Geforce 320, FreeBSD9 amd64, nvidia, Startx=autoreboot

P

pfnv

Every time I use the command startx the system will autoreboot immediately.

I have a Geforce 320m integrated card with shared memory of 256MB. I use FreeBSD 9.0 release for amd64. The nvidia driver is compiled using the ports. During the part of configuration, I have deselected all, i.e. no linux acpi heavy WVINVD and FreeBSD AGP support. The xorg.conf is generated by nvidia-xconfig automatically. And I manually added
Code: 
Option NvAGP "0"

I checked the log file of X, but I didn't find any useful information there. Besides, the VESA driver works for me.

Here is some dmesg information:
Code: 
vgapci0: <VGA-compatible display> port 0x1000-0x107f mem 0xd2000000-0xd2ffffff,0xc0000000-0xcfffffff,0xd0000000-0xd1ffffff
 irq 7 at device 0.0 on pci4
nvidia0: <GeForce 320M> on vgapci0
vgapci0: child nvidia0 requested pci_enable_io
vgapci0: <VGA-compatible display> port 0x1000-0x107f mem 0xd2000000-0xd2ffffff,0xc0000000-0xcfffffff,0xd0000000-0xd1ffffff
 irq 19 at device 0.0 on pci4

ehci0: <EHCI (generic) USB 2.0 controller> mem 0xd348b100-0xd348b1ff irq 19 at device 4.1 on pci0

Because I am using 64bit FreeBSD, this following trick makes no sense.
Code: 
machdep.disable_mtrrs="1"

I have also noticed that somebody have met a similar problem. Some people suggest to disable APIC ACPI. In my case, the system cannot boot without APIC ACPI, the error is something like it cannot find a timer. Does that mean I have to do something like "assign an irq to some device" manually instead of using ACPI?

Thanks for any comment and advice.
 
Try building a kernel without agp(4), I've noticed NVidia doesn't really like it.

I also suggest setting WITHOUT_NOUVEAU in /etc/make.conf and rebuilding x11/xorg. This will give you a slightly newer DRI and associated files.

Code: 
dri-7.6.1,2                         >   succeeds index (index has 7.4.4,2)
libGL-7.6.1                         >   succeeds index (index has 7.4.4)
libGLU-7.6.1                        >   succeeds index (index has 7.4.4)
libdrm-2.4.17_1                     >   succeeds index (index has 2.4.12_1)
libglut-7.6.1                       >   succeeds index (index has 7.4.4)
 
I'm curious about how the agp(4) driver could interfere if there's no matching hardware on the system, for example any newer machine with only PCI and PCIe buses?
 
No idea, but if the hardware isn't there it won't hurt removing it either ;)
 
SirDice said:
Try building a kernel without agp(4), I've noticed NVidia doesn't really like it.

I also suggest setting WITHOUT_NOUVEAU in /etc/make.conf and rebuilding x11/xorg. This will give you a slightly newer DRI and associated files.

Code: 
dri-7.6.1,2                         >   succeeds index (index has 7.4.4,2)
libGL-7.6.1                         >   succeeds index (index has 7.4.4)
libGLU-7.6.1                        >   succeeds index (index has 7.4.4)
libdrm-2.4.17_1                     >   succeeds index (index has 2.4.12_1)
libglut-7.6.1                       >   succeeds index (index has 7.4.4)
Click to expand...

Thank you for your reply. I begin with a fresh install of the base system. Next, I remove the following line from the kernel config file.
Code: 
device agp
Then I recompiled the kernel. I add
Code: 
WITHOUT_NOUVEAU=YES
to /etc/make.conf. At present, I have installed the x11/xorg and x11/nvidia-driver. I got the same version of the ports as those listed in your reply. Unfortunately, the same crash is repeated.
 
The nv driver does not support my card. Thank you for your advice. I will report this on the NVidia forum.
 
You must log in or register to reply here.
Back
Top