libata-tj-stable success report, with minor patch

libata-tj-stable success report, with minor patch

[Jim Paris]

Hello Tejun,

I recently purchased a Norco DS-1220[1] 12-bay SATA enclosure from
Newegg[2].  It comes with a sil3124-based PCI/X card and uses sil3726
port multipliers.  Ater unsuccesfully wrestling with buggy proprietry
drivers included in the box, I found your libata-tj-stable[3] patches
and applied them to 2.6.16.19.  With the minor fix below, it seems to
be working wonderfully.  Thank you!!

The enclosure takes 4 SATA connections.  Bays appear connected like this:
  ata1.00    = bay 5
  ata1.01-04 = bay 1-4
  ata2.00    = bay 10
  ata2.01-04 = bay 6-9
  ata3.00    = bay 11
  ata4.00    = bay 12

Your patches only detected the first 4 out of 5 ports on each PM.
To fix this, I made this (admittedly naïve) change:

--- drivers/scsi/libata-pm.orig	2006-06-11 21:01:28.000000000 -0400
+++ drivers/scsi/libata-pm.c	2006-06-11 21:03:57.000000000 -0400
@@ -324,7 +324,7 @@
 	 * it requires hardreset to resume PM links.
 	 */
 	if (vendor == 0x1095 && devid == 0x4726) {
-		*nr_ports -= 2;
+		*nr_ports -= 1;
 		*link_flags |= ATA_LFLAG_HRST_TO_RESUME;
 	}
 }

The patch is also available here[4], along with dmesg and lspci output.
With this, all 12 bays are working fine.  I also did some limited
hotplug testing and ran into no problems at all.  Again, thank you!

-jim

[1] http://www.norcotek.com/news.php?news_id=24
[2] http://www.newegg.com/Product/Product.asp?Item=N82E16816133001
[3] http://home-tj.org/wiki/index.php/Libata-tj-stable
[4] http://jim.sh/~jim/libata/

Re: libata-tj-stable success report, with minor patch

[Tejun Heo]

Jim Paris wrote:
> Hello Tejun,
> 
> I recently purchased a Norco DS-1220[1] 12-bay SATA enclosure from
> Newegg[2].  It comes with a sil3124-based PCI/X card and uses sil3726
> port multipliers.  Ater unsuccesfully wrestling with buggy proprietry
> drivers included in the box, I found your libata-tj-stable[3] patches
> and applied them to 2.6.16.19.  With the minor fix below, it seems to
> be working wonderfully.  Thank you!!
> 
> The enclosure takes 4 SATA connections.  Bays appear connected like this:
>   ata1.00    = bay 5
>   ata1.01-04 = bay 1-4
>   ata2.00    = bay 10
>   ata2.01-04 = bay 6-9
>   ata3.00    = bay 11
>   ata4.00    = bay 12
> 
> Your patches only detected the first 4 out of 5 ports on each PM.
> To fix this, I made this (admittedly naïve) change:
> 
> --- drivers/scsi/libata-pm.orig	2006-06-11 21:01:28.000000000 -0400
> +++ drivers/scsi/libata-pm.c	2006-06-11 21:03:57.000000000 -0400
> @@ -324,7 +324,7 @@
>  	 * it requires hardreset to resume PM links.
>  	 */
>  	if (vendor == 0x1095 && devid == 0x4726) {
> -		*nr_ports -= 2;
> +		*nr_ports -= 1;
>  		*link_flags |= ATA_LFLAG_HRST_TO_RESUME;
>  	}
>  }
> 
> The patch is also available here[4], along with dmesg and lspci output.
> With this, all 12 bays are working fine.  I also did some limited
> hotplug testing and ran into no problems at all.  Again, thank you!

Hmmm... sil3726 and 4726 report different number of ports depending on 
configuration.  I'll try to get it right in the next round.  That is a 
pretty cool enclosure you have there.  How's the performance?

-- 
tejun

Re: libata-tj-stable success report, with minor patch

[Jim Paris]

> Hmmm... sil3726 and 4726 report different number of ports depending on 
> configuration.  I'll try to get it right in the next round.

Sounds good to me, thanks again.  Let me know if there's anything I
can help test.

> That is a pretty cool enclosure you have there.  How's the
> performance?

Haven't used it much yet, but it seems decent.  "hdparm -t" gives
80MB/s on the RAID6 array across all disks.  Each disk reads at 60MB/s
alone, and reading from all 12 simultaneously gets 6 MB/s on each of
the ten multiplexed disks and 30 MB/s on each of the last two disks.
The card is in a normal PCI slot so that's probably maxing out my bus.

-jim