Congrats waever!! -- Another ATI card comes out swinging on Linux!!


-- i am usually pretty persistent, but i think i am giving up on this one!

It's your choice. I do like the great feeling I get when running Tux Racer downhill with this card and linux. That is the extent of my gaming. The desktop is what I enjoy with this card. Nvidia makes great drivers. But my Geforce 4 card is in my gateway server for a reason.

Waever was someone else we were trying to help. Glad one more is working.

synaptical, keep poking at it... It will work. Not sure what the issue is with the boot loader, but I've not heard of kernel compiling having anything to do with the issue you are describing.

cd /usr/src/linux
make menuconfig
make bzImage && make modules && make modules_install
mv /boot/System.map /boot/system.map.old
cp System.map /boot/
mv /boot/vmlinuz /boot/vmlinuz.old
cd arch/i386/boot
cp bzImage /boot/vmlinuz
lilo -v
Reboot

or upgrade from whatever to whatever...

cd /usr/src
rm -f linux
mv linux-2.6.x linux-2.6.x.old
cd /lib/modules
mv 2.6.x 2.6.x.old
cd /boot
rm -f System.map
mv System.map-ide-2.6.x System.map
cp System.map System.map.old
cp vmlinuz vmlinuz.old

vi /etc/lilo.conf
image = /boot/vmlinuz
root = /dev/hda6
label = Linux
read-only
# --add this --
image = /boot/vmlinuz.old
root = /dev/hda6
label = Linux-Old
read-only

:wq

cd /usr/src
tar xjvf linux-2.6.x.tar.bz2
ln -s linux-2.6.x linux
cd linux
make menuconfig
make bzImage && make modules && make modules_install
mv /boot/System.map /boot/system.map.old
cp System.map /boot/
mv /boot/vmlinuz /boot/vmlinuz.old
cd arch/i386/boot
cp bzImage /boot/vmlinuz
lilo -v
reboot
============


Hooper

hmmm.... I did a lot recompiling and lilo-ing, but I never messed up the MBR.. I don't know, maybe a typo could cause it? but I think lilo is good enough not to cause that error...
synaptical, I also had that thought for a while.. just few days ago I was thinking about selling my Radeon and get an nvidia card. but also on that day I saw the thread about 2.6.1 released and tried it right away. I believe you have the same problem as mine, so give it a shot..

waever...now that is good to hear

synaptical...sorry to hear your frustrations. I don't think you need the vmaddy patch though. Only the AMD patch. As waever stated abouve, he had almost identical errors in his output, and 2.6.1, 3.7.0 drivers, and the AMD patch seemed to have worked for him. As far as the mesa drivers..I don't know how to cleanly remove tham short of creating some sort of link pointing away from them, or extracting the glx.h files from XFree source to overwrite the mesa ones...but then I am not 100% clear on this right now either.

Either way, if I can think of anything else, I'll post it. For now though, I would agree that fixing the Lilo/boot issue is a matter of utmost importance.

congrats, weaver! tell me, what's your secret?

i recompiled with agp built in to see if i could get rid of those dmesg errors under the agpgart, and it seemed to work:

agpgart: Detected NVIDIA nForce2 chipset
agpgart: Maximum main memory to use for agp memory: 439M
agpgart: AGP aperture is 64M @ 0xe8000000

but now when booting i get a bunch of FATAL: module nvidia_agp already in kernel errors. i thought the nvidia option needed to be included as a module anyway? can i take that out of the kernel altogether?

i've never had it before, but it just started showing up for some reason, every time i run lilo.

thanks for the encouragement, i will probably keep trying until i run out of configurations options, at least. i love my radeon card, it has great IQ, and i really want to be getting the full potential out of it. Hooper, i can relate totally! it bugs me to no end to know that something isn't 100% functional and optimized! (that's also why i love slack ) just getting a little discouraged here and there -- please ignore.

FATAL: module nvidia_agp already in kernel errors.

These errors are usually due to a module trying to load with this compiled directly into the kernel, do you have this loading via /etc/rc.d/rc.modules?
If so you could try remming it.

Also do a lsmod and see what is loaded already into the kernel.

root@machine:/# lsmod
Module Size Used by
ip_conntrack_ftp 71444 0
ip_conntrack_irc 70676 0
ipt_LOG 5376 5
ipt_limit 2304 5
ipt_REJECT 6016 4
ipt_state 1792 32
iptable_filter 2432 1
ip_conntrack 28832 3 ip_conntrack_ftp,ip_conntrack_irc,ipt_state
ip_tables 16784 5 ipt_LOG,ipt_limit,ipt_REJECT,ipt_state,iptable_filter
snd_pcm_oss 48260 1
snd_mixer_oss 17152 2 snd_pcm_oss
fglrx 203084 183
root@machine

Note the last one.

Exactly...the error itself does no harm. Basically, it's telling you that whatever is calling to load the module, resulted in a "fatal" modprobe because the module is already there. It doesn't hurt anything...just looks ugly.

As far as getting it to work...good job.

no, that's what's confusing. it was built as a module, and so it should load as one, right? can i take it out, or do i absolutely need it?

still no luck:
display: :0.0 screen: 0
OpenGL vendor string: Mesa project: www.mesa3d.org
OpenGL renderer string: Mesa GLX Indirect
OpenGL version string: 1.3 Mesa 4.0.4

otherwise, everything seems to be "go." i don't think anything else is going to matter until i find out how to disable that mesa crap.

Give me an lsmod output...

May sound stupid...but you are re-installing the ATI drivers after every kernel compile...correct? And you are getting no errors during this?

lsmod (agpgart built in)
Module Size Used by
fglrx 210060 0
cs46xx 68940 1
ac97_codec 19276 1 cs46xx


yep, reinstalling the drivers every time. i've gotten so paranoid i'm even removing the entire directory before recompiling the kernel, and then re-untarring, repatching, and then reinstalling them every time. i could do the whole thing in my sleep by now, it's not that bad, really, just more of a nuisance at this point.

If built as a module, you need to load with modprobe via rc.modules. If compiled directly into the kernel, you need to rem on the rc.modules. So being you have it compiled as a module, you need to keep within rc.modules startup.

Mesa crap will leave when you get your fglrx loaded properly.

Something doesn't look right. When you load the agpgart module, your chipset module should load with it. You have agpgart built in...yet your chipset module doesn't load and fglrx does? When you ran fglrxconfig, you chose to use the external agpgart?

The other thing that gets me is this...I have NOT added any entries in rc.modules or modules.conf. After I log in, lsmod doesn't show agpgart, chipset-agp, or fglrx. Once I startx, the modules are loaded.

Ok... Here is my new 2.6.1 compile configuration. Take a look and see if you have something similar.

rjs@machine:~$ uname -a
Linux machine 2.6.1 #1 SMP Sun Jan 11 00:43:52 EST 2004

==Processor==[*] MTRR (Memory Type Range Register) support

==Character Devices==
<*> /dev/agpgart (AGP Support)
<M> ATI chipset support
<*> Intel 440LX/BX/GX, I8xx and E7x05 chipset support
[ ] Direct Rendering Manager (XFree86 4.1.0 and higher DRI support)

===Graphics Support====[*] Support for frame buffer devices [*] VESA VGA graphics support
<M> ATI Radeon display support

Remember this is for an Intel.

=============
rjs@machine:~$ fglrxinfo
display: :0.0 screen: 0
OpenGL vendor string: ATI Technologies Inc.
OpenGL renderer string: RADEON 9700 PRO x86/SSE2
OpenGL version string: 1.3 (X4.3.0-3.7.0)
=============
rjs@machine:~$ glxinfo |more |grep ATI
client glx vendor string: ATI
GLX_ARB_get_proc_address, GLX_ARB_multisample, GLX_ATI_pixel_format_float,
GLX_ATI_render_texture
OpenGL vendor string: ATI Technologies Inc.
GL_ARB_window_pos, GL_ATI_draw_buffers, GL_ATI_element_array,
GL_ATI_envmap_bumpmap, GL_ATI_fragment_shader, GL_ATI_map_object_buffer,
GL_ATI_separate_stencil, GL_ATI_texture_env_combine3,
GL_ATI_texture_float, GL_ATI_texture_mirror_once,
GL_ATI_vertex_array_object, GL_ATI_vertex_attrib_array_object,
GL_ATI_vertex_streams, GL_ATIX_texture_env_combine3,
GL_ATIX_texture_env_route, GL_ATIX_vertex_shader_output_point_size,
==============
rjs@machine:~$ glxgears
15786 frames in 5.0 seconds = 3157.200 FPS
19759 frames in 5.0 seconds = 3951.800 FPS
19620 frames in 5.0 seconds = 3924.000 FPS
25079 frames in 5.0 seconds = 5015.800 FPS
25740 frames in 5.0 seconds = 5148.000 FPS
25800 frames in 5.0 seconds = 5160.000 FPS
25734 frames in 5.0 seconds = 5146.800 FPS
25759 frames in 5.0 seconds = 5151.800 FPS
25729 frames in 5.0 seconds = 5145.800 FPS
25704 frames in 5.0 seconds = 5140.800 FPS
25768 frames in 5.0 seconds = 5153.600 FPS
25758 frames in 5.0 seconds = 5151.600 FPS
================

And for comparison...also for Intel...

# Loadable module support
#
CONFIG_MODULES=y
CONFIG_MODULE_UNLOAD=y
# CONFIG_MODULE_FORCE_UNLOAD is not set
CONFIG_OBSOLETE_MODPARM=y
# CONFIG_MODVERSIONS is not set
CONFIG_KMOD=y

# Processor type and features

# CONFIG_HIGHMEM64G is not set
CONFIG_HIGHMEM=y
# CONFIG_HIGHPTE is not set
CONFIG_MTRR=y


==Character Devices==

<M> /dev/agpgart (AGP Support)
<M> Intel 440LX/BX/GX, I8xx and E7x05 chipset support
[ ] Direct Rendering Manager (XFree86 4.1.0 and higher DRI support)


# Graphics support
#
CONFIG_FB=y
# CONFIG_FB_CYBER2000 is not set
# CONFIG_FB_IMSTT is not set
CONFIG_FB_VGA16=m
CONFIG_FB_VESA=y
CONFIG_VIDEO_SELECT=y
CONFIG_FB_RADEON=m
CONFIG_FB_VIRTUAL=m

# Console display driver support
#
CONFIG_VGA_CONSOLE=y
# CONFIG_MDA_CONSOLE is not set
CONFIG_DUMMY_CONSOLE=y
CONFIG_FRAMEBUFFER_CONSOLE=y
CONFIG_PCI_CONSOLE=y
# CONFIG_FONTS is not set
CONFIG_FONT_8x8=y
CONFIG_FONT_8x16=y