pci bars not assigned properly

pkan_1999 · Post by **pkan_1999** » 2013/10/07 22:19:18

My pcie graphics devices are not assigned by os somehow.
Attaching the error snippet from the dmesg below.
We also tried running the command "echo 1 > /sys/bus/pci/rescan" to force reenumeration but results are the same.

[ 0.000000] Linux version 2.6.40-3.0.9 (root@l-sim-33-025) (gcc version 4.1.2 20080704 (Red Hat 4.1.2-51)) #11 SMP Wed Nov 7 23:42:24 PST 2012

[ 0.849224] ACPI: ACPI bus type pnp unregistered
[ 0.857440] PCI: max bus depth: 3 pci_try_num: 4
[ 0.857620] pci 0000:00:02.0: BAR 9: can't assign mem pref (size 0x18000000)
[ 0.857623] pci 0000:00:02.0: BAR 7: assigned [io 0x1000-0x1fff]
[ 0.857625] pci 0000:01:00.0: BAR 9: can't assign mem pref (size 0x18000000)
[ 0.857627] pci 0000:01:00.0: BAR 8: can't assign mem (size 0x1000000)
[ 0.857629] pci 0000:01:00.0: BAR 7: assigned [io 0x1000-0x1fff]
[ 0.857631] pci 0000:02:00.0: BAR 9: can't assign mem pref (size 0x18000000)
[ 0.857632] pci 0000:02:00.0: BAR 8: can't assign mem (size 0x1000000)
[ 0.857634] pci 0000:02:00.0: BAR 7: assigned [io 0x1000-0x1fff]
[ 0.857636] pci 0000:03:00.0: BAR 1: can't assign mem pref (size 0x10000000)
[ 0.857637] pci 0000:03:00.0: BAR 3: can't assign mem pref (size 0x2000000)
[ 0.857639] pci 0000:03:00.0: BAR 0: can't assign mem (size 0x1000000)
[ 0.857640] pci 0000:03:00.0: BAR 6: can't assign mem pref (size 0x80000)
[ 0.857642] pci 0000:03:00.0: BAR 5: assigned [io 0x1000-0x107f]
[ 0.859772] pci 0000:03:00.0: BAR 5: set to [io 0x1000-0x107f] (PCI address [0x1000-0x107f])

Here is the lspsci output of our device which doesn't get assign properly

03:00.0 VGA compatible controller: nVidia Corporation Device 13bf (rev a1) (prog-if 00 [VGA controller])
Control: I/O- Mem- BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- SERR- (64-bit, prefetchable) [disabled]
Region 3: Memory at (64-bit, prefetchable) [disabled]
Region 5: I/O ports at 1000 [disabled] [size=128]
Capabilities: [60] Power Management version 3
Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
Capabilities: [68] MSI: Enable- Count=1/1 Maskable- 64bit+
Address: 0000000000000000 Data: 0000
Capabilities: [78] Express (v2) Endpoint, MSI 00
DevCap: MaxPayload 256 bytes, PhantFunc 0, Latency L0s unlimited, L1
Capabilities: [600 v1] Vendor Specific Information: ID=0001 Rev=1 Len=024
Capabilities: [900 v1] #19

I can send the detail dmesg, iomem and lspci -vv output if required.

Thanks.
-Paul Kan

AlanBartlett · Post by **AlanBartlett** » 2013/10/07 22:57:06

[quote]
[ 0.000000] Linux version 2.6.40-3.0.9 (root@l-sim-33-025) (gcc version 4.1.2 20080704 (Red Hat 4.1.2-51)) #11 SMP Wed Nov 7 23:42:24 PST 2012
[/quote]
That is not a [i]CentOS 5[/i] kernel. Please try using the distributed version, [b]kernel-2.6.18-348.18.1.el5[/b].

pkan_1999 · Post by **pkan_1999** » 2013/10/08 22:03:31

Thanks. We can try moving to the centos5 kernel.
However, do you know if there is some known issue with pci assignment with the current kernel version I am using?

AlanBartlett · Post by **AlanBartlett** » 2013/10/08 22:50:32

The kernel that you are using is a total mystery, as the last version 2.6 kernel source tarball released was [b]linux-2.6.39.4[/b]!

The [i]CentOS[/i] project, its developers and volunteers cannot support arbitrary software obtained from other sources.

Currently there is a [url=http://elrepo.org/tiki/kernel-lt]kernel-lt[/url] package available from the [url=http://elrepo.org]ELRepo Project[/url], which is compatible with [i]CentOS 5[/i], if you really cannot use the [i]CentOS[/i] distributed kernel . . .

CentOS