On 01/22/2011 11:31 AM, Toralf Förster wrote:

Robert Hancock wrote at 17:36:15

quoted

(CCing linux-ide)

On 01/22/2011 09:25 AM, Toralf Förster wrote:

quoted

Hello,

Henrique de Moraes Holschuh wrote at 14:11:12

quoted

I don't even know whom to refer you to.  Maybe ask in LKML directly?
Use the subject "libata EH running when switching VTs".  If one of
those VTs are in X.org control, say that as well.

I'm wondering why I get this in /var/log/messages w/ kernel 2.6.37 when I
switch from VT7 (running xorg ) to VT12 (or 1 or 2 or another number) and
back:


2011-01-22T16:20:53.854+01:00 n22 kernel: ata1: hard resetting link
2011-01-22T16:20:54.163+01:00 n22 kernel: ata1: SATA link up 1.5 Gbps
(SStatus 113 SControl 300)
2011-01-22T16:20:54.164+01:00 n22 kernel: ata1.00: ACPI cmd
ef/02:00:00:00:00:a0 (SET FEATURES) succeeded
2011-01-22T16:20:54.164+01:00 n22 kernel: ata1.00: ACPI cmd
f5/00:00:00:00:00:a0 (SECURITY FREEZE LOCK) filtered out
2011-01-22T16:20:54.165+01:00 n22 kernel: ata1.00: ACPI cmd
ef/5f:00:00:00:00:a0 (SET FEATURES) succeeded
2011-01-22T16:20:54.165+01:00 n22 kernel: ata1.00: ACPI cmd
ef/10:03:00:00:00:a0 (SET FEATURES) filtered out
2011-01-22T16:20:54.169+01:00 n22 kernel: ata1.00: ACPI cmd
ef/02:00:00:00:00:a0 (SET FEATURES) succeeded
2011-01-22T16:20:54.169+01:00 n22 kernel: ata1.00: ACPI cmd
f5/00:00:00:00:00:a0 (SECURITY FREEZE LOCK) filtered out
2011-01-22T16:20:54.169+01:00 n22 kernel: ata1.00: ACPI cmd
ef/5f:00:00:00:00:a0 (SET FEATURES) succeeded
2011-01-22T16:20:54.169+01:00 n22 kernel: ata1.00: ACPI cmd
ef/10:03:00:00:00:a0 (SET FEATURES) filtered out
2011-01-22T16:20:54.170+01:00 n22 kernel: ata1.00: configured for
UDMA/100 2011-01-22T16:20:54.171+01:00 n22 kernel: ata1: EH complete


I'm running an almost stable Gentoo on a ThinkPad T400 with current
kernel (2.6.37, earlier shows this behaviour too)

Is that the only output you're getting? Can you post the full dmesg log
from bootup?

Of course - attached here

Curious, there seems to be no indication of why EH was triggered, 
doesn't seem to be have been due to an error. Does this also happen in 
single-user mode or with minimal processes running? It's possible that 
userspace is triggering this somehow.