Fwd: Re: Problem with Linux aic94xx driver

Fwd: Re: Problem with Linux aic94xx driver

[Luben Tuikov]

--- Terje Røsten <terje.rosten@ntnu.no> wrote:
> Hi!
> 
> I have a Tyan Transport TA26 barebone, with an addon card that I think 
> has a AIC-94xx chip.
> I am using Fedora Core 6 with a kernel from Fedora rawhide (development):
> 
> $ uname -a
> Linux randi.phys.ntnu.no 2.6.19-1.2891.fc7 #1 SMP Thu Dec 21 10:59:56 
> EST 2006 x86_64 x86_64 x86_64 GNU/Linux
> 
> It's 2.6.20rc1-git5 kernel with some patches.
> 
> When I load the module I get:
> 
> aic94xx: Adaptec aic94xx SAS/SATA driver version 1.0.2 loaded
> ACPI: PCI Interrupt 0000:03:05.0[A] -> GSI 25 (level, low) -> IRQ 25
> aic94xx: found Adaptec AIC-9410W SAS/SATA Host Adapter, device 0000:03:05.0
> scsi7 : aic94xx
> aic94xx: using sequencer V17/10c6
> aic94xx: device 0000:03:05.0: SAS addr 500e081000011470, PCBA SN , 8 
> phys, 8 enabled phys, flash present, BIOS build 1608
> ERROR: Unidentified device type 5
> ERROR: Unidentified device type 5
> ERROR: Unidentified device type 5
> ERROR: Unidentified device type 5
> 
> (It's 4 Western Digital RE2 WD5000YS with NCQ (500 GB SATA) on the 
> controller,
> set up in RAID BIOS  to one RAID 1+0 logical drive).
> 
> Seems like I end up here:
> 
> drivers/scsi/libsas/sas_discover.c: sas_get_port_device() :
> 
> 256   switch (dev->dev_type) {
> 257         case SAS_END_DEV:
> 258                 rphy = sas_end_device_alloc(port->port);
> 259                 break;
> 260         case EDGE_DEV:
> 261                 rphy = sas_expander_alloc(port->port,
> 262                                           SAS_EDGE_EXPANDER_DEVICE);
> 263                 break;
> 264         case FANOUT_DEV:
> 265                 rphy = sas_expander_alloc(port->port,
> 266                                           SAS_FANOUT_EXPANDER_DEVICE);
> 267                 break;
> 268         case SATA_DEV:
> 269         default:
> 270                 printk("ERROR: Unidentified device type %d\n", 
> dev->dev_type);
> 271                 rphy = NULL;
> 272                 break;
> 273         }
> 
> Any ideas?

Terje,

In my SAS Stack, I _do not_ have that switch statement in that
function.  Apparently you are using "libsas", which is based
on an early, early version of my code, which bottomley and his
gang of SAS experts appropriated to "maintain".  And as your
problem shows has been "fixed" in many ways.

You should complain to them -- I've CC-ed the linux-scsi mailing
list.

BTW, I don't experience this bug with the SAS Stack as I maintain it.

    Luben

-
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

Re: Fwd: Re: Problem with Linux aic94xx driver

[Darrick J. Wong]

--- Terje Røsten <terje.rosten@ntnu.no> wrote:

> (It's 4 Western Digital RE2 WD5000YS with NCQ (500 GB SATA) on the
> controller,
> set up in RAID BIOS  to one RAID 1+0 logical drive).

I suppose this may be a moot point since Mr. Røsten bought a better
controller, but for the record, there exists another SAS stack that also
does not have this bug (namely, the entire lack of SATA support):

http://www.kernel.org/git/?p=linux/kernel/git/jejb/aic94xx-sas-2.6.git;a=summary

There are various stability/functionality fixes for that stack that are
under development as well:

http://sweaglesw.net/~djwong/docs/sas_ata-patches/

That said, even if Mr. Røsten had built a driver with SATA support, he
wrote that he had configured a RAID in the BIOS; to use that under
Linux, one needs the dmraid tool to set up BIOS RAID arrays as
device-mapper devices.  Of course, it is possible that at least one of
the many (five? six?) Adaptec 94xx driver variants supports both SATA
and HostRAID natively.

Better to use a nice 3ware card that takes care of all this in hardware.

--D
-
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

Re: Fwd: Re: Problem with Linux aic94xx driver

[Luben Tuikov]

--- "Darrick J. Wong" <djwong@us.ibm.com> wrote:
> --- Terje Røsten <terje.rosten@ntnu.no> wrote:
> 
> > (It's 4 Western Digital RE2 WD5000YS with NCQ (500 GB SATA) on the
> > controller,
> > set up in RAID BIOS  to one RAID 1+0 logical drive).
> 
> I suppose this may be a moot point since Mr. Røsten bought a better
> controller, but for the record, there exists another SAS stack that also
> does not have this bug (namely, the entire lack of SATA support):
> 
> http://www.kernel.org/git/?p=linux/kernel/git/jejb/aic94xx-sas-2.6.git;a=summary

And there exists the original SAS Stack _with_ SATA support, _without_ this
bug, which I distribute.  I'm not sure if Terje was aware of it.

I think it is clear from his letter that he wanted SATA support.  Why would
he use a SAS Stack which has no SATA support, and thus trivially wouldn't
exhibit this bug, as per your suggestion...

This all of course begs the question: Why is there such a blatant
bug in the "master" branch of the Linux kernel for this driver?
Yep, I'm looking at tag v2.6.20-rc2 and it is right there where he
pointed out that it is.  The commit introducing it is
2908d778ab3e244900c310974e1fc1c69066e450, and it is one blob commit,
without previous git history, only the log message.

You seem to have replied to an email which I posted to linux-scsi, i.e.
my email is in the To: header.  You also seem to have edited the mail
headers to remove me from the CC/To list which is automatically built when
one replies to a message.

Removing people from CC lists is considered impolite. Adding people 
is ok.

   Luben

> There are various stability/functionality fixes for that stack that are
> under development as well:
> 
> http://sweaglesw.net/~djwong/docs/sas_ata-patches/
> 
> That said, even if Mr. Røsten had built a driver with SATA support, he
> wrote that he had configured a RAID in the BIOS; to use that under
> Linux, one needs the dmraid tool to set up BIOS RAID arrays as
> device-mapper devices.  Of course, it is possible that at least one of
> the many (five? six?) Adaptec 94xx driver variants supports both SATA
> and HostRAID natively.
> 
> Better to use a nice 3ware card that takes care of all this in hardware.


-
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

Re: Fwd: Re: Problem with Linux aic94xx driver

[Darrick J. Wong]

Luben Tuikov wrote:

> You seem to have replied to an email which I posted to linux-scsi, i.e.
> my email is in the To: header.  You also seem to have edited the mail
> headers to remove me from the CC/To list which is automatically built when
> one replies to a message.
> 
> Removing people from CC lists is considered impolite. Adding people
> is ok.

I apologize if you took offense; no slight was intended.  I had thought
that you were not interested in further discussion of support issues
relating to the aic94xx driver in mainline because you maintain your own
driver that does not have these issues.

In any case, thank you for catching the omission.

--D