K8 ECC error with linux-2.6.32

[Johannes Hirte]

With kernel 2.6.32 I get now:

Dec 11 21:26:37 datengrab kernel: Northbridge Error, node 0, core: -1
Dec 11 21:26:37 datengrab kernel: K8 ECC error.

First I thought this triggered by radeon KMS, since with this driver I get 
lots of this entries in the log together with screen corruptions. It doesn't 
happen on X start up but after a while working with X.

Now I've seen that the ECC errors also appear with the proprietary fglrx 
driver. It only occours one time at X start up here

Dec 11 21:26:37 datengrab kernel: [fglrx] AGP detected, AgpState   = 
0x1f000b3b (hardware caps of chipset)
Dec 11 21:26:37 datengrab kernel: [fglrx] [agp] enabling AGP with 
mode=0x1f000b3a
Dec 11 21:26:37 datengrab kernel: agpgart-amd64 0000:00:00.0: AGP 3.0 bridge
Dec 11 21:26:37 datengrab kernel: agpgart-amd64 0000:00:00.0: putting AGP V3 
device into 8x mode
Dec 11 21:26:37 datengrab kernel: pci 0000:01:00.0: putting AGP V3 device into 
8x mode
Dec 11 21:26:37 datengrab kernel: [fglrx] AGP enabled,  AgpCommand = 
0x1f000312 (selected caps)
Dec 11 21:26:37 datengrab kernel: [fglrx] Setup AGP aperture
Dec 11 21:26:37 datengrab kernel: Northbridge Error, node 0, core: -1
Dec 11 21:26:37 datengrab kernel: K8 ECC error.
Dec 11 21:26:38 datengrab kernel: [fglrx] Could not enable MSI; System 
prevented initialization
Dec 11 21:26:38 datengrab kernel: [fglrx] Firegl kernel thread PID: 2565
Dec 11 21:26:39 datengrab kernel: [fglrx] Gart cacheable size:1316 M.
Dec 11 21:26:39 datengrab kernel: [fglrx] Reserved FB block: Shared offset:0, 
size:1000000
Dec 11 21:26:39 datengrab kernel: [fglrx] Reserved FB block: Unshared 
offset:fd0b000, size:2f5000
Dec 11 21:26:39 datengrab kernel: [fglrx] Reserved FB block: Unshared 
offset:1fffb000, size:5000

After forcing AGP 8x to 4x mode, it doesn't happen again with fglrx. I've 
changed drivers/char/agp/generic.c for this. For curiosity the radeon driver 
with KMS initialized AGP in 4x mode itself without the need to force it.

Dec  7 22:50:59 datengrab kernel: agpgart-amd64 0000:00:00.0: AGP 3.0 bridge
Dec  7 22:50:59 datengrab kernel: agpgart-amd64 0000:00:00.0: putting AGP V3 
device into 4x mode
Dec  7 22:50:59 datengrab kernel: radeon 0000:01:00.0: putting AGP V3 device 
into 4x mode

Nevertheless the ECC errors happen here together with the screen corruptions 
which make a restart of X necessary.

Any ideas whats going wrong here?

regards,
  Johannes
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/