Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 11-22-05, 04:12 AM   #1
hklygre
Registered User
 
Join Date: Oct 2005
Posts: 16
Default Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Hi,

I've got an HP XW9300 workstation with dual Quadro FX 4500 cards which we intend to use to run two projectors. I had this setup with another XW9300 (but with quadro 3400s, different CPUs, memory and disks) where things worked well, but now I'm unable to get a picture on the second display. The driver fails with the following message:

(EE) NVIDIA(1): Failed to determine FB and I/O addresses!
(EE) NVIDIA(1): *** Aborting ***

- which I've never seen before, and Google hasn't heard of.

Any hints?


(And it is just _too_ ridiculous that I'm not allowed to upload nvidia bug reports without renaming the file. This forum does not allow files with the "log" extension)
Attached Files
File Type: txt nvidia-bug-report.txt (78.9 KB, 119 views)
hklygre is offline   Reply With Quote
Old 11-22-05, 05:13 AM   #2
Thunderbird
 
Join Date: Jul 2002
Location: Netherlands, Europe
Posts: 2,105
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Could you show the output of lspci -v -x of the second videocard? The problem is that somehow the second videocard can't be initialized. If I'm right the system initializes the first card on system startup and the second needs to be initialized when X starts up. (through the int10 module I think) For some reason no memory addresses have been allocated for your card.
Thunderbird is offline   Reply With Quote
Old 11-22-05, 06:02 AM   #3
hklygre
Registered User
 
Join Date: Oct 2005
Posts: 16
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Um... yeah. lspci doesn't show the second card at all, it appears.

0000:00:00.0 Memory controller: nVidia Corporation CK804 Memory Controller (rev a3)
0000:00:01.0 ISA bridge: nVidia Corporation CK804 ISA Bridge (rev a3)
0000:00:01.1 SMBus: nVidia Corporation CK804 SMBus (rev a2)
0000:00:02.0 USB Controller: nVidia Corporation CK804 USB Controller (rev a2)
0000:00:02.1 USB Controller: nVidia Corporation CK804 USB Controller (rev a3)
0000:00:04.0 Multimedia audio controller: nVidia Corporation CK804 AC'97 Audio Controller (rev a2)
0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev a2)
0000:00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller (rev a3)
0000:00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller (rev a3)
0000:00:09.0 PCI bridge: nVidia Corporation CK804 PCI Bridge (rev a2)
0000:00:0a.0 Bridge: nVidia Corporation CK804 Ethernet Controller (rev a3)
0000:00:0e.0 PCI bridge: nVidia Corporation CK804 PCIE Bridge (rev a3)
0000:05:05.0 FireWire (IEEE 1394): Texas Instruments TSB43AB22/A IEEE-1394a-2000 Controller (PHY/Link)
0000:0a:00.0 VGA compatible controller: nVidia Corporation: Unknown device 009d (rev a1)


The only display adapter in the lspci-output is this:

0000:0a:00.0 VGA compatible controller: nVidia Corporation: Unknown device 009d (rev a1) (prog-if 00 [VGA])
Subsystem: nVidia Corporation: Unknown device 02af
Flags: bus master, fast devsel, latency 0, IRQ 20
Memory at c0000000 (32-bit, non-prefetchable) [size=16M]
Memory at b0000000 (64-bit, prefetchable) [size=256M]
Memory at c1000000 (64-bit, non-prefetchable) [size=16M]
I/O ports at 3000 [size=128]
Capabilities: [60] Power Management version 2
Capabilities: [68] Message Signalled Interrupts: 64bit+ Queue=0/0 Enable-
Capabilities: [78] #10 [0001]
00: de 10 9d 00 07 00 10 00 a1 00 00 03 00 00 00 00
10: 00 00 00 c0 0c 00 00 b0 00 00 00 00 04 00 00 c1
20: 00 00 00 00 01 30 00 00 00 00 00 00 de 10 af 02
30: 00 00 00 00 60 00 00 00 00 00 00 00 05 01 00 00

- which is the first card.

Do you know if there should be two PCIE-bridges when there's two X16-slots?

Maybe I should try to pull down the latest kernel sources - they may have fixed this. 2.6.14.1 was the first kernel I could get to boot properly (with all CPUs), so the latest rc or git-kernel may have more fixes for this system.
hklygre is offline   Reply With Quote
Old 11-22-05, 07:03 AM   #4
Thunderbird
 
Join Date: Jul 2002
Location: Netherlands, Europe
Posts: 2,105
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

I'm not fully upto spec with the nforce workstation chipsets. In case of normal desktops you useally have one PCI-E 16x slot and in a SLI system you use two PCI-E 16x slots at 8x speed I think. You need to use some sort of 'jumper'-card to enable this slot I think.

The professional nforce chipset supports two 16x slots but at some stage you had to use multiple bridges perhaps now it is builtin. (there are chipsets with two 16x slots) I don't know what is used on your board.
Thunderbird is offline   Reply With Quote
Old 11-22-05, 08:04 AM   #5
hklygre
Registered User
 
Join Date: Oct 2005
Posts: 16
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Apparently the HP XW9300 uses a Tyan Thunder K8WE (http://www.tyan.com/products/html/thunderk8we.html) mainboard, which claims to have two full speed x16 slots. It also looks like they are on different chips, which I presume mean different buses.

Anyway - I've now tried 2.6.15-rc2-git1, and there's no difference.

My contact in HP is going to try to get me a beta version of the nvidia driver, and tomorrow I'm going to a conference with some AMD people, so I'll have to try and fish some information out of them as well.
hklygre is offline   Reply With Quote
Old 11-22-05, 08:34 AM   #6
zander
NVIDIA Corporation
 
zander's Avatar
 
Join Date: Aug 2002
Posts: 3,740
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

If only one NVIDIA graphics device is found, as is the case on your system, then there's nothing the NVIDIA Linux graphics driver can do. You need to determine why the second graphics card doesn't show up on the bus. Please note that this is a different problem than those where one or more of the PCI I/O memory regions haven't been assigned.
zander is offline   Reply With Quote
Old 11-22-05, 09:04 AM   #7
hklygre
Registered User
 
Join Date: Oct 2005
Posts: 16
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Quote:
Originally Posted by zander
If only one NVIDIA graphics device is found, as is the case on your system, then there's nothing the NVIDIA Linux graphics driver can do. You need to determine why the second graphics card doesn't show up on the bus. Please note that this is a different problem than those where one or more of the PCI I/O memory regions haven't been assigned.
That's just it - the nvidia driver apparently finds both cards, even though lspci doesn't show them.

From the XFree86 log:

(--) PCI:*(10:0:0) nVidia Corporation unknown chipset (0x009d) rev 161, Mem @ 0xc0000000/24, 0xb0000000/28, 0xc1000000/24, I/O @ 0x3000/7
(--) PCI: (129:0:0) nVidia Corporation unknown chipset (0x009d) rev 161, Mem @ 0xc3000000/24, 0xd0000000/28, 0xc4000000/24, I/O @ 0x1000/7

- so both cards are detected by the graphics driver. The kernel apparently doesn't detect things properly, so that might still be the problem. I'll try to contact linux-kernel as well.
hklygre is offline   Reply With Quote
Old 11-22-05, 09:12 AM   #8
zander
NVIDIA Corporation
 
zander's Avatar
 
Join Date: Aug 2002
Posts: 3,740
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Judging from the log file, the NVIDIA driver only sees one device (it depends on information provided by the kernel's PCI sub system); the log excerpt you're referring to is generated by the X server's PCI probe code. That's still an interesting data point, though; does the card show up with `lspci -H1`, as well? Can you try a Linux kernel older than Linux 2.6.13?
zander is offline   Reply With Quote

Old 11-22-05, 03:35 PM   #9
hklygre
Registered User
 
Join Date: Oct 2005
Posts: 16
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Quote:
Originally Posted by zander
Judging from the log file, the NVIDIA driver only sees one device (it depends on information provided by the kernel's PCI sub system); the log excerpt you're referring to is generated by the X server's PCI probe code. That's still an interesting data point, though; does the card show up with `lspci -H1`, as well?
Quote:
Can you try a Linux kernel older than Linux 2.6.13?
Not really - 2.6.14.1 was the first that could find more than 64 megs(!) of RAM.

However, I've solved (of sorts) the problem - booting with pci=noacpi gives me one more PCIE bridge and one more VGA card:

0000:00:00.0 Memory controller: nVidia Corporation CK804 Memory Controller (rev a3)
0000:00:01.0 ISA bridge: nVidia Corporation CK804 ISA Bridge (rev a3)
0000:00:01.1 SMBus: nVidia Corporation CK804 SMBus (rev a2)
0000:00:02.0 USB Controller: nVidia Corporation CK804 USB Controller (rev a2)
0000:00:02.1 USB Controller: nVidia Corporation CK804 USB Controller (rev a3)
0000:00:04.0 Multimedia audio controller: nVidia Corporation CK804 AC'97 Audio Controller (rev a2)
0000:00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev a2)
0000:00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller (rev a3)
0000:00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller (rev a3)
0000:00:09.0 PCI bridge: nVidia Corporation CK804 PCI Bridge (rev a2)
0000:00:0a.0 Bridge: nVidia Corporation CK804 Ethernet Controller (rev a3)
0000:00:0e.0 PCI bridge: nVidia Corporation CK804 PCIE Bridge (rev a3)
0000:05:05.0 FireWire (IEEE 1394): Texas Instruments TSB43AB22/A IEEE-1394a-2000 Controller (PHY/Link)
0000:0a:00.0 VGA compatible controller: nVidia Corporation: Unknown device 009d (rev a1)
0000:80:00.0 Memory controller: nVidia Corporation CK804 Memory Controller (rev a3)
0000:80:01.0 Memory controller: nVidia Corporation CK804 Memory Controller (rev a3)
0000:80:0e.0 PCI bridge: nVidia Corporation CK804 PCIE Bridge (rev a3)
0000:81:00.0 VGA compatible controller: nVidia Corporation: Unknown device 009d (rev a1)

So apparently all is now well, X finds both cards, but I'll have to test some more tomorrow.
hklygre is offline   Reply With Quote
Old 11-22-05, 04:09 PM   #10
zander
NVIDIA Corporation
 
zander's Avatar
 
Join Date: Aug 2002
Posts: 3,740
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

Thanks for the update; can you post the output of `lspci -t` when you've booted with pci=noacpi?
zander is offline   Reply With Quote
Old 11-22-05, 07:38 PM   #11
hklygre
Registered User
 
Join Date: Oct 2005
Posts: 16
Default Re: Problem with dual Quadro FX 4500: "Failed to determine FB and I/O addresses!"

lspci -t:

-+-[80]-+-00.0
| +-01.0
| \-0e.0-[81]----00.0
\-[00]-+-00.0
+-01.0
+-01.1
+-02.0
+-02.1
+-04.0
+-06.0
+-07.0
+-08.0
+-09.0-[05]----05.0
+-0a.0
\-0e.0-[0a]----00.0
hklygre is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -5. The time now is 12:31 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.