Official FreeBSD drivers

[xpto]

Why this driver doesn't support riva128?

riva as in ati ?

try the dri drivers dri.sf.net

[darius]

I would say he means nVidia Riva128.

Note that the Windows drivers for Rivas are different to TNT/GeForce (NV4+) so it's not very suprising.

The XFree86 drivers support Rivas, doubt they do 3d though.

[errodr]

I would strongly recommend upgrading your ports tree with cvsup (see the handbook) and using the port
cd /usr/ports/x11/nvidia-driver
make install clean

After that you just have to edit your XF86Config file to use the nvidia driver (vs the nv one) and you should be set.

Thanks, but I still get the same error when I try to compile the src.

Any ideas??

[darius]

Thanks, but I still get the same error when I try to compile the src.

Any ideas??

Can you paste the last 20 lines of output when you are building the port please?

[nullpt]

Hello,

I'm having some issues with the nvidia driver for freebsd. I install the driver trough the ports system .

phobos# cd /usr/ports/x11/nvidia-driver/
phobos# make install

Eveything install fine except the make setup that seems to be broken.

phobos# pkg_info | grep nvidia
nvidia-driver-1.0.4365_2 NVidia graphics card binary drivers for hardware OpenGL ren
phobos# cd work/NVIDIA-FreeBSD-x86-1.0-4365
phobos# make setup
===> module
"Makefile", line 17: Malformed conditional (${OSVERSION} > 500011)
"Makefile", line 17: Need an operator
"Makefile", line 19: if-less endif
"Makefile", line 19: Need an operator
make: fatal errors encountered -- cannot continue
*** Error code 1

Stop in /usr/ports/x11/nvidia-driver/work/NVIDIA-FreeBSD-x86-1.0-4365.

Bu the make setup is option i think.

phobos# kldload nvidia
Jul 18 07:03:17 phobos kernel: nvidia0: <GeForce2 MX/MX 400> mem 0xd0000000-0xd7ffffff,0xde000000-0xdeffffff irq 4 at device 0.0 on pci1
# kldstat
Id Refs Address Size Name
1 3 0xc0400000 65a558 kernel
2 1 0xc4711000 187000 nvidia.ko
phobos# kldunload nvidia
Jul 18 07:04:09 phobos kernel: nvidia0: detached
phobos# kldstat
Id Refs Address Size Name
1 1 0xc0400000 65a558 kernel
phobos#

All fine till here (except for the make setup).
I setup X and gnome and it works fine with the default nv driver. When I switch to the nvidia driver my XFreeConfig is equal to the default one, I only change the nv driver to nvidia and i take out dri, record and xtrap from the Secton "Module".
Here is the sections:
Section "Module"
Load "dbe"
Load "extmod"
Load "glx"
Load "speedo"
Load "freetype"
Load "type1"
EndSection

Section "Device"
Option "NvAGP" "1"
Identifier "Card0"
Driver "nvidia"
VendorName "nVidia Corporation"
BoardName "NV11 [GeForce2 MX/MX 400]"
BusID "PCI:1:0:0"
EndSection

The rest of the config is the default.
When I startx I don't see any output it just leads me to a screen full of crazy characters and symbols and I can't get out (Ctrl^c, ctrl+alt+backspace) until it shuts down the machine. It seems like there is some sort of problem somewhere and I can't find it. I had this machine as my freebsd desktop with freebsd 5.2.1-releasep9 for almost a month and it worked just fine with nvidia driver but then i had to reinstall everything coz i only had 6GB space for freebsd and I installed everything in a new disk. I have more than one freebsd desktop and they all run with nvidia driver, never had a problem and now this is bugging me like hell!

If anyone knows anything on this please post.

System info:

phobos# uname -a
FreeBSD phobos 5.2.1-RELEASE-p9 FreeBSD 5.2.1-RELEASE-p9 #0: Sun Jul 18 05:43:12 GMT 2004 alex@phobos:/usr/obj/usr/src/sys/phobos i386
phobos#

nvidia driver:

nvidia-driver-1.0.4365_2 NVidia graphics card binary drivers for hardware OpenGL ren

Kernel Options:

device pcm
options COMPAT_LINUX
device agp

(I tried without this options as well)

BIOS settings:

PnP-OS Aware: NO
Primary Graphic Adapter: PCI
Allocate IRQ to PCI VGA: yes

dmesg errors:

(dont know if this is an error)
pci0: <PCI bus> on pcib0
pci_cfgintr: 0:2 INTD BIOS irq 11
pci_cfgintr: 0:2 INTA BIOS irq 5
pci_cfgintr: 0:2 INTC BIOS irq 11
pci_cfgintr: 0:2 INTC BIOS irq 11
pci_cfgintr: 0:3 INTA BIOS irq 5
pci_cfgintr: 0:9 INTA BIOS irq 5

AND

vga0: <Generic ISA VGA> at port 0x3c0-0x3df iomem 0xa0000-0xbffff on isa0
unknown: <PNP0303> can't assign resources (port)
unknown: <PNP0501> can't assign resources (port)
unknown: <PNP0501> can't assign resources (port)
unknown: <PNP0401> can't assign resources (port)
unknown: <PNP0700> can't assign resources (port)
unknown: <PNP0f13> can't assign resources (irq)

Graphic card:

AGP Geforce MX 400 64MB SDRAM
It has an onboard card but it's not being used.

Thanks for your help and thanks for reading my thread!

cheers

PS: For the sake of hard trying I edited the Makefile from the modules area and fixed the condition. But the problem still persists!

[zander]

Try booting with ACPI disabled and/or change the AGP GART driver.

[nullpt]

Hello, I have agp compiled into my kernel. I tried to use the system AGP:
Option "NvAGP" "3"

but i went to the same problem.
I always boot without ACPI

phobos# cat /boot/loader.conf
hint.acpi.0.disabled=1
nvidia_load="YES"
phobos#

But I will try it again!
Another thing that is messing with me is that in the nvidia-freebsd documentation says that USER_LDT is needed and it is not default in FreeBSD 5.x releases and my freebsd kernel config utility says that is an invalid kernel option. Anyone shed a light on this?

cheers

[nullpt]

Hello,

I just compiled a new kernel with only one modification:

device pcm # sound device support

I compiled the nvidia-driver like this:
# cd /usr/ports/x11/nvidia-driver/
# make -DWITH_FREEBSD_AGP install

It compiled ok.
I added to my loader.conf:

hint.acpi.0.disabled=1
linux_load="YES"
agp_load="YES"
nvidia_load="YES"

and I'm getting the same troubles. I will post here my log files:

http://sublimesp.net/~alex/dmesg.txt
http://sublimesp.net/~alex/XF86Config
http://sublimesp.net/~alex/nvidia.output.installation.txt

Thanks for your help.

PS: Is USER_LDT enabled for default in the 5.2.1-RELEASE?
PS2: My card is AGP but In my XF86Config it says that the card is at BusID PCI, my nvidia0 was created at pc1. I know I have Primary graphic adapter set to PCI at the BIOS and that the Allocate IRC to PCI VGA is enabled in my BIOS. Does it anything to do with that?

thanks for your help!
cheers

[darius]

nullpt - 5.x has USER_LDT "on all the time" - it's nolonger an option.

AGP cards look a LOT like PCI cards so, yes, nvidia0 will attach to a PCI bus.

And.. What troubles are you getting?

[nullpt]

I have some freebsd desktops and I built them all ith nvidia driver ok.
I just started building this one and when I install and configure X with the driver it gives me a screen with lots of crazy characters and i cannot escape from this, only with reboot.

thanks!

PS: This machine was working with nvidia driver until some days ago when i had to reinstall everything due to a hard disk upgrade, the one i had was too small. From 1 month till now did the developers released some new driver with some changes? The one I have asap is nvidia-driver-1.0.4365_2 .

cheers

[zander]

The current release still is 1.0-4365. You should try disabling AGP to see if it's the source of your problem and verify that the NVIDIA driver sees the card's interrupt requests (I'm guessing the last thing you see in the XFree86 log file is "(II) NVIDIA(0): Setting mode ...").

[nullpt]

Yes I have that release installed.
During the night i rebuilt the world and the kernel and reinstall XFree86-4 and nvidia driver. Didn't worked.
I tried to disable system AGP byt puttyng agp_load="NO" in loader.conf and I am analising the logs asap.

Will post soon.

[zander]

You disabled FreeBSD's AGP GART driver; if you want to use NvAGP, you'll also need to rebuild the NVIDIA kernel module with support for it. In order to disable AGP support, you'll need to set the "NvAGP" option to "0" in your XFree86 configuration file.

[nullpt]

Well I compiled the nvidia driver without support for the FREEBSD_AGP gart. I installed it normally and I have been using "NvAGP" "1" that stands for the nvidia driver AGP gart.
But I will try to set it to 0 and see what happens. Thanks for the help.
PS: My XFree86.0.log and :0.log.4 is attached.

[nullpt]

Same thing happens with NvAGP 0.

Thanks for your help.

[nullpt]

Hello folks,

Thanks for the help on this issue. Unfortunatly i didn't worked it out.
I tried to reinstall it and did everything by the book and I still got the same problem.

If any developer or experienced person reads this please try to answer to this question:

I install X and the nvidia driver correctly by the book but when i load the driver nvidia on the X server it appears a screen full of carachters with no sense and I can only stop it with a cold reboot?

I think it's something related with some BIOS settings or something, so I leave you the details here:

PNP-OS Aware: OFF
Primary graphics adapter: PCI
Allocate IRQ TO PCI VGA: YES

Thanks once again for your time,
Cheers

[zander]

A black or garbled screen can be indicative of numerous problems, unfortunately; from your description, I suspect the NVIDIA card either doesn't generate interrupts or they don't reach the driver - in that case X gets stuck in the middle of the initial mode set, but the system remains intact, i.e. you should still be able to e.g. log in remotely. Looking at your dmesg log, it seems that the NVIDIA card is assigned IRQ 4 and that the first serial port is enabled; this combination cannot work, which is reflected by sio0's confusion. I guess you could try disabling the serial port (assuming you don't need it) and/or check your BIOS settings.

[nullpt]

Yes I have that enabled indeed. I will check that right away. Thanks a lot!

[nullpt]

Hello,

It was exactly that! Thanks a lot zander. I have no problem now.

Cheers

[zander]

No problem, I'm glad I could help.

[hishnik]

I bought nVidia card only because there are official drivers for FreeBSD. Hope it will work on my ASUS A8V.

[martin71]

does anyone know, if the freebsd driver supports the "low voltage" (powermizer mode for geforce 4 on laptop?

[zipster]

Anyone out there try to build the 1.0-4365 release of the nVidia driver for a FreeBSD 5.3-RELEASE system ?

I have a Toshiba Satellite Pro 6100 with a 1600x1200 TFT display, and the (current) 1.0-6113 release does not allow
the full 1600x1200 resolution -- rather, it simply drops the the right 12-pixel edge of the display.

The 1.0-4365 release of the driver did not do this under 5.3.1-RELEASE, so I've been having a look at building the kernel module for the driver under 5.3-RELEASE (and gcc-3.4.2).

When I try to make the kernel module, the make fails with the following errors:

nvidia_ctl.c:61: error: conflicting types for 'nvidia_ctl_open'
nvidia_ctl.c:16: error: previous declaration of 'nvidia_ctl_open' was here
nvidia_ctl.c:61: error: conflicting types for 'nvidia_ctl_open'
nvidia_ctl.c:16: error: previous declaration of 'nvidia_ctl_open' was here
nvidia_ctl.c:78: error: conflicting types for 'nvidia_ctl_close'
nvidia_ctl.c:17: error: previous declaration of 'nvidia_ctl_close' was here
nvidia_ctl.c:78: error: conflicting types for 'nvidia_ctl_close'
nvidia_ctl.c:17: error: previous declaration of 'nvidia_ctl_close' was here
nvidia_ctl.c:96: error: conflicting types for 'nvidia_ctl_ioctl'
nvidia_ctl.c:18: error: previous declaration of 'nvidia_ctl_ioctl' was here
nvidia_ctl.c:96: error: conflicting types for 'nvidia_ctl_ioctl'
nvidia_ctl.c:18: error: previous declaration of 'nvidia_ctl_ioctl' was here
nvidia_ctl.c:168: error: conflicting types for 'nvidia_ctl_poll'
nvidia_ctl.c:19: error: previous declaration of 'nvidia_ctl_poll' was here
nvidia_ctl.c:168: error: conflicting types for 'nvidia_ctl_poll'
nvidia_ctl.c:19: error: previous declaration of 'nvidia_ctl_poll' was here
nvidia_ctl.c: In function `nvidia_ctl_attach':
nvidia_ctl.c:211: warning: assignment makes integer from pointer without a cast
nvidia_ctl.c: In function `nvidia_ctl_detach':
nvidia_ctl.c:248: warning: passing arg 1 of `destroy_dev' makes pointer from integer without a cast
nvidia_ctl.c: At top level:
nvidia_ctl.c:16: warning: 'nvidia_ctl_open' used but never defined
nvidia_ctl.c:17: warning: 'nvidia_ctl_close' used but never defined
nvidia_ctl.c:18: warning: 'nvidia_ctl_ioctl' used but never defined
nvidia_ctl.c:19: warning: 'nvidia_ctl_poll' used but never defined
*** Error code 1

The 1.0-6113 driver, on the other hand, builds without complaint on this system.
I'm afraid I'm at a loss to explain what's causing the compiler to balk at the 4365 flavor of nvidia_ctl.c.
Is some "patch" available that will "fix" 1.0-4365 for gcc-3.4.2 ?