Re: Bad write performance with libata-tj-stable-2.6.17.4-20060710, pcmcia based sata_sil24, PMP, and NCQ drive

[Derek Taubert <taubert@geeks.org>]

On Mon, Aug 28, 2006 at 09:58:16PM -0700, Derek Taubert wrote:
> On Tue, Aug 29, 2006 at 11:39:57AM +0900, Tejun Heo wrote:
> > The result doesn't seem to indicate any problem in libata or any storage 
> > related kernel subsystem.  I would track down the reader first.
> 
> This drive has only been partitioned; there is no filesystem on it.  So,
> it certainly isn't mounted anywhere.  There honestly isn't _anything_
> other than the dd going on to sda1, and that's the only partition on
> sda.
> 
> After starting dd:
> 
> # fuser -v /dev/sda1
> 
>                      USER        PID ACCESS COMMAND
> /dev/sda1            root      21753 f....  dd
> 
> avg-cpu:  %user   %nice    %sys %iowait   %idle
>            0.90    0.00   11.01   88.09    0.00
> 
> Device:            tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
> sda            1306.61      1295.60      1310.11      12943      13088

Another thought about this...

The sd cache_type is set to "write back" for this drive.  I suspect that
the reads are coming from the block driver attempting to fill lines.

# cat /sys/bus/scsi/devices/0:0:0:0/scsi_disk:0:0:0:0/cache_type
write back

Also, a problem here:

# echo "none" > /sys/bus/scsi/devices/0:0:0:0/scsi_disk:0:0:0:0/cache_type
echo: write error: invalid argument

dmesg says:

sda: Current: sense key: Illegal Request
    Additional sense: Invalid field in cdb

Derek