Re: SATA disk/controller problems

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Wed, Aug 13, 2008 at 08:36:49PM +0200, Jaap Boender wrote:
> Hello,
> 
> I've just bought a computer with an ICH7 disk controller, and a 250 Gb 
> Hitachi HDP725025GLA380 SATA disk. At first glance, everything works 
> fine, but at times of great disk activity, this error pops up:
> 
> piixide1:0:0: bus-master DMA error: status=0x26
> wd0e: error writing fsbn 46755296 of 46755296-46755327 (wd0 bn 57242591; 
> cn 56788 tn 4 sn 35), retrying
> wd0: (aborted command, interface CRC error)
> piixide1:0:0: bus-master DMA error: status=0x26
> wd0e: error writing fsbn 46755296 of 46755296-46755327 (wd0 bn 57242591; 
> cn 56788 tn 4 sn 35), retrying
> wd0: (aborted command, interface CRC error)
> piixide1:0:0: bus-master DMA error: status=0x26
> wd0e: error writing fsbn 46755296 of 46755296-46755327 (wd0 bn 57242591; 
> cn 56788 tn 4 sn 35), retrying
> wd0: (aborted command, interface CRC error)
> wd0: soft error (corrected)
> piixide1:0:0: bus-master DMA error: status=0x26
> wd0: transfer error, downgrading to Ultra-DMA mode 5
> wd0(piixide1:0:0): using PIO mode 4, Ultra-DMA mode 5 (Ultra/100) (using 
> DMA)
> 
> This goes on until the disk arrives in PIO mode 4 without any DMA, and 
> then it's just the "error writing fsbn..." messages. Sometimes, I also 
> see this:
> 
> piixide1:0:0: lost interrupt
> type: ata tc_bcount: 8192 tc_skip: 0
> wd0e: device timeout writing fsbn 47607424 of 47607424-47607439 (wd0 bn 
> 58094719; cn 57633 tn 10 sn 25), retrying
> wd0: soft error (corrected)
> 
> First I thought that the disk had just been damaged in transport, but 
> the SMART self-test doesn't find anything, and it really only happens 
> when I do things like runring two big compilations together.
> 
> I've tried activating the FORCE_LBA48 quirk, but that doesn't help, and 
> I've also tried disabling the PCI IDE controller (piixide0 in the dmesg 
> below) in the BIOS, but that doesn't help either. ahcisata(4) is enabled 
> in the kernel, but not attached.
> 
> Could this be a problem on the NetBSD side, or is the disk just broken?

I don't have issues with piixide SATA controllers. Things to
check:
- remplace the cable
- make sure the power supply is strong enough to properly power the
  motherboard and peripherals. I've seen this kind of behavior go away
  after remplacing the power supply by a stonger one (before deciding
  how strong your PSU is, check how much it can provide on the
  +5V and +12V rails: I've seen 500W PSU that would provide less current
  on 5V and 12V than a 400W one; the extra watts were all for the
  +3.3V used by the motherboard).

if noone of this help, try a different drive.

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--