1 |
Hi, |
2 |
|
3 |
> I tried these kernels (all vanilla): |
4 |
> 2.6.32.13 |
5 |
> 2.6.33.5 |
6 |
> 2.6.34.0 |
7 |
|
8 |
So it's not a known problem that has been fixed. |
9 |
|
10 |
Just a wild guess... can you try recompiling the kernel *without* |
11 |
pata_via? Some people have reported having problems with sata drives on |
12 |
VIA controllers when pata_via is also loaded. If you need to access |
13 |
devices on the PATA ports, you can use the the old non-libata via82cxxx |
14 |
IDE driver. |
15 |
|
16 |
> This is, what I cut from the dmegs out (I hope to get all relevant |
17 |
> infos...if you missing something, I save the complete dmesg output to |
18 |
> disk for later refrence...): |
19 |
|
20 |
This is just the controller initialization part, it does not include the |
21 |
actual information about the drives. Look for something like "ata1: SATA |
22 |
link up..." a bit further in the log. |
23 |
|
24 |
> fdisk -cu -S 56 /dev/sda |
25 |
|
26 |
Well, fdisk -cu defaults to creating the first partition beginning at |
27 |
sector 2048, which is fine as it's divisible by 8 (and thus is aligned |
28 |
on a 4096-byte boundary). I don't think you need to specify the number |
29 |
of sectors per track. |
30 |
|
31 |
But this is not really relevant, as creating unaligned partitions merely |
32 |
results in lower performance, not DMA errors :) |
33 |
|
34 |
> Do I have to replace my motherboard/graphics card/RAM/... |
35 |
> only because WD cannot talk to and/or vice versa? |
36 |
|
37 |
If the pata_via trick above does not work, and an upgrade is out of |
38 |
question, my suggestion is to find a cheap PCI sata2 controller and to |
39 |
use that instead of the on-board sata ports. |
40 |
|
41 |
> Can I use PCI cards on newer motherboards with PCIe-slots??? |
42 |
|
43 |
You cannot use PCI cards in PCIe slots, but most new motherboards still |
44 |
have at least one PCI slot. |
45 |
|
46 |
andrea |