Red Hat Knowledgebase

In Red Hat Enterprise Linux 4.5 support for Native Command Queueing (NCQ) was added to the nVidia SATA controller driver. Some disks do not properly respond to NCQ commands and cause problems, commonly expressed as delays in IO to the disks and sometimes system instability.

Errors similar to the following will be in /var/log/messages if this problem is present on the system:

Jul 13 09:56:55 localhost kernel: ata1: CPB flags CMD err, flags=0x11
Jul 13 09:57:25 localhost kernel: ata1: EH in ADMA mode, notifier 0x0 notifier_error 0x0 gen_ctl 0x1501000 status 0x400
Jul 13 09:57:25 localhost kernel: ata1: CPB 0: ctl_flags 0x1f, resp_flags 0x2
Jul 13 09:57:25 localhost kernel: ata1: CPB 1: ctl_flags 0x1f, resp_flags 0x11
Jul 13 09:57:25 localhost kernel: ata1: CPB 2: ctl_flags 0x1f, resp_flags 0x2
Jul 13 09:57:25 localhost kernel: ata1: CPB 3: ctl_flags 0x1f, resp_flags 0x2
Jul 13 09:57:25 localhost kernel: ata1: CPB 4: ctl_flags 0x1f, resp_flags 0x2

A workaround is available. Perform the following actions:

1. Add the following line to /etc/modprobe.conf:

   options sata_nv adma=0
   

2. Rebuild the initrd:

   mkinitrd -f /boot/initrd-(kernel version).img (kernel version)
   

3. Reboot the system.

This will prevent NCQ from being used. We have identified the following hard drives which are affected by this problem:

"HITACHI HDS7250SASUN500G 0621KTAWSD", "K2AOAJ0AHITACHI"
"HITACHI HDS7250SASUN500G 0607K2BMDD", "K2AOA20AHITACHI"

If the system has a different drive, please contact Red Hat Support so it can be identified, and added to the list of drives to disable NCQ for automatically in a future update of Red Hat Enterprise Linux 4.