aic94xx status

aic94xx status

[James Bottomley]

The current status is this:

The biggest problem I've been pursuing all week was the hangs and
softlockups on driver insertion.  This actually turns out to be all
caused by the generic timer code and have nothing to do with the actual
aic94xx (the pitfalls of working in Linus' git head, I'm afraid).

The only two remaining serious issues I can now find that need to be
fixed before it can go into scsi-misc are:

1) SATA support.  We at least need this stubbed out so seeing a SATA
device won't cause nasty things to happen (IBM is working on this).
Ideally, we need to integrate this driver with Brian King's SATA/SAS
code, but no-one who has this board has a sata device and vice versa.

2) Error handling.  I seem to have a somewhat flakey connection on my
SAS cage that seems to cause intermittent transport errors.  The current
behaviour is for this driver to find the transport problem, clear the
nexus and then complete the command (i.e. fail it).  So if this happens
at boot time in the initial INQUIRY (and it does for me about 25% of the
time) it never brings up my root disk, so the box panics.  The error
handler needs to be taught to retry the commands for transport failures.
(I can look into doing this unless there are ready volunteers).

3) The wide port formation and deformation issues, which I've described
in a separate email.

Other than this, most of the remaining work is code refactoring, which I
think can be done in-tree, unless anyone else has critical issues to add
to the list.

James

Re: aic94xx status

[Mike Anderson]

James Bottomley <James.Bottomley@SteelEye.com> wrote:
> 1) SATA support.  We at least need this stubbed out so seeing a SATA
> device won't cause nasty things to happen (IBM is working on this).
> Ideally, we need to integrate this driver with Brian King's SATA/SAS
> code, but no-one who has this board has a sata device and vice versa.
> 

Currently I am running a loaned SATA drive with the patch below. It just
generates the "Unidentified device type" type message when it detects a
SATA device and appears in my configuration to not cause any bad things to
happen.

-andmike
--
Michael Anderson
andmike@us.ibm.com

 Do not drop phy_list_lock is we have not taken it. Also added a SATA_DEV
 case as marker for future calls.

 Signed-off-by: Mike Anderson <andmike@us.ibm.com>

 drivers/scsi/sas/sas_discover.c |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

Index: aic94xx-sas-2.6-patched/drivers/scsi/sas/sas_discover.c
===================================================================
--- aic94xx-sas-2.6-patched.orig/drivers/scsi/sas/sas_discover.c	2006-06-23 11:12:01.000000000 -0700
+++ aic94xx-sas-2.6-patched/drivers/scsi/sas/sas_discover.c	2006-07-05 13:19:20.000000000 -0700
@@ -265,6 +265,7 @@ static int sas_get_port_device(struct as
 		rphy = sas_expander_alloc(port->port,
 					  SAS_FANOUT_EXPANDER_DEVICE);
 		break;
+	case SATA_DEV:
 	default:
 		printk("ERROR: Unidentified device type %d\n", dev->dev_type);
 		rphy = NULL;
@@ -272,7 +273,6 @@ static int sas_get_port_device(struct as
 	}
 
 	if (!rphy) {
-		spin_unlock_irqrestore(&port->phy_list_lock, flags);
 		kfree(dev);
 		return -ENODEV;
 	}

Re: aic94xx status

[James Bottomley]

On Mon, 2006-07-10 at 09:09 -0700, Mike Anderson wrote:
> Currently I am running a loaned SATA drive with the patch below. It
> just
> generates the "Unidentified device type" type message when it detects
> a
> SATA device and appears in my configuration to not cause any bad
> things to
> happen.

That seems to cover the directly attached case, thanks.

What happens if the SATA device is remote over an expander using STP?
That seems to have a separate case in sas_ex_discover_end_dev.

Thanks,

James

Re: aic94xx status

[Mike Anderson]

James Bottomley <James.Bottomley@SteelEye.com> wrote:
> On Mon, 2006-07-10 at 09:09 -0700, Mike Anderson wrote:
> > Currently I am running a loaned SATA drive with the patch below. It
> > just
> > generates the "Unidentified device type" type message when it detects
> > a
> > SATA device and appears in my configuration to not cause any bad
> > things to
> > happen.
> 
> That seems to cover the directly attached case, thanks.
> 
> What happens if the SATA device is remote over an expander using STP?
> That seems to have a separate case in sas_ex_discover_end_dev.

I tried the SATA drive connected off a edge device prior to having to
return the device. I saw the devices address in the attached message, but
nothing else was printed. At least it did not cause any other issues.

-andmike
--
Michael Anderson
andmike@us.ibm.com

RE: aic94xx status

[Robert Tarte]

> 1) SATA support.  We at least need this stubbed out so seeing a SATA
> device won't cause nasty things to happen (IBM is working on this).
> Ideally, we need to integrate this driver with Brian King's SATA/SAS
> code, but no-one who has this board has a sata device and vice versa.

[Tarte, Robert] I'm not sure what Brian King's SATA/SA code is, but I do
have boards that use the razor chip and I have SATA devices.

Re: aic94xx status

[Brian King]

Robert Tarte wrote:
>> 1) SATA support.  We at least need this stubbed out so seeing a SATA
>> device won't cause nasty things to happen (IBM is working on this).
>> Ideally, we need to integrate this driver with Brian King's SATA/SAS
>> code, but no-one who has this board has a sata device and vice versa.
> 
> [Tarte, Robert] I'm not sure what Brian King's SATA/SA code is, but I do
> have boards that use the razor chip and I have SATA devices.

The patch referred to here is a patch to libata that adds a set of APIs
that allows a SAS HBA to hook into libata to control SATA devices. It
was written so that I could get ipr to hook into libata, and the hope
is that it could be used by other SAS drivers as well. Unfortunately,
no other SAS drivers that support SATA passthrough have really tried
using it yet.

http://marc.theaimsgroup.com/?l=linux-ide&m=115279779219927&w=2

Brian

Re: aic94xx status

[Luben Tuikov]

--- Brian King <brking@us.ibm.com> wrote:
> Robert Tarte wrote:
> >> 1) SATA support.  We at least need this stubbed out so seeing a SATA
> >> device won't cause nasty things to happen (IBM is working on this).
> >> Ideally, we need to integrate this driver with Brian King's SATA/SAS
> >> code, but no-one who has this board has a sata device and vice versa.
> > 
> > [Tarte, Robert] I'm not sure what Brian King's SATA/SA code is, but I do
> > have boards that use the razor chip and I have SATA devices.
> 
> The patch referred to here is a patch to libata that adds a set of APIs
> that allows a SAS HBA to hook into libata to control SATA devices. It
> was written so that I could get ipr to hook into libata, and the hope
> is that it could be used by other SAS drivers as well. Unfortunately,
> no other SAS drivers that support SATA passthrough have really tried
> using it yet.

Maybe this is because they have SATL a la proper.

   Luben
 
> http://marc.theaimsgroup.com/?l=linux-ide&m=115279779219927&w=2
> 
> Brian
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>