ATI radeon Sticky

Message

Grilo · Post by **Grilo** » Wed Aug 13, 2003 1:54 am

ok after a few trys 3.2.0 works with dri.. but i get a huge cut in glxgears with the beta3 kernel.. any ideas. . and yes i have all the tweaked lines working and the proper kernel settings.

Grilo

Admiral LSD · Post by **Admiral LSD** » Wed Aug 13, 2003 2:08 am

Dunno if this is the right place for this but is it just me or have the 3.2.4 ATi drivers, which are supposed to be in the Portage tree, become strangely scarce as of late? If they are available in Portage would someone be kind enough to give us a mirror that I can add to my make.conf to grab them?

dirtboy · Post by **dirtboy** » Wed Aug 13, 2003 6:21 am

Same here. For the past two days no mirror has had the file fglrx-glc22-4.3.0-3.2.4.i586.tar.gz.

Wedge_ · Post by **Wedge_** » Wed Aug 13, 2003 6:57 am

The link in the howto thread should still have them - http://odin.prohosting.com/wedge01/fglr ... 586.tar.gz

antipop · Post by **antipop** » Wed Aug 13, 2003 7:02 am

G-LiTe` wrote:
yardbird wrote:Just look at G-Lite posts. I don't know if he was able to solve the problem, though...
Nah no fix. We'll have to wait either for SiS to release some official drivers or some kernel hacker to... well.. start hacking.
I've been keeping an eye on it for months. It's going slow, let me tell you.
The chip involved is the SiS 5591/5592 btw, it's mentioned in the kernel config. There was a message somewhere in 2.6's changelogs that SiS 648 is supported. The question is, was thet the chip or the chipset? Because my chipset is SiS 648.
But as long as I have to use agp_try_unsupported=1 it'll probably stink.

Code: Select all

cat /proc/pci | grep AGP
    Host bridge: Silicon Integrated Systems [SiS] SiS645DX Host & Memory & AGP Controller (rev 0).
    PCI bridge: Silicon Integrated Systems [SiS] 5591/5592 AGP (rev 0).

Did you get past the blank screen in starting X?

G-LiTe` · Post by **G-LiTe`** » Wed Aug 13, 2003 7:42 am

antipop wrote:

Code: Select all

cat /proc/pci | grep AGP
    Host bridge: Silicon Integrated Systems [SiS] SiS645DX Host & Memory & AGP Controller (rev 0).
    PCI bridge: Silicon Integrated Systems [SiS] 5591/5592 AGP (rev 0).

Did you get past the blank screen in starting X?

X starts fine, it just crashes in some applications after a few seconds of 3D acceleration. (tuxracer and some screensavers do)
Try disabling dga by putting this:

Code: Select all

SubSection   "extmod"
     Option   "omit xfree86-dga"
EndSubSection

instead of:

Code: Select all

Load   "extmod"

in your XF86Config file.
Also make sure option UseInternalAGPGART is set to False.

yardbird · Post by **yardbird** » Wed Aug 13, 2003 9:39 am

fca wrote:It works! Now I have a beautiful KDE from CVS, with a nice 2.6.0test3 kernel beneath it, and 3D acceleration using the 3.2.4 drivers!

Now, only the framebuffer has to work with the new kernel, and I have everything working, but then bleeding edge.

Many thanks!

Nice to hear that

About the framebuffer: personally I don't use it, but you can try latest -mm sources (2.6.0-test3-mm2). In the changelog:

http://www.kernel.org/pub/linux/kernel/ ... nounce.txt

it is mentioned that framebuffer drivers got lots of fixes and updates.

antipop · Post by **antipop** » Wed Aug 13, 2003 1:58 pm

It seem like binary ati-drivers are working for me now.
I'm using 2.6.0-test3-mm2 kernel and compiled agpgart into kernel and SiS chipset support as a module.
At first I forgot to modprobe sis-agp, but even without doing that my screen has gone blank but now it just gave me an AGP error. I modprobed sis-agp and now it's working. I also set UseInternalAGPGART off from XF86Config.

Hopefully this will work for me from now on and I'll get to test the tv-out.

Edit: Is this behaviour normal:

Code: Select all

fglrxinfo
Error: unable to open display :0

I've done opengl-update ati, and I get ~2400 fps from glxgears and ~230 fps from fgl_glxgears.
I also have

Code: Select all

[fglrx:firegl_agp_lock_pages] *ERROR* agp_allocate_memory_phys_list failed

in my dmesg output.

G-LiTe` · Post by **G-LiTe`** » Wed Aug 13, 2003 2:32 pm

that's not right.. did you restart the x server after opengl-update?
Did you run dmesg to see if both sis-agp and fglrx are loaded properly? You might need to pass agp_try_unsupported=1 as an option to the sis-agp driver.
after all that check the output of "glxinfo" and tell us if direct rendering is turned on.

Wedge_ · Post by **Wedge_** » Wed Aug 13, 2003 2:37 pm

antipop wrote:I also have
Code: Select all
[fglrx:firegl_agp_lock_pages] *ERROR* agp_allocate_memory_phys_list failed
in my dmesg output.

You can probably ignore this one. I get sh*tloads of them every time I run fgl_glxgears, but it doesn't seem to affect anything. It happens with both 3.2.0 and 3.2.4.

Edit: I remember seeing a post somewhere on another forum that said this error could be fixed by changing your AGP aperture BIOS setting, but I've tried 64, 128, 256 and 512MB and it didn't help.

antipop · Post by **antipop** » Wed Aug 13, 2003 3:11 pm

G-LiTe` wrote:that's not right.. did you restart the x server after opengl-update?
Did you run dmesg to see if both sis-agp and fglrx are loaded properly? You might need to pass agp_try_unsupported=1 as an option to the sis-agp driver.
after all that check the output of "glxinfo" and tell us if direct rendering is turned on.

Now I re-ran opengl-update ati and restarted x server. Still same error.
sis-agp and fglrx are loaded properly.

Code: Select all

Aug 13 16:45:42 [kernel] fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel.
Aug 13 16:45:42 [kernel] [fglrx] Maximum main memory to use for locked dma buffers: 431 MBytes.
Aug 13 16:45:42 [kernel] [fglrx] module loaded - fglrx 3.2.4 [Jul 24 2003] on minor 0
Aug 13 16:45:42 [kernel] mtrr: 0xe0000000,0x4000000 overlaps existing 0xe0000000,0x1000000
Aug 13 16:45:44 [kernel] Console: switching to colour frame buffer device 160x64
                - Last output repeated twice -
Aug 13 16:46:59 [kernel] agpgart: Detected SiS 646 chipset
Aug 13 16:47:10 [kernel] Console: switching to colour frame buffer device 160x64
                - Last output repeated twice -
Aug 13 16:47:16 [kernel] mtrr: 0xe0000000,0x4000000 overlaps existing 0xe0000000,0x1000000
Aug 13 16:47:16 [kernel] [fglrx] AGP detected, AgpState   = 0x1f000201 (hardware caps of chipset)
Aug 13 16:47:16 [kernel] agpgart: Found an AGP 2.0 compliant device at 0000:00:00.0.
Aug 13 16:47:16 [kernel] [fglrx] free  AGP = 54800384
Aug 13 16:51:35 [kernel] [fglrx:firegl_agp_lock_pages] *ERROR* agp_allocate_memory_phys_list failed
                - Last output repeated 7697 times -
Aug 13 18:04:13 [kernel] mtrr: no MTRR for e0000000,400000 found
Aug 13 18:04:14 [kernel] Console: switching to colour frame buffer device 160x64
                - Last output repeated twice -
Aug 13 18:04:20 [kernel] mtrr: 0xe0000000,0x4000000 overlaps existing 0xe0000000,0x1000000
Aug 13 18:04:20 [kernel] [fglrx] free  AGP = 54800384

glxinfo says that Direct Rendering is enabled.
Is it normal that running glxinfo takes ~6 seconds to give any output?

Code: Select all

 $ glxinfo
name of display: :1.0
display: :1  screen: 0
direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.2
server glx extensions:
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_EXT_import_context
client glx vendor string: SGI
client glx version string: 1.2
client glx extensions:
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_EXT_import_context, 
    GLX_ARB_get_proc_address, GLX_ARB_multisample, GLX_ATI_pixel_format_float, 
    GLX_ATI_render_texture
GLX extensions:
    GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_EXT_import_context
OpenGL vendor string: ATI Technologies Inc.
OpenGL renderer string: Radeon 8500 DDR Pentium 4 (SSE2)
OpenGL version string: 1.3 (X4.3.0-3.2.4)

Wedge_ · Post by **Wedge_** » Wed Aug 13, 2003 3:50 pm

antipop:

bda wrote:when trying to load fglrx, disable mtrr for your framebuffer and/or make sure radeonfb is not compiled into the kernel. I had vesafb AND radeonfb compiled in (which is a bad idea - there should only be one fb driver), and was using vesafb with mtrr.

bda posted that a couple of pages back - he was getting the same mtrr overlap messages that you are. Try doing what he says and see if that helps. Is that log section just from loading the module? I don't get any of the agp_lock_pages errors until I run fgl_glxgears.

antipop wrote:glxinfo says that Direct Rendering is enabled.
Is it normal that running glxinfo takes ~6 seconds to give any output?

The output looks fine, but it shouldn't take any time to run, so it looks like something is still not working.

bda · Post by **bda** » Thu Aug 14, 2003 2:30 pm

bda wrote:when trying to load fglrx, disable mtrr for your framebuffer and/or make sure radeonfb is not compiled into the kernel. I had vesafb AND radeonfb compiled in (which is a bad idea - there should only be one fb driver), and was using vesafb with mtrr.

Actually, I discovered that the only way to get rid of the mtrr overlap message was to disable the frame buffer completely (pass vga=normal to the kernel). Even when using vesafb WITHOUT the mtrr option, I would get that error message.

I did some testing, and I'm pretty sure it doesn't affect performance, so it's probably save to ignore those messages. Note that I'm using 3.2.0 - it might behave differently in 3.2.4.

Guess I will add this to the HOWTO...

bushwakko · Post by **bushwakko** » Tue Aug 19, 2003 1:41 am

My X starts and all but glxinfo says no DRI and I get this in my log:

[/code]
(II) fglrx(0): driver needs XFree86 version: 4.3.x
(II) fglrx(0): detected XFree86 version: 4.3.0
(II) Loading extension ATIFGLRXDRI
(II) fglrx(0): doing DRIScreenInit
drmOpenDevice: minor is 0
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 6, (OK)
drmOpenDevice: minor is 0
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 6, (OK)
drmOpenDevice: minor is 0
drmOpenDevice: node name is /dev/dri/card0
drmOpenDevice: open result is 6, (OK)
drmOpenDevice: minor is 1
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 result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 2
drmOpenDevice: node name is /dev/dri/card2
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 3
drmOpenDevice: node name is /dev/dri/card3
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 4
drmOpenDevice: node name is /dev/dri/card4
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 5
drmOpenDevice: node name is /dev/dri/card5
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 6
drmOpenDevice: node name is /dev/dri/card6
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 7
drmOpenDevice: node name is /dev/dri/card7
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 8
drmOpenDevice: node name is /dev/dri/card8
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 9
drmOpenDevice: node name is /dev/dri/card9
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 10
drmOpenDevice: node name is /dev/dri/card10
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 11
drmOpenDevice: node name is /dev/dri/card11
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 12
drmOpenDevice: node name is /dev/dri/card12
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 13
drmOpenDevice: node name is /dev/dri/card13
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: open result is -1, (No such device)
drmOpenDevice: Open failed
drmOpenDevice: minor is 14
drmOpenDevice: node name is /dev/dri/card14
drmOpenDevice: open result isdrmOpenDevice: 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
(II) fglrx(0): [drm] drmOpen failed
(EE) fglrx(0): DRIScreenInit failed!
(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): ********************************************* *
(II) fglrx(0): FBADPhys: 0xf0000000 FBMappedSize: 0x04000000
(WW) fglrx(0): Failed to set up write-combining range (0xf0000000,0x4000000)
(II) fglrx(0): FBMM initialized for area (0,0)-(1408,8191)
(II) fglrx(0): FBMM auto alloc for area (0,0)-(1408,1050) (front color buffer - assumption)
(==) fglrx(0): Backing store disabled
(==) fglrx(0): Silken mouse enabled
(II) fglrx(0): Using hardware cursor (scanline 1050)
(II) fglrx(0): Largest offscreen area available: 1408 x 7138
(**) Option "dpms"
(**) fglrx(0): DPMS enabled
(II) fglrx(0): Using XFree86 Acceleration Architecture (XAA)
Screen to screen bit blits
Solid filled rectangles
Solid Horizontal and Vertical Lines
Offscreen Pixmaps
Setting up tile and stipple cache:
32 128x128 slots
32 256x256 slots
16 512x512 slots
(II) fglrx(0): Acceleration enabled
(II) fglrx(0): Direct rendering disabled
(II) Loading extension FGLRXEXTENSION
(II) Loading extension ATITVOUT
(==) RandR enabled
(II) Setting vga for screen 0.
-1, (No such device)

Code: Select all


I'm using 2.6-test3 and the newest ati drivers: media-video/ati-drivers-3.2.4 

I have compiled in support for agpgart and my intel xhipset and my chipset is M9 (it's worked before, and everything but DRI seems to).

any ideas?

Wedge_ · Post by **Wedge_** » Tue Aug 19, 2003 7:20 am

Have you set UseInternalAGPGART to no? It's enabled if you pick the default settings in fglrxconfig, and it won't use the kernel AGP support that way. Also check you don't have any DRM settings enabled in your kernel.

Rüpel · Post by **Rüpel** » Tue Aug 19, 2003 12:04 pm

i couldn't get my radeon 9800 pro to work on my via kt400 board yesterday. i have installed the current 2.4.20 gentoo kernel and emerged the 3.2.0 drivers. does this kernel support the via kt400?

the driver-installation itself was easy following the guide listed somewhere here. i also set the kernel-config as suggested (mtrr=y, agpgart=m, viaagp=y, dri=n)- but X gave me

Code: Select all

(EE) fglrx(0): [agp] unable to acquire AGP, error "xf86_ENODEV"
(EE) fglrx(0): cannot init AGP

which some people in this thread already experienced. i guess something about my kt400 and agp is wrong. i tried to use ati's internal agpgart. does this mean, that ati's internal agpgart doesn't support via's kt400?

no agp-related modules were loaded. fglrx is loaded after startx quit.

what's the best way to proceed now? i'll try loading via-agp and use the external agpgart first. if that doesn't work, i should go with a 2.6 kernel, shouldn't i? are they stable enough to work the next months? i want to setup my system now and have it running stable. i don't want to configure a new kernel once a week.

any tips and hints appreciated.

i think, i'll post again this evening, when i'm at home and have tried a few things...

odd thing is: i bought a kt400 board instead of an nforce2 because someone told me, that via-chipset-support is better than nvidias under linux...

Kihaji · Post by **Kihaji** » Tue Aug 19, 2003 12:26 pm

Rüpel wrote:i couldn't get my radeon 9800 pro to work on my via kt400 board yesterday. i have installed the current 2.4.20 gentoo kernel and emerged the 3.2.0 drivers. does this kernel support the via kt400?

the driver-installation itself was easy following the guide listed somewhere here. i also set the kernel-config as suggested (mtrr=y, agpgart=m, viaagp=y, dri=n)- but X gave me
Code: Select all
(EE) fglrx(0): [agp] unable to acquire AGP, error "xf86_ENODEV"
(EE) fglrx(0): cannot init AGP
which some people in this thread already experienced. i guess something about my kt400 and agp is wrong. i tried to use ati's internal agpgart. does this mean, that ati's internal agpgart doesn't support via's kt400?

no agp-related modules were loaded. fglrx is loaded after startx quit.

what's the best way to proceed now? i'll try loading via-agp and use the external agpgart first. if that doesn't work, i should go with a 2.6 kernel, shouldn't i? are they stable enough to work the next months? i want to setup my system now and have it running stable. i don't want to configure a new kernel once a week.

any tips and hints appreciated.

i think, i'll post again this evening, when i'm at home and have tried a few things...

odd thing is: i bought a kt400 board instead of an nforce2 because someone told me, that via-chipset-support is better than nvidias under linux...

2.4 kernel does not work with AGP3.0 (8x), which is what your card is running at. Use the latest 2.6 kernel or try your luck with the 2.5 line. I have a KT400 board and just now got my 9700pro to run great with 2.6 test 3.

Wedge_ · Post by **Wedge_** » Tue Aug 19, 2003 12:27 pm

i couldn't get my radeon 9800 pro to work on my via kt400 board yesterday. i have installed the current 2.4.20 gentoo kernel and emerged the 3.2.0 drivers. does this kernel support the via kt400?
...
what's the best way to proceed now? i'll try loading via-agp and use the external agpgart first. if that doesn't work, i should go with a 2.6 kernel, shouldn't i? are they stable enough to work the next months? i want to setup my system now and have it running stable. i don't want to configure a new kernel once a week.

There are people that have gotten KT400 AGP working, but as far as I know, only with 2.6 kernels and at least the 3.2.0 drivers (you might need to use 3.2.4, I don't know). There are patches floating around for 2.4 to add KT400 AGP support, but a lot of people seem to have problems getting them to work. If you just emerge one of the beta kernels, use the same AGP settings as before, emerge the 3.2.0 or 3.2.4 drivers again, and set UseInternalAGPGART to "no", then it should (:)) work for you. As for the 2.6 kernels, I've been running mm-sources for weeks and haven't had any major problems. They seem stable enough, and the increased responsiveness is also rather nice

G-LiTe` · Post by **G-LiTe`** » Tue Aug 19, 2003 1:59 pm

Holy shiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiit!!!! It works! no crashes! w000T!
I recommend everyone with a SiS 5591/5592 agp bridge to try the mm sources. 2.6.0test3-mm3 solved my problems, tuxracer works, ut2003 works. EVERYTHING WORKS. WOOHOO!

I haven't been able to test it with mm2 or before because they both crashed my system.
I wouldn't know about 2.4 mm either. But it's definitly worth checking out. \o/

Wedge_ · Post by **Wedge_** » Tue Aug 19, 2003 2:22 pm

Holy shiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiit!!!! It works! no crashes! w000T!
I recommend everyone with a SiS 5591/5592 agp bridge to try the mm sources. 2.6.0test3-mm3 solved my problems, tuxracer works, ut2003 works. EVERYTHING WORKS. WOOHOO!

I can see you're happy about it

I didn't even know -mm3 was out yet.

Rüpel · Post by **Rüpel** » Tue Aug 19, 2003 5:26 pm

maybe slightly off-topic, but anyways.

when i build that kernel (mm2.6.0test3-rc2 from 08/16/2003) it builds the bzImage fine (although there are some strange errors while booting) but gives a sh*tload of undefined symbols when building the modules. i have alsa and agpgart as modules and each and every file related to these "packages" seem to have no access to external symbols. is it me?

Wedge_ · Post by **Wedge_** » Tue Aug 19, 2003 5:57 pm

How did you build it? For the 2.6 kernels, you just need to do

Code: Select all

make && make modules_install

without all the make clean dep etc that the 2.4 kernels need.

Rüpel · Post by **Rüpel** » Tue Aug 19, 2003 8:24 pm

uuhhh. well, that should help, lets see...

NuclearFusi0n · Post by **NuclearFusi0n** » Tue Aug 19, 2003 8:46 pm

Rüpel wrote:maybe slightly off-topic, but anyways.

when i build that kernel (mm2.6.0test3-rc2 from 08/16/2003) it builds the bzImage fine (although there are some strange errors while booting) but gives a sh*tload of undefined symbols when building the modules. i have alsa and agpgart as modules and each and every file related to these "packages" seem to have no access to external symbols. is it me?

emerge module-init-tools if you haven't already.

Rüpel · Post by **Rüpel** » Tue Aug 19, 2003 8:55 pm

NuclearFusi0n wrote:

emerge module-init-tools if you haven't already.

no, i haven't. thx. will try that. (since the make && make modules_install didn't solve the problem)

btw: funny to post here from lynx...