dual card problems

[aaronj]

Hello everyone

I have been trying to get 2 nvidia cards, fx3000 and a fx5200pci to work. This is on a 2885 tyan board. With Fedora core 3 x86_64.

My probelm is that whenever I modprobe nvidia (the latest 6629) driver I get the following error.

NVRM: request mem region failed for 0x0:0x1, aborting
nvidia: probe of 0000:02:07.0 failed with erorr -1
NVRM: the Nvidia probe was not called for 1 device.
NVRM: loading NVIDIA Linux x86_64 NVIDIA ernel Module.

Note the PCI address for fx5200pci is 02:07.0
and for fx3000 is 05:00.0

The AGP card works no problems its just the PCI card.

Are there any suggestion out there as to how I can resolve this issue. I am dying to get 3 screens working.

Cheers
aaronj

[ifaxsoln]

I'm actually seeing something similar with a single PCI card ... a Quadro NVS running on fedora core 3, x86_64. Apparently this card has two GPUs onboard, so it's supposed to look like two separate graphics cards. The output of lspci supports this:

04:00.0 VGA compatible controller: nVidia Corporation NV17GL [Quadro4 200/400 NVS] (rev a3)
04:04.0 VGA compatible controller: nVidia Corporation NV17GL [Quadro4 200/400 NVS] (rev a3)

When I load the nvidia module though, only the 04:04.0 device is recognized, ... for the 04:00.0 one I get:

nvidia: module license 'NVIDIA' taints kernel.
ACPI: PCI interrupt 0000:04:00.0[A] -> GSI 16 (level, low) -> IRQ 185
NVRM: request_mem_region failed for 0x0:0x1, aborting
nvidia: probe of 0000:04:00.0 failed with error -1
ACPI: PCI interrupt 0000:04:04.0[A] -> GSI 16 (level, low) -> IRQ 185
NVRM: the NVIDIA probe routine was not called for 1 device(s)!!
NVRM: loading NVIDIA Linux x86_64 NVIDIA Kernel Module 1.0-6629 Wed Nov 3 11:43:48

An engineer @ PNY suggested I try upgrading the BIOS on my Tyan 2885, so I went from 2.02 to their 2.02w (beta release). That made no difference.

I'm completely stumped.

I do manage to get TwinView running with the GPU on 04:04.0 but performance is terrible ... hopefully I'll be able to figure out why.

-Darren

[zander]

NVRM: request_mem_region failed for 0x0:0x1, aborting

The invalid range (0MB@0x00000000) suggests that the SBIOS didn't configure the secondary graphics device.

[ifaxsoln]

Originally Posted by zander

NVRM: request_mem_region failed for 0x0:0x1, aborting

The invalid range (0MB@0x00000000) suggests that the SBIOS didn't configure the secondary graphics device.

Thanks for the reply! Are you suggesting that the BIOS on my Tyan 2885 should be doing something differently? After all, it _does_ appear in the list of PCI devices ...

If so, any hints on how I might try to get it to configure that device? Typical tweaks? Or is this likely to be a limitation of the motherboard that might never be addressed?

-Darren

[zander]

The installed SBIOS doesn't appear to assign a valid address range for BAR0 (device registers); what does lspci -nvvv -s 4:00.0 report?

[geowiz]

Same problem on a Tyan 2875 with a 5200 PCI card & a Quadro 980 AGP card. Whichever is the first graphics card in the bios works but the secondary card apparently has memory initialization issues.

Output of lspci -nvv -s 02:00.0 (secondary card per bios option) is attached. It indicates the I/O and memory are not being used and memory is disabled.

Tried kernel 2.6.6 & 2.6.9, Nvidia 5332 (on 2.6.6 kernel) and 6629 on both (x86_64). Tried using the NV AGP, agpgart and no agp options in the XF86Config - no joy. Also booted with "noapic pci=noapic noacpi" with the same (non) results.

We know the bios on the Tyan 2875 is out of date on this motherboard and are going to update it tomorrow(Dec 10 -updated bios, still no joy). I'm not holding my breath that it will fix the problem.

Any hints where to go next?

Thanks - Geowiz

[ifaxsoln]

Originally Posted by zander

The installed SBIOS doesn't appear to assign a valid address range for BAR0 (device registers); what does lspci -nvvv -s 4:00.0 report?

Thanks for your interest Zander! I moved it to a different PCI slot so the IDs are different, but the problem's the same and here's the output you requested:

[root@nexus ~]# lspci -nvvv -s 2:00.0
02:00.0 Class 0300: 10de:017a (rev a3)
Subsystem: 10de:014c
Control: I/O- Mem- BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Interrupt: pin A routed to IRQ 209
Region 0: Memory at <ignored> (32-bit, non-prefetchable) [disabled]
Region 1: Memory at <ignored> (32-bit, prefetchable) [disabled]
Region 2: Memory at <ignored> (32-bit, prefetchable) [disabled]
Expansion ROM at <ignored> [disabled]
Capabilities: [60] Power Management version 2
Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

Compare this to the one that works:

[root@nexus ~]# lspci -nvvv -s 2:04.0
02:04.0 Class 0300: 10de:017a (rev a3)
Subsystem: 10de:014c
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 248 (1250ns min, 250ns max)
Interrupt: pin A routed to IRQ 209
Region 0: Memory at fd000000 (32-bit, non-prefetchable) [size=16M]
Region 1: Memory at d0000000 (32-bit, prefetchable) [size=128M]
Region 2: Memory at d8f80000 (32-bit, prefetchable) [size=512K]
Expansion ROM at fe5e0000 [disabled] [size=128K]
Capabilities: [60] Power Management version 2
Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

[root@nexus ~]#

[jschellhaass]

Have you tried this option?

Option "UseInt10Module" "boolean"
Enable use of the X Int10 module to soft-boot all
secondary cards, rather than POSTing the cards through
the NVIDIA kernel module. Default: off (POSTing is done
through the NVIDIA kernel module).


jeff

[geowiz]

Yes, I have tried "UseInt10Module" and I got exactly the same results.

Guess we'll stick to simple dual head from one graphics card for now.

Thanks,

Geowiz

[zander]

This really looks like a problem you'll want to talk to Tyan about.

[ifaxsoln]

Originally Posted by jschellhaass

Have you tried this option?

Option "UseInt10Module" "boolean"
Enable use of the X Int10 module to soft-boot all
secondary cards, rather than POSTing the cards through
the NVIDIA kernel module. Default: off (POSTing is done
through the NVIDIA kernel module).


jeff

Well Jeff, I'm not sure I know how to apply that option to my particular problem (unable to get second GPU on the NVS-400 to work) but here's what I tried. The GPU that I can't seem to get running is at PCI:2:4:0, so I tried to run on just that screen:

Section "ServerLayout"
Identifier "Default Layout"
Screen "Screen1"
InputDevice "Mouse0" "CorePointer"
InputDevice "Keyboard0" "CoreKeyboard"
EndSection


Section "Device"
Identifier "NV PCI 1"
VendorName "nvidia"
Driver "nvidia"
# update this with the PCI id of your card. Consult the output
# of the 'lspci' command. The BusID is usually optional when
# only using one graphics card.
BusID "PCI:2:4:0"
Option "UseInt10Module" "yes"
EndSection

Section "Screen"
Identifier "Screen0"
Device "NV PCI 1"
Monitor "MyMonitor"
DefaultDepth 24

Subsection "Display"
Depth 8
Modes "1280x1024" "1024x768" "800x600" "640x400"
EndSubsection
Subsection "Display"
Depth 16
Modes "1280x1024" "1024x768" "800x600" "640x480"
EndSubsection
Subsection "Display"
Depth 24
Modes "1280x1024" "1024x768" "800x600" "640x480"
EndSubsection

EndSection

And here's a snippet of the Xorg.0 logfile:

(**) NVIDIA(0): Option "UseInt10Module" "yes"
(II) Loading sub module "int10"
(II) LoadModule: "int10"
(II) Loading /usr/X11R6/lib64/modules/linux/libint10.a
(II) Module int10: vendor="X.Org Foundation"
compiled for 6.8.1, module version = 1.0.0
ABI class: X.Org Video Driver, version 0.7
(--) NVIDIA(0): Linear framebuffer at 0xB8800000
(--) NVIDIA(0): MMIO registers at 0xFA500000
(EE) NVIDIA(0): Failed to initialize the NVIDIA graphics device!
(EE) NVIDIA(0): *** Aborting ***
(II) UnloadModule: "nvidia"
(II) UnloadModule: "int10"
(II) Unloading /usr/X11R6/lib64/modules/linux/libint10.a
(EE) Screen(s) found, but none have a usable configuration.

Fatal server error:
no screens found


I'll attach the complete logfile to this message.

Is this what you had in mind?

-Darren

[ifaxsoln]

Sorry, I excerpted the wrong Screen and Device sections there ... I'll just attach my .conf file so you can see how I was configured.