: setting latency timer to 64
[drm] Initialized radeon 1.29.0 20080613 on minor 0
[drm] Used old pci detect: framebuffer loaded
[drm] Setting GART location based on new memory map
[drm] Loading RV630 CP Microcode
[drm] Loading RV630 PFP Microcode
[drm] Resetting GPU
[drm] writeback test succeeded in 1 usecs

I did following procedure to install it:
git clone git://anongit.freedesktop.org/mesa/drm
git checkout -b r6xx-r7xx-support origin/r6xx-r7xx-support
cd drm/linux-core
make
cp drm.ko radeon.ko /lib/modules/2.6.27-gentoo-r8/kernel/drivers/gpu/
depmod

Ah, I think you didn't overwrite the old modules. drm.ko should be in /lib/modules/2.6.27-gentoo-r8/kernel/drivers/gpu/drm, while radeon.ko should be in /lib/modules/2.6.27-gentoo-r8/kernel/drivers/gpu/drm/radeon.
Just to be sure you can rmmod the two modules (without X running) and load them from the compiled sources (using insmod); in this way you are sure to load the correct drivers ;)
If it's still not working check that your card is listed in shared-core/drm_pciids.txt (if you don't know how post the output of lspci -nn).

Hmm, I just had a thought that my errors have something to do with an outdated mesa? I am using mesa-7.2.

Nope, ATM there's only the software renderer.

[bridgman]

Choosing XRender and clicking Apply just gives "Can't enable desktop effects" error. It is the same With OpenGL mode. But if I check the "Disable functionality check" and choose Opengl, it enables Desktop effects but as it is in software rendering mode

Thanks; I'll try to play with the KDE settings here.

Just for completeness I have to ask.. is disabling functionality check and choosing XRender an option ?

[tball]

@bridgman

I got an Ati HD3650 mobility.
Any successhistories about running EXA and xv accelerated desktop with git version of radeonhd?

I'm very excited of this opensource driver, and would like to test it. But should I just use fglrx for now and wait? I have no idea of the current state of the r6xx/r7xx branch.

[bridgman]

As of yesterday the 6xx-7xx open source code is still showing varying degrees of corruption when EXA Composite is enabled, but Xv seems to be pretty solid. Blits with overlapping src/dst areas have not been optimized, so scrolling and dragging are still slow unless you are running an EXA/XRender-based compositing desktop (metacity, KWin etc..). There are a number of people hanging out on #radeonhd who have successfully built and installed the drivers so if you want to give them a try you won't be alone.

It's still a pain moving back and forth from open source to fglrx so dual-booting would probably be a better option if you want to follow the progress of the open source drivers while still using fglrx.

I don't think we can say the drivers are ready for normal use yet, but they are getting pretty close and having at least a small number of people use them helps us to find and fix issues more quickly. Having developers (or even would-be developers) look at the code also helps; a number of problems have been found and fixed that way.

[schwarzygesetzlos]

Ah, I think you didn't overwrite the old modules. drm.ko should be in /lib/modules/2.6.27-gentoo-r8/kernel/drivers/gpu/drm, while radeon.ko should be in /lib/modules/2.6.27-gentoo-r8/kernel/drivers/gpu/drm/radeon.
Just to be sure you can rmmod the two modules (without X running) and load them from the compiled sources (using insmod); in this way you are sure to load the correct drivers ;)
If it's still not working check that your card is listed in shared-core/drm_pciids.txt (if you don't know how post the output of lspci -nn).


Thanks for the input! That was one of my first thoughts too. But I recompiled the kernel completely without the DRM and RADEON module. So there was no gpu/ dir in the corresponding /lib/modules/2.6.27-gentoo-r8/ I created the gpu/ dir myself and simply copied the modules over. Of course did a depmod -a after that. Just to be sure:
yea 2.6.27-gentoo-r8 # cd /lib/modules/2.6.27-gentoo-r8/
yea 2.6.27-gentoo-r8 # find -name 'drm.ko'
./kernel/drivers/gpu/drm.ko
yea 2.6.27-gentoo-r8 # find -name 'radeon.ko'
./kernel/drivers/gpu/radeon.ko
Here's my lspci -nn output:
yea 2.6.27-gentoo-r8 # lspci -nn
00:00.0 Host bridge [0600]: Advanced Micro Devices [AMD] RS780 Host Bridge [1022:9600]
00:01.0 PCI bridge [0604]: ASRock Incorporation Device [1849:9602]
00:0a.0 PCI bridge [0604]: Advanced Micro Devices [AMD] RS780 PCI to PCI bridge (PCIE port 5) [1022:9609]
00:11.0 SATA controller [0106]: ATI Technologies Inc SB700/SB800 SATA Controller [AHCI mode] [1002:4391]
00:12.0 USB Controller [0c03]: ATI Technologies Inc SB700/SB800 USB OHCI0 Controller [1002:4397]
00:12.1 USB Controller [0c03]: ATI Technologies Inc SB700 USB OHCI1 Controller [1002:4398]
00:12.2 USB Controller [0c03]: ATI Technologies Inc SB700/SB800 USB EHCI Controller [1002:4396]
00:13.0 USB Controller [0c03]: ATI Technologies Inc SB700/SB800 USB OHCI0 Controller [1002:4397]
00:13.1 USB Controller [0c03]: ATI Technologies Inc SB700 USB OHCI1 Controller [1002:4398]
00:13.2 USB Controller [0c03]: ATI Technologies Inc SB700/SB800 USB EHCI Controller [1002:4396]
00:14.0 SMBus [0c05]: ATI Technologies Inc SBx00 SMBus Controller [1002:4385] (rev 3a)
00:14.1 IDE interface [0101]: ATI Technologies Inc SB700/SB800 IDE Controller [1002:439c]
00:14.2 Audio device [0403]: ATI Technologies Inc SBx00 Azalia (Intel HDA) [1002:4383]
00:14.3 ISA bridge [0601]: ATI Technologies Inc SB700/SB800 LPC host controller [1002:439d]
00:14.4 PCI bridge [0604]: ATI Technologies Inc SBx00 PCI to PCI Bridge [1002:4384]
00:14.5 USB Controller [0c03]: ATI Technologies Inc SB700/SB800 USB OHCI2 Controller [1002:4399]
00:18.0 Host bridge [0600]: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] HyperTransport Technology Configuration [1022:1100]
00:18.1 Host bridge [0600]: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Address Map [1022:1101]
00:18.2 Host bridge [0600]: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] DRAM Controller [1022:1102]
00:18.3 Host bridge [0600]: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Miscellaneous Control [1022:1103]
01:05.0 VGA compatible controller [0300]: ATI Technologies Inc Radeon HD 3300 Graphics [1002:9614]

As you suggested I looked up the ID of my HD 3300 in the database. Seems it's not listed ATM, only the older models:
yea shared-core # cat drm_pciids.txt | grep "961"
0x1002 0x5961 CHIP_RV280 "ATI Radeon RV280 9200"
0x1002 0x9610 CHIP_RS780|RADEON_NEW_MEMMAP|RADEON_IS_IGP "ATI Radeon HD 3200 Graphics"
0x1002 0x9611 CHIP_RS780|RADEON_NEW_MEMMAP|RADEON_IS_IGP "ATI Radeon 3100 Graphics"
0x1002 0x9612 CHIP_RS780|RADEON_NEW_MEMMAP|RADEON_IS_IGP "ATI Radeon HD 3200 Graphics"
0x1002 0x9613 CHIP_RS780|RADEON_NEW_MEMMAP|RADEON_IS_IGP "ATI Radeon 3100 Graphics"

So it is the missing PCI-ID which prevents the drm module from picking up my card?

My main motivation to run this development drivers is to get the XVIDEO-Extension working. If I don't need the drm module to achieve this, then it's ok for me.

But as long as xvinfo says yea ~ # xvinfo

X-Video Extension version 2.2
screen #0
no adaptors present
i have to continue ;)

[bridgman]

Yep, sounds like the missing ID is a problem. You could probably confirm that by looking at the drm log messages with dmesg.

You can update the ID list on your local copy of the source and see if that helps; I'll also mention it to agd5f.

[agd5f]

I've added the id to drm git.

[bridgman]

OK, guess I won't bother mentioning it to agd5f then :D

[rahman.duran]

Thanks; I'll try to play with the KDE settings here.

Just for completeness I have to ask.. is disabling functionality check and choosing XRender an option ?

Nope, It is not :) BTW I played with that again but this time it could not enabled composite with "OpenGL" and "Disable test" options.

[schwarzygesetzlos]

I've added the id to drm git.

Thanks bridgman & agd5f! The drm module gets loaded now - awesome dudes!

But now I got another problem ;) After X has started with DRI enabled & radeonhd, switching to terminal via CTRL-F1 and switching back to X crashes the machine completely. It turns the screen to black, though it gets some kind of signal (the monitor LED shows it's still active). The only way to get out is a soft reset.

Here's my xorg.0.log:
http://pastebin.ch/969

Out of curiosity I tried using the new drm module in connection with the xf86-video-ati-6.10.0. According to the log the drm module gets utilized too, but at some point there's an error and the xf86-video-ati-6.10.0 doesn't get to X using drm.

Here's the xorg.log with xf86-video-ati-6.10.0:
http://pastebin.ch/970

The dmesg output with the ati driver is:
[drm] Initialized radeon 1.29.0 20080613 on minor 0
[drm] Used old pci detect: framebuffer loaded
[drm] Setting GART location based on new memory map
[drm] GART aligned down from 0x00010000 to 0x00000000
[drm] Num pipes: 2
[drm] writeback test failed
[drm] Setting GART location based on new memory map
[drm] GART aligned down from 0x00010000 to 0x00000000
[drm] Num pipes: 2
[drm] writeback test failed
[drm] Setting GART location based on new memory map
[drm] GART aligned down from 0x00010000 to 0x00000000
[drm] Num pipes: 2
[drm] writeback test failed


BTW I only compiled the kernel modules from the drm-git sources. Is it necessary to use the libdrm-git too? ATM I am using libdrm-2.3.1

[tettamanti]

Out of curiosity I tried using the new drm module in connection with the xf86-video-ati-6.10.0. According to the log the drm module gets utilized too, but at some point there's an error and the xf86-video-ati-6.10.0 doesn't get to X using drm.

Here's the xorg.log with xf86-video-ati-6.10.0:
http://pastebin.ch/970

This is expected, RADEONAccelInit fails for anything >= R600; IOW ATM only radeonhd supports acceleration with R600 and newer.

[PWMx]

Todays git :

EXA corruption is still there. (But not as bad :))
The font corruption seems to be replaced with wrong letters :confused:
Switching consoles ended with black screen and killed keyboard. (Init 0 via acpid power button worked ..)

EDIT: My mistake - XV won't work with "XAA" as agd5f correctly pointed out.
(I just wonder why it is faster than "none" and video tearing isn't there as with "shadowfb" ?)

[agd5f]

There is no XAA support. If you specify XAA you're probably falling back to shadowfb.

[bridgman]

Alex just pushed fixes for all the known corruption issues, and also some performance optimizations for scrolling and dragging. For anyone running 6xx/7xx open source drivers this is a good time to update your drm and radeonhd drivers.

[bridgman]

Looks like one of the performance optimizations introduced some *different* corruption so agd5f has reverted that one change. Anyone who picked up source in the last few minutes should pull the latest.

[PuckPoltergeist]

Looks like one of the performance optimizations introduced some *different* corruption so agd5f has reverted that one change. Anyone who picked up source in the last few minutes should pull the latest.

Gna, so I was a little bit to fast. :D

You're speaking about corruption like this?
http://www.stud.tu-ilmenau.de/~johi-in/Bildschirmfoto4.jpg
(after scrolling in konqueror)

[bridgman]

Probably. Alex tried a few different cards and hasn't been able to see corruption on any of them. That's not unusual though...

[monraaf]

Looks like one of the performance optimizations introduced some *different* corruption so agd5f has reverted that one change. Anyone who picked up source in the last few minutes should pull the latest.

I bet some compiz users would have liked the smearing effect :)

Ok with latest git looks like all corruption and hangs are gone, Xv is working and scrolling performance in Firefox looks acceptable but improvement is welcome.

[PuckPoltergeist]

Probably. Alex tried a few different cards and hasn't been able to see corruption on any of them. That's not unusual though...

If I can help debugging this, let me know. The logs didn't show anything so far.

edit: Still does not survive switching back from console to X. The switch from X to console (Ctrl+Alt+F1) works, but when switching back to X I only get a black screen. :(

[bridgman]

PuckPoltergeist, are you still getting corruption after the latest commit (when Alex reverted an earlier commit) ?

[rahman.duran]

I just tried the latest git code. I must say, impressive work :) Thanks Alex. All the corruptions gone. I can enable Desktop Effects on kde 4.1.3 with Xrender and Disable Functionality Check options.

Here some feedback about speed. With composite the performance is worse. Scrolling is not usable. Window drawings are less responsive (if this is correct word for it, excuse my english). Without composite, the overall performanse is good enough. Not fast as radeonhd driver anyway but it is usable. The scrolling it absolutely far better than yesterdays build :) it gives me nearly the same taste of performance as radeonhd driver. So I have no real complain about scrolling as before. I didn't tried XV video yet.

So, again thanks. Keep up good work :D

Edit: more feedback about performance :)

It takes about 2 seconds in fireforx to switch tabs and now after abit longer test scrolling seems not that fast. And switching windows on desktop is taking more, about 3 sec to render window contends. Yang Zhao posted a patch to mail-list about R600OverlapCopy() optimisation. I am going to test it now.

[PuckPoltergeist]

PuckPoltergeist, are you still getting corruption after the latest commit (when Alex reverted an earlier commit) ?
No, corruptions are gone. Only problem so far is when switching back from console to X. Then X hangs with black screen.

[rahman.duran]

No, corruptions are gone. Only problem so far is when switching back from console to X. Then X hangs with black screen.

I can comfirm this issue on my hd3200 igp. The only way to swithcing VT and back to X is while you are in VT kill the X server first ( one way is to going to "runlevel 3" and then "runlevel 5" but this time you loose your running session).

[PWMx]

Current radeon git r6xx-r7xx-branch seems to work fine ! Rs780: EXA + Xv :D

But for some reason similar radeonhd causes massive corruption.

[agd5f]

Weird. Both radeon and radeonhd work perfectly for me on my RS780. I think I have the same board you do (Gigabyte MA78GM-S2H).

[d2kx]

Is the mesa/drm r6xx-r7xx branch also needed for EXA/Xv? I think so but PWMx didn't mention it. Also, do you have to just compile and make install the mesa/drm thing or do you need to patch in the kernel somehow? Would be interested in trying that out as my ASUS HD 4850 1GB with Glaciator cooling (the default one for ASUS now) is an "unknown model" with fglrx and RadeonHD 1.2.4 (but works without any problems with both).

[bridgman]

Is the mesa/drm r6xx-r7xx branch also needed for EXA/Xv? I think so but PWMx didn't mention it. Also, do you have to just compile and make install the mesa/drm thing or do you need to patch in the kernel somehow? Would be interested in trying that out as my ASUS HD 4850 1GB with Glaciator cooling (the default one for ASUS now) is an "unknown model" with fglrx and RadeonHD 1.2.4 (but works without any problems with both).

Yes, you need latest mesa/drm and radeonhd/ati 6xx-7xx branches. The drm loads the microcode and manages the ring buffer used for command submission to the 3D engine.

EDIT - see next post for build/install info

[agd5f]

You just need the kernel modules ffrom teh r6xx-r7xx-support branch; you don't need to update your libdrm. So all you need to do with the r6xx-r7xx-support branch is:

cd linux-core
make drm.o radeon.o
sudo cp drm.ko /path/to/your/kernel/modules
sudo cp radeon.ko /path/to/your/kernel/modules
sudo depmod -a

[tball]

Latest git works fine for me :-D
Radeon HD3200 mobility.

Thank you so much for this driver.

[tball]

I was to fast writing, I forgot to mention my kwin composite doesn't work. Not even with xrender. Anybody got it working with jaunty or intrepid? The error message says something about my X configuration. Do I need something specific in xorg.conf to activate composite?

[tball]

Hmm, I get thiese error lines in my Xorg log:

[1 sec: 253577 usec](EE) AIGLX error: dlopen of /usr/lib/dri/r600_dri.so failed (/usr/lib/dri/r600_dri.so: cannot open shared object file: No such file or directory)
[1 sec: 253665 usec](EE) AIGLX: reverting to software rendering

Am I installing the radeonhd modules wrong? Or whats going on.

[agd5f]

You can ignore those errors for now. There is no 3D driver yet. The current code just provides EXA and Xv support.

[bridgman]

I think those are normal. There is no 3D driver yet, so AIGLX is kinda hypothetical for now.

Someone mentioned that composite was working with radeon but not radeonhd, which (if true) is another mystery because the added code is the same in the two drivers.

[tball]

Ok will I have to compile another driver than radeonhd? I mean could I just type radeon instead of radeonhd in my xorg.conf after following the instructions at: http://www.x.org/wiki/radeonhd%3Ar6xx_r7xx_branch?

Great support!

[bridgman]

Yeah, you would need to download and build source from the 6xx-7xx branch of xorg/driver/xf86-video-ati rather than from the same-named branch of xorg/driver/xf86-video-radeonhd.

Alternatively, just hang in with radeonhd and watch for updates; it shouldn't be long before someone figures out why the two drivers are behaving differently since so much of the code is identical.

[tball]

Thanks Bridgman and agd5f.

I'll just hang on with the radeonhd driver for now. Just tried radeon, and it makes my computer reboot when Xorg starts.

Nothing unusually in the xorg.0.log though. Weird.

[RealNC]

The argument about having two drivers was that radeon was only there to support up to R500.

Er, now what's the catch?

[rahman.duran]

For those who interested my kde4.2 composite on OpenSuse 11.1 with hd3200 igp works with "XRender" and " Disable Fuctionality Check " options. But don't ask me about performance :) I am talking about radeonhd driver BTW.

[tball]

For those who interested my kde4.2 composite on OpenSuse 11.1 with hd3200 igp works with "XRender" and " Disable Fuctionality Check " options. But don't ask me about performance :) I am talking about radeonhd driver BTW.

Okay interesting.
xserver version? xorg.conf configuration?

[rahman.duran]

Okay interesting.
xserver version? xorg.conf configuration?

X.Org X Server 1.5.2
nothing special in xorg.conf just default.

# /.../
# SaX generated X11 config file
# Created on: 2009-02-05T17:26:22+0200.
#
# Version: 8.1
# Contact: Marcus Schaefer <sax@suse.de>, 2005
# Contact: SaX-User list <https://lists.berlios.de/mailman/listinfo/sax-users>
#
# Automatically generated by [ISaX] (8.1)
# PLEASE DO NOT EDIT THIS FILE!
#

Section "Files"
FontPath "/usr/share/fonts/misc:unscaled"
FontPath "/usr/share/fonts/local"
FontPath "/usr/share/fonts/75dpi:unscaled"
FontPath "/usr/share/fonts/100dpi:unscaled"
FontPath "/usr/share/fonts/Type1"
FontPath "/usr/share/fonts/URW"
FontPath "/usr/share/fonts/Speedo"
FontPath "/usr/share/fonts/PEX"
FontPath "/usr/share/fonts/cyrillic"
FontPath "/usr/share/fonts/latin2/misc:unscaled"
FontPath "/usr/share/fonts/latin2/75dpi:unscaled"
FontPath "/usr/share/fonts/latin2/100dpi:unscaled"
FontPath "/usr/share/fonts/latin2/Type1"
FontPath "/usr/share/fonts/latin7/75dpi:unscaled"
FontPath "/usr/share/fonts/baekmuk:unscaled"
FontPath "/usr/share/fonts/japanese:unscaled"
FontPath "/usr/share/fonts/kwintv"
FontPath "/usr/share/fonts/truetype"
FontPath "/usr/share/fonts/uni:unscaled"
FontPath "/usr/share/fonts/CID"
FontPath "/usr/share/fonts/ucs/misc:unscaled"
FontPath "/usr/share/fonts/ucs/75dpi:unscaled"
FontPath "/usr/share/fonts/ucs/100dpi:unscaled"
FontPath "/usr/share/fonts/hellas/misc:unscaled"
FontPath "/usr/share/fonts/hellas/75dpi:unscaled"
FontPath "/usr/share/fonts/hellas/100dpi:unscaled"
FontPath "/usr/share/fonts/hellas/Type1"
FontPath "/usr/share/fonts/misc/sgi:unscaled"
FontPath "/usr/share/fonts/xtest"
FontPath "/opt/kde3/share/fonts"
InputDevices "/dev/gpmdata"
InputDevices "/dev/input/mice"
EndSection

Section "ServerFlags"
Option "AllowMouseOpenFail" "on"
Option "ZapWarning" "on"
EndSection

Section "Module"
Load "dri"
Load "dbe"
Load "freetype"
Load "extmod"
Load "glx"
EndSection

Section "InputDevice"
Driver "kbd"
Identifier "Keyboard[0]"
Option "Protocol" "Standard"
Option "XkbLayout" "tr"
Option "XkbModel" "microsoftpro"
Option "XkbOptions" "caps:shift"
Option "XkbRules" "xfree86"
EndSection


Section "InputDevice"
Driver "mouse"
Identifier "Mouse[1]"
Option "Buttons" "12"
Option "Device" "/dev/input/mice"
Option "Name" "A4 Tech PS/2+USB Mouse"
Option "Protocol" "explorerps/2"
Option "Vendor" "Sysp"
Option "ZAxisMapping" "4 5"
EndSection


Section "Monitor"
DisplaySize 518 324
HorizSync 30-81
Identifier "Monitor[0]"
ModelName "SAMSUNG SYNCMASTER"
Option "DPMS"
Option "PreferredMode" "1920x1200"
VendorName "SAM"
VertRefresh 43-60
UseModes "Modes[0]"
EndSection


Section "Modes"
Identifier "Modes[0]"
Modeline "1920x1200" 151.63 1920 2032 2240 2560 1200 1201 1204 1234
Modeline "1920x1200" 119.75 1920 1968 2000 2080 1200 1203 1209 1227 +HSync -Vsync
Modeline "1600x1200" 152.67 1600 1704 1880 2160 1200 1201 1204 1240
Modeline "1600x1200" 121.25 1600 1648 1680 1760 1200 1203 1207 1232 +HSync -Vsync
Modeline "1280x1024" 108.88 1280 1360 1496 1712 1024 1025 1028 1060
Modeline "1280x1024" 89.25 1280 1328 1360 1440 1024 1027 1034 1053 +HSync -Vsync
Modeline "1280x960" 102.10 1280 1360 1496 1712 960 961 964 994
Modeline "1280x960" 83.75 1280 1328 1360 1440 960 963 967 987 +HSync -Vsync
Modeline "1024x768" 64.11 1024 1080 1184 1344 768 769 772 795
Modeline "1024x768" 55.00 1024 1072 1104 1184 768 771 775 790 +HSync -Vsync
Modeline "800x600" 38.22 800 832 912 1024 600 601 604 622
Modeline "800x600" 34.75 800 848 880 960 600 603 607 617 +HSync -Vsync
Modeline "1920x1200" 154 1920 1968 2000 2080 1200 1203 1209 1235 -hsync +vsync
EndSection


Section "Screen"
DefaultDepth 24
SubSection "Display"
Depth 15
Modes "1920x1200" "1600x1200" "1280x1024" "1280x960" "1024x768" "800x600"
Virtual 3840 1200
EndSubSection
SubSection "Display"
Depth 16
Modes "1920x1200" "1600x1200" "1280x1024" "1280x960" "1024x768" "800x600"
Virtual 3840 1200
EndSubSection
SubSection "Display"
Depth 24
Modes "1920x1200" "1600x1200" "1280x1024" "1280x960" "1024x768" "800x600"
Virtual 3840 1200
EndSubSection
SubSection "Display"
Depth 8
Modes "1920x1200" "1600x1200" "1280x1024" "1280x960" "1024x768" "800x600"
Virtual 3840 1200
EndSubSection
Device "Device[0]"
Identifier "Screen[0]"
Monitor "Monitor[0]"
EndSection


Section "Device"
BoardName "ATI Radeon HD 3200 Graphics"
Driver "radeonhd"
Identifier "Device[0]"
Option "monitor-DVI-D_1" "Monitor[0]"
VendorName "ATI"
Option "AccelMethod" "exa"
Option "DRI" "on"
EndSection



Section "ServerLayout"
Identifier "Layout[all]"
InputDevice "Keyboard[0]" "CoreKeyboard"
InputDevice "Mouse[1]" "CorePointer"
Option "Clone" "off"
Option "Xinerama" "off"
Screen "Screen[0]"
EndSection


Section "DRI"
Group "video"
Mode 0660
EndSection

Section "Extensions"
EndSection


By the way I am really impressed the development of the driver. Just one week ago it was far away from usable for me. But now it is really matured. Thanks all the devs.

[tball]

Hmm weird.

I don't got the:
Section "DRI"
Group "video"
Mode 0660
EndSection


But thats about it.

[rahman.duran]

I have noticed image quality differences between x11 video out and xv vo on latest git( btw I dont know if it was the same before as I had never a chance to test it before ). On x11 -vo the images darker parts are darker than on xv -vo. Is it some king of gama stuf in xv code of the branch? I can post screenshots if you want.

[rahman.duran]

Hmm weird.

I don't got the:
Section "DRI"
Group "video"
Mode 0660
EndSection


But thats about it.

I dont think it is related with composite it seems it is about XV video permissions. Ofcourse I am just guessing.

[d2kx]

Ok, I've got RadeonHD running with EXA now (with new drm and radeon modules of course) on my RV770 (had to add AccelMethod "EXA") and it is much faster than VESA, but Xv doesn't seem to be avaible, because xvinfo tells me there's no adapters present. Also interesting in the Xorg.0.log was:

(II) RADEONHD(0): Unknown card detected: 0x9442:0x1043:0x0264.
If - and only if - your card does not work or does not work optimally
please contact radeonhd@opensuse.org to help rectify this.
Use the subject: 0x9442:0x1043:0x0264: <name of board>
and *please* describe the problems you are seeing
in your message.
(--) RADEONHD(0): Detected an RV770 on an unidentified card

and

(II) RADEONHD(0): The detected amount of videoram exceeds the PCI BAR aperture.
(II) RADEONHD(0): Using only 262144kB of the total 1048576kB.
(--) RADEONHD(0): VideoRAM: 262144 kByte

[bridgman]

Sounds like DRI is not initialized and you're running with shadowfb.

Do you have DRI enabled in your xorg.conf file (according to the radeonhd wiki you just need :

Option "DRI"

... with no true/false value.

Maybe pastebin your log and dmesg output ?

EDIT - regarding the two messages you quoted :

For the first message, "unrecognized hardware" really means "I recognize the GPU, but I don't have your specific card in my list of boards which need non-standard treatment, so I'm just going to run the normal code for your GPU". No problem.

For the second message, that's normal as well. The PCI BARs only cover up to 256MB of address space, but modern cards have more than that. The implication is that your CPU can only easily access the first 256MB, but your GPU can use the entire memory on the card. Again, not a problem; the CPU usually only needs to access frame buffer and offscreen buffers, and occasionally a depth buffer; most of the RAM is used for textures, which are usually blitted up to video memory and accessed only by the GPU.

Video acceleration also uses a frame cache for motion compensation, but again this information is normally only accessed by the GPU so the BAR limit is no problem.

[d2kx]

DRI is enabled now, but still no Xv (if xvinfo is not lying).

Xorg.0.log
http://pastebin.com/m72692a1c

dmesg
http://pastebin.com/m79ed6837

I just booted up a openSUSE 11.1 KDE4 LiveCD and 2D performance and overall responsiveness seemed much better, even with ShadowFB. Sometimes I think Ubuntu's Xserver is just one big piece of... cake.

Thanks for your help, bridgman & agd5f!

[bridgman]

I think you have an old version of drm. Are you building both radeonhd and drm from the 6xx-7xx branches ?

[d2kx]

I did exactly what agd5f said. Yes, I used the r6xx-r7xx-support branch from mesa/drm, too. I replaced the modules that were in

/lib/modules/2.6.28-something/kernel/drivers/gpu/drm/drm.ko
/lib/modules/2.6.28-something/kernel/drivers/gpu/drm/radeon/radeon.ko

With the newer ones...

[bridgman]

Sounds reasonable. You definitely seem to be running with an old drm, though -- dmesg output showed none of the messages (eg microcode loading) that come with the latest drm code.

The message from your X log was :

[0 sec: 572990 usec](EE) RADEONHD(0): RHDDRIVersionCheck: drmOpen("radeon", "pci:0000:01:00.0") failed.
[0 sec: 573042 usec](WW) RADEONHD(0): RHDDRIPreInit: Version check failed. Disabling DRI.

EDIT - did you reboot after updating the kernel modules ?

Ignore my comment about the commit; you do have the right radeonhd source. Sorry.

[d2kx]

Yes, I ran depmod -a and rebooted. Maybe it's just an Ubuntu thing... and someone else knows what's wrong. Well, it's too late now (11.30pm) to look into the issue.

[schwarzygesetzlos]

I can comfirm this issue on my hd3200 igp. The only way to swithcing VT and back to X is while you are in VT kill the X server first ( one way is to going to "runlevel 3" and then "runlevel 5" but this time you loose your running session).

Yep, can confirm this issue too on my HD3300 IGP using todays (09.02.2009) git.

Hope this bug gets squashed next if the git gets updated by the devs as massive as the last few days. ;)

[susikala]

Since I see the people here use xserver 1.5.3., I wanted to ask if any of the new code supports xserver 1.6. yet? With HD3200 on Jaunty I get horrible performance with both radeon and radeonhd. Is this code only useable right now with the old xserver?

[bridgman]

I have seen the code used on very recent X servers, so it's probably OK to run with Jaunty. In general the latest open source driver code tends to track the latest X server code since the two evolve together.

Remember that you will need the 6xx-7xx support branches for both DRM and the X driver (radeonhd or ati).

[tball]

Since I see the people here use xserver 1.5.3., I wanted to ask if any of the new code supports xserver 1.6. yet? With HD3200 on Jaunty I get horrible performance with both radeon and radeonhd. Is this code only useable right now with the old xserver?

I use the xserver 1.6 included in Jaunty. It runs fast for me.

[susikala]

Yes, I ran depmod -a and rebooted. Maybe it's just an Ubuntu thing... and someone else knows what's wrong. Well, it's too late now (11.30pm) to look into the issue.

I run Ubuntu too. For some reason, Ubuntu wouldn't autoload the compiled drm module. I went into the folder where I copied it and insmod'd it. Restarted X, and xvinfo works. -vo xv in mplayer works like a charm, so I added it into /etc/modules (or does anyone know how to get Ubuntu to accept it?). I get somewhat teary video with radeonhd now, so I think I'm going to test the 6xx/7xx branch in radeon too. But it's quite blazingly fast.

So, I can confirm this is working for me on HD3200 / Jaunty / latest xserver as of 11.02.09.

P.S. bridgman: are the steps to get radeon to work with 6xx/7xx hardware basically the same?

[susikala]

I get somewhat teary video with radeonhd now, so I think I'm going to test the 6xx/7xx branch in radeon too.

I'm referring to this (http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?h=r6xx-r7xx-support&id=8e9ef8ff581892cbe1b7ea56d48b9a1abd70179d) commit, I was pointed to it by someone in chat. I couldn't find it in the radeonhd tree, does anyone know if it's going to be ported there too?

[Kano]

@susikala

It is highly unlikely that you need to load the drm module on your own as it is a depend of the radeon module. If that is not autoloaded by the radeon driver, then it is an error of that driver. It does not load of course when a fglrx module was loaded.

[schwarzygesetzlos]

With todays drm source swichting between X and VT no longer poses a problem! Finally got radeonhd & XV working flawlessly - JAY :D Thanks to all for your help!

[highlandsun]

The drm_pciids.txt file has the Mobility HD3400 lines in the wrong order in the file, so the cards are not recognized. Moving them to the proper location makes them work:


diff --git a/shared-core/drm_pciids.txt b/shared-core/drm_pciids.txt
index 5792eb3..20a51fe 100644
--- a/shared-core/drm_pciids.txt
+++ b/shared-core/drm_pciids.txt
@@ -286,12 +286,12 @@
0x1002 0x958E CHIP_RV630|RADEON_NEW_MEMMAP "ATI Radeon HD 2600 LE"
0x1002 0x958F CHIP_RV630|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Fir
0x1002 0x95C0 CHIP_RV620|RADEON_NEW_MEMMAP "ATI Radeon HD 3470"
+0x1002 0x95C2 CHIP_RV620|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Rad
+0x1002 0x95C4 CHIP_RV620|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Rad
0x1002 0x95C5 CHIP_RV620|RADEON_NEW_MEMMAP "ATI Radeon HD 3450"
0x1002 0x95C6 CHIP_RV620|RADEON_NEW_MEMMAP "ATI Radeon HD 3450"
0x1002 0x95C7 CHIP_RV620|RADEON_NEW_MEMMAP "ATI Radeon HD 3430"
0x1002 0x95C9 CHIP_RV620|RADEON_NEW_MEMMAP "ATI Radeon HD 3450"
-0x1002 0x95C2 CHIP_RV620|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Rad
-0x1002 0x95C4 CHIP_RV620|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Rad
0x1002 0x95CC CHIP_RV620|RADEON_NEW_MEMMAP "ATI FirePro V3700"
0x1002 0x95CD CHIP_RV620|RADEON_NEW_MEMMAP "ATI FireMV 2450"
0x1002 0x95CE CHIP_RV620|RADEON_NEW_MEMMAP "ATI FireMV 2260"


Note: I haven't gotten any further than loading drm.ko and radeon.ko yet, dunno if the radeonhd driver will actually work for me. I just finally see the drm init in dmesg now.

[agd5f]

The drm_pciids.txt file has the Mobility HD3400 lines in the wrong order in the file, so the cards are not recognized. Moving them to the proper location makes them work:

Note: I haven't gotten any further than loading drm.ko and radeon.ko yet, dunno if the radeonhd driver will actually work for me. I just finally see the drm init in dmesg now.

The order doesn't matter. The drm doesn't actually get initialized until X loads.

[PuckPoltergeist]

With todays drm source swichting between X and VT no longer poses a problem! Finally got radeonhd & XV working flawlessly - JAY :D Thanks to all for your help!

Yeah, finally it works now. But I've still some kind of screen corruption. For example, there is something like a horizontal dash on top of the cursor. It isn't visible on the screenshots cause no cursor is shown there. Will try if I can take a screenshot, where this is visible.

[bridgman]

Does the corruption leave "trails" or is it only visible in the cursor's current location ?

[PuckPoltergeist]

Yeah, finally it works now. But I've still some kind of screen corruption. For example, there is something like a horizontal dash on top of the cursor. It isn't visible on the screenshots cause no cursor is shown there. Will try if I can take a screenshot, where this is visible.

With more then one tab in konqueror it is visible too:

http://www.stud.tu-ilmenau.de/~johi-in/Bildschirmfoto6.jpg

You can see it in the tab-bar.

Oh, its much more worse:

http://www.stud.tu-ilmenau.de/~johi-in/Bildschirmfoto7.jpg

[agd5f]

Make sure you're using the latest drm bits as well.

[highlandsun]

The order doesn't matter. The drm doesn't actually get initialized until X loads.

OK, my mistake. I had another older drm.ko/radeon.ko in /lib/modules/xx/updates so the ones I had just built and copies into xx/kernel/driver/gpu weren't actually loading.

I confirmed by reading the drm_init() code that order doesn't matter, and insmod'ing the freshly built files worked fine.

As a footnote, the radeon.ko module logs an init message displaying its version # when it's loaded/probed successfully, before the X server starts. I wasn't even seeing that before, with the old module. But it mostly seems to be working now.

In /var/log/Xorg.log I see a message that AIGLX couldn't init with DRI because of some other missing shared library. I guess there's more I need to rebuild on my Ubuntu 8.10 system. But now Xv is working, and mplayer is using a lot less CPU than before, which is all I really cared about.


(EE) AIGLX error: dlopen of /usr/lib/dri/r600_dri.so failed (/usr/lib/dri/r600_dri.so: cannot open shared object file: No such file or directory)
(EE) AIGLX: reverting to software rendering
(II) AIGLX: Loaded and initialized /usr/lib/dri/swrast_dri.so

[agd5f]

In /var/log/Xorg.log I see a message that AIGLX couldn't init with DRI because of some other missing shared library. I guess there's more I need to rebuild on my Ubuntu 8.10 system. But now Xv is working, and mplayer is using a lot less CPU than before, which is all I really cared about.


(EE) AIGLX error: dlopen of /usr/lib/dri/r600_dri.so failed (/usr/lib/dri/r600_dri.so: cannot open shared object file: No such file or directory)
(EE) AIGLX: reverting to software rendering
(II) AIGLX: Loaded and initialized /usr/lib/dri/swrast_dri.so


You can ignore that. There is no 3D driver yet for r6xx/r7xx.

[PuckPoltergeist]

Make sure you're using the latest drm bits as well.

I have done a git pull just before building the source, drm as well as radeonhd.

[schwarzygesetzlos]

With newest git-source I get following error when stopping the X-Server, e.g. rebooting the machine:
[drm] wait idle failed status: 0xA0003030 0x00000003
Which gets repeated very often during the whole shutdown-process.

[DanL]

For those having trouble getting the latest drm to work, make sure you're configuring/autogen.sh with --prefix=/usr

I found this guide helpful for building drm, though it's a bit outdated: http://wiki.x.org/wiki/radeonhd:DRI

[prakaushal]

Very very Thanx

[schwarzygesetzlos]

For those having trouble getting the latest drm to work, make sure you're configuring/autogen.sh with --prefix=/usr

I found this guide helpful for building drm, though it's a bit outdated: http://wiki.x.org/wiki/radeonhd:DRI

Do I need to do that if I merely need the kernel module living inside the linux-core sub-dir? Up to now compiling the module went fine if I did a make inside the linux-core sub-dir.

[schwarzygesetzlos]

With newest git-source I get following error when stopping the X-Server, e.g. rebooting the machine:
[drm] wait idle failed status: 0xA0003030 0x00000003
Which gets repeated very often during the whole shutdown-process.

Erm, dunno if the recent git-entry "r6xx/r7xx: re-init vm regs on resume, should fix resume" does have something to do with this error but it still persists. Killing X is no longer possible and after trying to do that VT output is veeery sloowly, e.g. a dmesg tages ages. The whole system seems to slow down afterwards...

[agd5f]

The bug is in radeonhd, not the drm; something in the tear down path. radeon works fine. I'll sort out radeonhd later tonight or tomorrow.

[schwarzygesetzlos]

The bug is in radeonhd, not the drm; something in the tear down path. radeon works fine. I'll sort out radeonhd later tonight or tomorrow.

Mmkay. Thanks for the info & the fast reply!

[agd5f]

I fixed it last night:
62f401aae135c0fd4fdc9536c11f5563080714c7

[schwarzygesetzlos]

I fixed it last night:
62f401aae135c0fd4fdc9536c11f5563080714c7

Yep, can confirm it's working! Thanks again maen!

[doddles]

Yep, can confirm it's working! Thanks again maen!
Yes! A big word of thanks from me too! My GA-MA78GM-S2H Radeon HD3200 is now giving me tear-free video in MythTV. It also seems to have fixed up a sound problem I'd been having when using the Catalyst drivers (is that possible?).

Anyway, interestingly, when I built the radeon driver, I couldn't get DRM to load. But switching to the radeonhd driver worked straight up. Don't know why the difference, but I'm happy it's working.

Are there any useful performance tests I could do or log/config files I could post here that might be useful for others with the same motherboard?

[algury]

Yes! A big word of thanks from me too! My GA-MA78GM-S2H Radeon HD3200 is now giving me tear-free video in MythTV. It also seems to have fixed up a sound problem I'd been having when using the Catalyst drivers (is that possible?).

Anyway, interestingly, when I built the radeon driver, I couldn't get DRM to load. But switching to the radeonhd driver worked straight up. Don't know why the difference, but I'm happy it's working.

Are there any useful performance tests I could do or log/config files I could post here that might be useful for others with the same motherboard?

Hi there,

I'm trying to install RadeonHD for my GA-MA78GM-S2H video chipset (Radeon HD3200) - so basically the same hardware as you - on Ubuntu 8.10:

1) I followed the wiki page:
http://www.x.org/wiki/radeonhd%3Ar6xx_r7xx_branch
(everything seems to be ok)

2) Driver is now installed (no screen corruption, better screen support than ATI's proprietary driver) but everything is very slow (e.g. windows are refreshing slowly when moved) - options "AccelMethod" "XAA" and "DRI" "on" are present in xorg.conf - xvinfo returns "no adaptors present"

So my question is: what did you do to make it work?

Thanks a lot,
Pierre

[algury]

Edit to my previous post:

I also found this in Xorg.0.log:
(WW) RADEONHD(0): DRI support has been disabled at compile time

I'm really not an expert but I guess it's related although I followed very closely the install wiki...

[algury]

BTW... here what I was referring to as my previous post :)

Yes! A big word of thanks from me too! My GA-MA78GM-S2H Radeon HD3200 is now giving me tear-free video in MythTV. It also seems to have fixed up a sound problem I'd been having when using the Catalyst drivers (is that possible?).

Anyway, interestingly, when I built the radeon driver, I couldn't get DRM to load. But switching to the radeonhd driver worked straight up. Don't know why the difference, but I'm happy it's working.

Are there any useful performance tests I could do or log/config files I could post here that might be useful for others with the same motherboard?

Hi there,

I'm trying to install RadeonHD for my GA-MA78GM-S2H video chipset (Radeon HD3200) - so basically the same hardware as you - on Ubuntu 8.10:

1) I followed the wiki page:
http://www.x.org/wiki/radeonhd%3Ar6xx_r7xx_branch
(everything seems to be ok)

2) Driver is now installed (no screen corruption, better screen support than ATI's proprietary driver) but everything is very slow (e.g. windows are refreshing slowly when moved) - options "AccelMethod" "XAA" and "DRI" "on" are present in xorg.conf - xvinfo returns "no adaptors present"

So my question is: what did you do to make it work?

Thanks a lot,
Pierre

[agd5f]

BTW... here what I was referring to as my previous post :)



Hi there,

I'm trying to install RadeonHD for my GA-MA78GM-S2H video chipset (Radeon HD3200) - so basically the same hardware as you - on Ubuntu 8.10:

1) I followed the wiki page:
http://www.x.org/wiki/radeonhd%3Ar6xx_r7xx_branch
(everything seems to be ok)

2) Driver is now installed (no screen corruption, better screen support than ATI's proprietary driver) but everything is very slow (e.g. windows are refreshing slowly when moved) - options "AccelMethod" "XAA" and "DRI" "on" are present in xorg.conf - xvinfo returns "no adaptors present"

So my question is: what did you do to make it work?

Thanks a lot,
Pierre

Make you installed the drm kernel modules correctly. Next, there is no XAA support for r6xx/r7xx, only EXA, so you want:
Option "AccelMethod" "EXA"
and DRI is required for EXA and Xv:
Option "DRI" "TRUE"

I also found this in Xorg.0.log:
(WW) RADEONHD(0): DRI support has been disabled at compile time

You also need to compile the driver with DRI support. you'll need the libdrm devel package installed.

[rahman.duran]

Edit to my previous post:

I also found this in Xorg.0.log:
(WW) RADEONHD(0): DRI support has been disabled at compile time

I'm really not an expert but I guess it's related although I followed very closely the install wiki...

I don't know if the wiki is up to date but you should use radeonhd master branch now, not r6xx-r7xx branch. And you still need to use r6xx-r7xx branch of drm. It is also important to overwrite existing radeon.ko and drm.ko. The wiki is generic the place to copy .ko files depends on your distro.

Btw, there is no XAA. You should use EXA instead.

Edit: Opps seems Alex faster than me :)

[algury]

Make you installed the drm kernel modules correctly. Next, there is no XAA support for r6xx/r7xx, only EXA, so you want:
Option "AccelMethod" "EXA"
and DRI is required for EXA and Xv:
Option "DRI" "TRUE"

You also need to compile the driver with DRI support. you'll need the libdrm devel package installed.

Sorry for the "AccelMethod" option, I knew that the right value was "EXA" -- this was just a little test.

About the need to compile the driver with DRI support, how do you specify the DRI support? is there an option?
(I've already installed libdrm-dev)
(I'm sorry if my questions are a bit naive but I'm very new to linux)

Thx
Pierre

[Edit]
after these commands:
cd xf86-video-radeonhd
./autogen.sh --prefix=/usr
I have the following output at the end of the process:
NOTE: DRI support is disabled

-> how to enable it then?

[algury]

Sorry for the "AccelMethod" option, I knew that the right value was "EXA" -- this was just a little test.

About the need to compile the driver with DRI support, how do you specify the DRI support? is there an option?
(I've already installed libdrm-dev)
(I'm sorry if my questions are a bit naive but I'm very new to linux)

Thx
Pierre

[Edit]
after these commands:
cd xf86-video-radeonhd
./autogen.sh --prefix=/usr
I have the following output at the end of the process:
NOTE: DRI support is disabled

-> how to enable it then?


Yes, I did it!

Actually, I was just missing some of the prerequisites packages.

So, to sum-up:
1) Hardware: Gigabyte GA-MA78GM-S2H (i.e. AMD780G chipset with Radeon HD3200 IGP), using the HDMI output with a Samsung LCD TV
2) OS: Ubuntu 8.10
3) I followed this wiki (just for the prerequisites packages section):
https://help.ubuntu.com/community/RadeonH
4) Then I followed this one (to build driver and drm, and set xorg.conf options):
http://www.x.org/wiki/radeonhd%3Ar6xx_r7xx_branch
5) After a restart, typing xvinfo return the detected adaptor (Adaptor #0: "RadeonHD Textured Video"), windows are no longer refreshed slowly... and last but not least, HD videos are tear-free even in 1920x1080 fullscreen

Two words: great job guys! (damn, that's three words)

It's almost easier to setup than the proprietary drivers: no black borders and no tearing in FullHD without searching for options.

Thanks for your replies, and keep up the good work! (I(ll probably test the 3D acceleration when it's ready!)
Cheers,
Pierre

PS: I'm very excited to be able to switch completely from Windows to Linux thanks to these drivers and because there are excellent media center softwares too, like XBMC which is very good looking.

[Boerkel]

Is it normal that i get a black screen, when i load a gnome-session with a dual-head configuration and radeonhd? If i try radeonhd + fluxbox everything works fine, but as soon as i try to use gnome the screen is black and i can't see anything. (but everything loads, because if can restart the xserver with alt+backspace)
Also if i just boot with one monitor everything works great.
Any ideas?

I'm using Ubuntu Hardy.

[bridgman]

I don't think this is normal. Can you pls pastebin an xorg log ?

[Boerkel]

I hope it isn't normal. ;)
Here is is the Xorg.0.log:
http://pastebin.com/m11fbbd9d

[agd5f]

Unless you posted the wrong log, you're running the vesa driver which doesn't support dualhead.

[Boerkel]

Ooops... mixed it up, sorry!
Here:
http://pastebin.com/m114a5e80