Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Jeff Garzik]

Mark Nelson wrote:
> Add an appropriate entry for the Promise PDC42819 controller. It has an
> AHCI mode and seems to work correctly with board_ahci.
> 
> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650 (4 port)
> software-based RAID cards (for which there is a binary driver, t3sas) and
> can be found on some motherboards, for example the MSI K9A2 Platinum,
> which calls the chip a Promise T3 controller.
> 
> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
> ---
> Apparently this chip also supports SAS disks, will these work with the
> ahci driver (I've only tested with normal desktop SATA disks)?

ahci is definitely SATA-only, so I wonder how they modified AHCI to 
support SAS transport...

	Jeff

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Alan Cox]

> > Apparently this chip also supports SAS disks, will these work with the
> > ahci driver (I've only tested with normal desktop SATA disks)?
> 
> ahci is definitely SATA-only, so I wonder how they modified AHCI to 
> support SAS transport...

Jeff while are on that subject a further reminder - AHCI is also breaking
Marvell PATA device support for this reason and we now have months of
regressions because of that.

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Jeff Garzik]

Alan Cox wrote:
>>> Apparently this chip also supports SAS disks, will these work with the
>>> ahci driver (I've only tested with normal desktop SATA disks)?
>> ahci is definitely SATA-only, so I wonder how they modified AHCI to 
>> support SAS transport...
> 
> Jeff while are on that subject a further reminder - AHCI is also breaking
> Marvell PATA device support for this reason and we now have months of
> regressions because of that.

Refresh my memory...  what is the problem with loading pata_marvell first?

AFAICS, if you load pata_marvell before ahci, it will grab the PATA 
disks and the PCI device itself.

	Jeff

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Alan Cox]

> Refresh my memory...  what is the problem with loading pata_marvell first?

The fact its almost impossible to do with a distribution kernel. eg
you can't make it work with Fedora (see the Fedora bugs assigned to
you)

If pata_marvell should be driving these chips please remove the entries
from the ahci.c driver as I asked before.

If ahci.c should be driving them then we need the documentation for
whatever extension they use driving the ports via AHCI. Clearly there is
one as its possible under Windows.

> AFAICS, if you load pata_marvell before ahci, it will grab the PATA 
> disks and the PCI device itself.

If you have a degree in initrd-foo

Alan

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Jeff Garzik]

Alan Cox wrote:
>> Refresh my memory...  what is the problem with loading pata_marvell first?
> 
> The fact its almost impossible to do with a distribution kernel. eg
> you can't make it work with Fedora (see the Fedora bugs assigned to
> you)

Hogwash -- just load one driver before the other.


> If pata_marvell should be driving these chips please remove the entries
> from the ahci.c driver as I asked before.
> 
> If ahci.c should be driving them then we need the documentation for
> whatever extension they use driving the ports via AHCI. Clearly there is
> one as its possible under Windows.

Situation before ahci mod:

	Users with PATA may use pata_marvell.
	Users wanting SATA screwed.
	Users wanting PATA+SATA screwed.

Situation after ahci mod:

	Users with PATA may use pata_marvell
	Users with SATA may use ahci
	Users wanting PATA+SATA screwed


>> AFAICS, if you load pata_marvell before ahci, it will grab the PATA 
>> disks and the PCI device itself.
> 
> If you have a degree in initrd-foo

Just switch the order of 'ahci' and 'pata_marvell' in 
/etc/modprobe.conf, then use Fedora's tools regenerate the initrd.

See?  It's not rocket science, and the current configuration can be 
easily made to work for Fedora users.

	Jeff

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Alan Cox]

>> Situation before ahci mod:
> 
> 	Users with PATA may use pata_marvell.
> 	Users wanting SATA screwed.

WRONG as I've said repeatedly. pata_marvell handles the SATA ports too as
legacy

> Just switch the order of 'ahci' and 'pata_marvell' in 
> /etc/modprobe.conf, then use Fedora's tools regenerate the initrd.

> See?  It's not rocket science, and the current configuration can be 
> easily made to work for Fedora users.

Well they don't seem to be able to work it out, and your basic premise is
incorrect anyway. Both ports work with pata_marvell.

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Alan Cox]

Oh and some small other PS items from going back over mail


- You previously (a year ago) stated you had the docs for the extension

- You've been ignoring bugs assigned to you about this for months (not
even replying to them)

- You can't even install most distros to work around this mess you
created to sort out the initrd


So I'm simply at this point going to give up and send patches and an
explanation to Linus. If you care to find the docs you said you had then
I can fix it nicely, if not then I'll fix it the blunter way.

Alan

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mark Nelson]

On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
> Mark Nelson wrote:
>>
>> Add an appropriate entry for the Promise PDC42819 controller. It has an
>> AHCI mode and seems to work correctly with board_ahci.
>>
>> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650 (4
>> port)
>> software-based RAID cards (for which there is a binary driver, t3sas) and
>> can be found on some motherboards, for example the MSI K9A2 Platinum,
>> which calls the chip a Promise T3 controller.
>>
>> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
>> ---
>> Apparently this chip also supports SAS disks, will these work with the
>> ahci driver (I've only tested with normal desktop SATA disks)?
>
> ahci is definitely SATA-only, so I wonder how they modified AHCI to support
> SAS transport...

That is a good question. From my understanding the controller is a SAS
controller that can drive SATA disks (as I guess all SAS controllers can
do...). Could it be that the controller can be setup to act as an AHCI
SATA controller or if the right registers are poked a SAS controller?

Thanks!

Mark

>
>        Jeff
>
>
>
>

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mikael Pettersson]

Mark Nelson writes:
 > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
 > > Mark Nelson wrote:
 > >>
 > >> Add an appropriate entry for the Promise PDC42819 controller. It has an
 > >> AHCI mode and seems to work correctly with board_ahci.
 > >>
 > >> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650 (4
 > >> port)
 > >> software-based RAID cards (for which there is a binary driver, t3sas) and
 > >> can be found on some motherboards, for example the MSI K9A2 Platinum,
 > >> which calls the chip a Promise T3 controller.
 > >>
 > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
 > >> ---
 > >> Apparently this chip also supports SAS disks, will these work with the
 > >> ahci driver (I've only tested with normal desktop SATA disks)?
 > >
 > > ahci is definitely SATA-only, so I wonder how they modified AHCI to support
 > > SAS transport...
 > 
 > That is a good question. From my understanding the controller is a SAS
 > controller that can drive SATA disks (as I guess all SAS controllers can
 > do...). Could it be that the controller can be setup to act as an AHCI
 > SATA controller or if the right registers are poked a SAS controller?

Based on how the sata_promise-supported chips work I'd guess that they
have dual programming interfaces with AHCI for JBOD SATA and a native
one (perhaps just additional control registers) for SAS and raid.

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mark Nelson]

On Tue, Sep 2, 2008 at 10:08 PM, Mikael Pettersson <mikpe@it.uu.se> wrote:
> Mark Nelson writes:
>  > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
>  > > Mark Nelson wrote:
>  > >>
>  > >> Add an appropriate entry for the Promise PDC42819 controller. It has an
>  > >> AHCI mode and seems to work correctly with board_ahci.
>  > >>
>  > >> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650 (4
>  > >> port)
>  > >> software-based RAID cards (for which there is a binary driver, t3sas) and
>  > >> can be found on some motherboards, for example the MSI K9A2 Platinum,
>  > >> which calls the chip a Promise T3 controller.
>  > >>
>  > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
>  > >> ---
>  > >> Apparently this chip also supports SAS disks, will these work with the
>  > >> ahci driver (I've only tested with normal desktop SATA disks)?
>  > >
>  > > ahci is definitely SATA-only, so I wonder how they modified AHCI to support
>  > > SAS transport...
>  >
>  > That is a good question. From my understanding the controller is a SAS
>  > controller that can drive SATA disks (as I guess all SAS controllers can
>  > do...). Could it be that the controller can be setup to act as an AHCI
>  > SATA controller or if the right registers are poked a SAS controller?
>
> Based on how the sata_promise-supported chips work I'd guess that they
> have dual programming interfaces with AHCI for JBOD SATA and a native
> one (perhaps just additional control registers) for SAS and raid.
>

That's interesting; and makes sense. I'm guessing that we won't end up
with an open source driver that can do SAS on these chips then... But
for now at least we can use them as SATA controllers and if Promise
wants to come to the table with specs or datasheets they can in the
future. I'm just happy to be able to use my eSATA ports (and perhaps
the possibility of getting my hard disk off the SB600, although I should
benchmark both controllers before jumping to any conclusions).

Thanks!

Mark

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Jeff Garzik]

Mark Nelson wrote:
> On Tue, Sep 2, 2008 at 10:08 PM, Mikael Pettersson <mikpe@it.uu.se> wrote:
>> Mark Nelson writes:
>>  > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
>>  > > Mark Nelson wrote:
>>  > >>
>>  > >> Add an appropriate entry for the Promise PDC42819 controller. It has an
>>  > >> AHCI mode and seems to work correctly with board_ahci.
>>  > >>
>>  > >> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650 (4
>>  > >> port)
>>  > >> software-based RAID cards (for which there is a binary driver, t3sas) and
>>  > >> can be found on some motherboards, for example the MSI K9A2 Platinum,
>>  > >> which calls the chip a Promise T3 controller.
>>  > >>
>>  > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
>>  > >> ---
>>  > >> Apparently this chip also supports SAS disks, will these work with the
>>  > >> ahci driver (I've only tested with normal desktop SATA disks)?
>>  > >
>>  > > ahci is definitely SATA-only, so I wonder how they modified AHCI to support
>>  > > SAS transport...
>>  >
>>  > That is a good question. From my understanding the controller is a SAS
>>  > controller that can drive SATA disks (as I guess all SAS controllers can
>>  > do...). Could it be that the controller can be setup to act as an AHCI
>>  > SATA controller or if the right registers are poked a SAS controller?
>>
>> Based on how the sata_promise-supported chips work I'd guess that they
>> have dual programming interfaces with AHCI for JBOD SATA and a native
>> one (perhaps just additional control registers) for SAS and raid.
>>
> 
> That's interesting; and makes sense. I'm guessing that we won't end up
> with an open source driver that can do SAS on these chips then... But
> for now at least we can use them as SATA controllers and if Promise
> wants to come to the table with specs or datasheets they can in the
> future. I'm just happy to be able to use my eSATA ports (and perhaps
> the possibility of getting my hard disk off the SB600, although I should
> benchmark both controllers before jumping to any conclusions).

FWIW, I'll poke my Promise contacts, but they have gone mostly silent in 
the past year, so don't get your hopes up...

	Jeff

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mark Nelson]

On Thu, Sep 4, 2008 at 8:37 PM, Jeff Garzik <jeff@garzik.org> wrote:
> Mark Nelson wrote:
>>
>> On Tue, Sep 2, 2008 at 10:08 PM, Mikael Pettersson <mikpe@it.uu.se> wrote:
>>>
>>> Mark Nelson writes:
>>>  > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
>>>  > > Mark Nelson wrote:
>>>  > >>
>>>  > >> Add an appropriate entry for the Promise PDC42819 controller. It
>>> has an
>>>  > >> AHCI mode and seems to work correctly with board_ahci.
>>>  > >>
>>>  > >> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650
>>> (4
>>>  > >> port)
>>>  > >> software-based RAID cards (for which there is a binary driver,
>>> t3sas) and
>>>  > >> can be found on some motherboards, for example the MSI K9A2
>>> Platinum,
>>>  > >> which calls the chip a Promise T3 controller.
>>>  > >>
>>>  > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
>>>  > >> ---
>>>  > >> Apparently this chip also supports SAS disks, will these work with
>>> the
>>>  > >> ahci driver (I've only tested with normal desktop SATA disks)?
>>>  > >
>>>  > > ahci is definitely SATA-only, so I wonder how they modified AHCI to
>>> support
>>>  > > SAS transport...
>>>  >
>>>  > That is a good question. From my understanding the controller is a SAS
>>>  > controller that can drive SATA disks (as I guess all SAS controllers
>>> can
>>>  > do...). Could it be that the controller can be setup to act as an AHCI
>>>  > SATA controller or if the right registers are poked a SAS controller?
>>>
>>> Based on how the sata_promise-supported chips work I'd guess that they
>>> have dual programming interfaces with AHCI for JBOD SATA and a native
>>> one (perhaps just additional control registers) for SAS and raid.
>>>
>>
>> That's interesting; and makes sense. I'm guessing that we won't end up
>> with an open source driver that can do SAS on these chips then... But
>> for now at least we can use them as SATA controllers and if Promise
>> wants to come to the table with specs or datasheets they can in the
>> future. I'm just happy to be able to use my eSATA ports (and perhaps
>> the possibility of getting my hard disk off the SB600, although I should
>> benchmark both controllers before jumping to any conclusions).
>
> FWIW, I'll poke my Promise contacts, but they have gone mostly silent in the
> past year, so don't get your hopes up...

Thanks Jeff!

No need for hope - we have ahci now :)

(speaking of which, should I resend the patch without the RFC?)

Mark

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Jeff Garzik]

Mark Nelson wrote:
> On Thu, Sep 4, 2008 at 8:37 PM, Jeff Garzik <jeff@garzik.org> wrote:
>> Mark Nelson wrote:
>>> On Tue, Sep 2, 2008 at 10:08 PM, Mikael Pettersson <mikpe@it.uu.se> wrote:
>>>> Mark Nelson writes:
>>>>  > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
>>>>  > > Mark Nelson wrote:
>>>>  > >>
>>>>  > >> Add an appropriate entry for the Promise PDC42819 controller. It
>>>> has an
>>>>  > >> AHCI mode and seems to work correctly with board_ahci.
>>>>  > >>
>>>>  > >> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650
>>>> (4
>>>>  > >> port)
>>>>  > >> software-based RAID cards (for which there is a binary driver,
>>>> t3sas) and
>>>>  > >> can be found on some motherboards, for example the MSI K9A2
>>>> Platinum,
>>>>  > >> which calls the chip a Promise T3 controller.
>>>>  > >>
>>>>  > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
>>>>  > >> ---
>>>>  > >> Apparently this chip also supports SAS disks, will these work with
>>>> the
>>>>  > >> ahci driver (I've only tested with normal desktop SATA disks)?
>>>>  > >
>>>>  > > ahci is definitely SATA-only, so I wonder how they modified AHCI to
>>>> support
>>>>  > > SAS transport...
>>>>  >
>>>>  > That is a good question. From my understanding the controller is a SAS
>>>>  > controller that can drive SATA disks (as I guess all SAS controllers
>>>> can
>>>>  > do...). Could it be that the controller can be setup to act as an AHCI
>>>>  > SATA controller or if the right registers are poked a SAS controller?
>>>>
>>>> Based on how the sata_promise-supported chips work I'd guess that they
>>>> have dual programming interfaces with AHCI for JBOD SATA and a native
>>>> one (perhaps just additional control registers) for SAS and raid.
>>>>
>>> That's interesting; and makes sense. I'm guessing that we won't end up
>>> with an open source driver that can do SAS on these chips then... But
>>> for now at least we can use them as SATA controllers and if Promise
>>> wants to come to the table with specs or datasheets they can in the
>>> future. I'm just happy to be able to use my eSATA ports (and perhaps
>>> the possibility of getting my hard disk off the SB600, although I should
>>> benchmark both controllers before jumping to any conclusions).
>> FWIW, I'll poke my Promise contacts, but they have gone mostly silent in the
>> past year, so don't get your hopes up...
> 
> Thanks Jeff!
> 
> No need for hope - we have ahci now :)

Well, I still think this PCI ID addition leaves open the possibility 
that plugging in a SAS disk could hang the system or similar misbehavior.

	Jeff

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mikael Pettersson]

Jeff Garzik writes:
 > Mark Nelson wrote:
 > > On Thu, Sep 4, 2008 at 8:37 PM, Jeff Garzik <jeff@garzik.org> wrote:
 > >> Mark Nelson wrote:
 > >>> On Tue, Sep 2, 2008 at 10:08 PM, Mikael Pettersson <mikpe@it.uu.se> wrote:
 > >>>> Mark Nelson writes:
 > >>>>  > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
 > >>>>  > > Mark Nelson wrote:
 > >>>>  > >>
 > >>>>  > >> Add an appropriate entry for the Promise PDC42819 controller. It
 > >>>> has an
 > >>>>  > >> AHCI mode and seems to work correctly with board_ahci.
 > >>>>  > >>
 > >>>>  > >> This chip is found on Promise's FastTrak TX2650 (2 port) and TX4650
 > >>>> (4
 > >>>>  > >> port)
 > >>>>  > >> software-based RAID cards (for which there is a binary driver,
 > >>>> t3sas) and
 > >>>>  > >> can be found on some motherboards, for example the MSI K9A2
 > >>>> Platinum,
 > >>>>  > >> which calls the chip a Promise T3 controller.
 > >>>>  > >>
 > >>>>  > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
 > >>>>  > >> ---
 > >>>>  > >> Apparently this chip also supports SAS disks, will these work with
 > >>>> the
 > >>>>  > >> ahci driver (I've only tested with normal desktop SATA disks)?
 > >>>>  > >
 > >>>>  > > ahci is definitely SATA-only, so I wonder how they modified AHCI to
 > >>>> support
 > >>>>  > > SAS transport...
 > >>>>  >
 > >>>>  > That is a good question. From my understanding the controller is a SAS
 > >>>>  > controller that can drive SATA disks (as I guess all SAS controllers
 > >>>> can
 > >>>>  > do...). Could it be that the controller can be setup to act as an AHCI
 > >>>>  > SATA controller or if the right registers are poked a SAS controller?
 > >>>>
 > >>>> Based on how the sata_promise-supported chips work I'd guess that they
 > >>>> have dual programming interfaces with AHCI for JBOD SATA and a native
 > >>>> one (perhaps just additional control registers) for SAS and raid.
 > >>>>
 > >>> That's interesting; and makes sense. I'm guessing that we won't end up
 > >>> with an open source driver that can do SAS on these chips then... But
 > >>> for now at least we can use them as SATA controllers and if Promise
 > >>> wants to come to the table with specs or datasheets they can in the
 > >>> future. I'm just happy to be able to use my eSATA ports (and perhaps
 > >>> the possibility of getting my hard disk off the SB600, although I should
 > >>> benchmark both controllers before jumping to any conclusions).
 > >> FWIW, I'll poke my Promise contacts, but they have gone mostly silent in the
 > >> past year, so don't get your hopes up...
 > > 
 > > Thanks Jeff!
 > > 
 > > No need for hope - we have ahci now :)
 > 
 > Well, I still think this PCI ID addition leaves open the possibility 
 > that plugging in a SAS disk could hang the system or similar misbehavior.

Agreed, there's no telling what Promise-specific magic is needed
to properly drive the T3.

/Mikael

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Alan Cox]

> > No need for hope - we have ahci now :)
> 
> Well, I still think this PCI ID addition leaves open the possibility 
> that plugging in a SAS disk could hang the system or similar misbehavior.

And not adding it means it doesn't work at all and you'll never find out
as you won't get the bug report from the SAS user 8)

Alan

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mark Nelson]

On Fri, Sep 5, 2008 at 12:28 AM, Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
>> > No need for hope - we have ahci now :)
>>
>> Well, I still think this PCI ID addition leaves open the possibility
>> that plugging in a SAS disk could hang the system or similar misbehavior.
>
> And not adding it means it doesn't work at all and you'll never find out
> as you won't get the bug report from the SAS user 8)
>

Another fair point. I'll let you guys work out which is the better way
to do things as you are the gurus :)

Thanks!

Mark

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mark Nelson]

On Thu, Sep 4, 2008 at 11:43 PM, Jeff Garzik <jeff@garzik.org> wrote:
> Mark Nelson wrote:
>>
>> On Thu, Sep 4, 2008 at 8:37 PM, Jeff Garzik <jeff@garzik.org> wrote:
>>>
>>> Mark Nelson wrote:
>>>>
>>>> On Tue, Sep 2, 2008 at 10:08 PM, Mikael Pettersson <mikpe@it.uu.se>
>>>> wrote:
>>>>>
>>>>> Mark Nelson writes:
>>>>>  > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
>>>>>  > > Mark Nelson wrote:
>>>>>  > >>
>>>>>  > >> Add an appropriate entry for the Promise PDC42819 controller. It
>>>>> has an
>>>>>  > >> AHCI mode and seems to work correctly with board_ahci.
>>>>>  > >>
>>>>>  > >> This chip is found on Promise's FastTrak TX2650 (2 port) and
>>>>> TX4650
>>>>> (4
>>>>>  > >> port)
>>>>>  > >> software-based RAID cards (for which there is a binary driver,
>>>>> t3sas) and
>>>>>  > >> can be found on some motherboards, for example the MSI K9A2
>>>>> Platinum,
>>>>>  > >> which calls the chip a Promise T3 controller.
>>>>>  > >>
>>>>>  > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
>>>>>  > >> ---
>>>>>  > >> Apparently this chip also supports SAS disks, will these work
>>>>> with
>>>>> the
>>>>>  > >> ahci driver (I've only tested with normal desktop SATA disks)?
>>>>>  > >
>>>>>  > > ahci is definitely SATA-only, so I wonder how they modified AHCI
>>>>> to
>>>>> support
>>>>>  > > SAS transport...
>>>>>  >
>>>>>  > That is a good question. From my understanding the controller is a
>>>>> SAS
>>>>>  > controller that can drive SATA disks (as I guess all SAS controllers
>>>>> can
>>>>>  > do...). Could it be that the controller can be setup to act as an
>>>>> AHCI
>>>>>  > SATA controller or if the right registers are poked a SAS
>>>>> controller?
>>>>>
>>>>> Based on how the sata_promise-supported chips work I'd guess that they
>>>>> have dual programming interfaces with AHCI for JBOD SATA and a native
>>>>> one (perhaps just additional control registers) for SAS and raid.
>>>>>
>>>> That's interesting; and makes sense. I'm guessing that we won't end up
>>>> with an open source driver that can do SAS on these chips then... But
>>>> for now at least we can use them as SATA controllers and if Promise
>>>> wants to come to the table with specs or datasheets they can in the
>>>> future. I'm just happy to be able to use my eSATA ports (and perhaps
>>>> the possibility of getting my hard disk off the SB600, although I should
>>>> benchmark both controllers before jumping to any conclusions).
>>>
>>> FWIW, I'll poke my Promise contacts, but they have gone mostly silent in
>>> the
>>> past year, so don't get your hopes up...
>>
>> Thanks Jeff!
>>
>> No need for hope - we have ahci now :)
>
> Well, I still think this PCI ID addition leaves open the possibility that
> plugging in a SAS disk could hang the system or similar misbehavior.
>

Fair point.

I'm not sure if this is any better but I could change the PCI ID to
specify the subvendor as 0x1462 (MSI) so that it will only match with
T3 controllers on MSI boards (like my K9A2 Platinum). I'd guess that
most people who bought the board would use the T3 for SATA disks
rather than SAS while most people who bought the dedicated
Promise controller would have bought it to use with SAS disks (but
admittedly that assumption could be fundamentally flawed - maybe
people bought the MSI board for the sole purpose of using SAS
without having to buy a discrete controller...)

Thanks!

Mark

Re: [RFC][PATCH] ahci: Add support for Promise PDC42819

[Mark Nelson]

On Fri, Sep 5, 2008 at 12:43 AM, Jeff Garzik <jeff@garzik.org> wrote:
> Mark Nelson wrote:
>>
>> On Thu, Sep 4, 2008 at 8:37 PM, Jeff Garzik <jeff@garzik.org> wrote:
>>>
>>> Mark Nelson wrote:
>>>>
>>>> On Tue, Sep 2, 2008 at 10:08 PM, Mikael Pettersson <mikpe@it.uu.se>
>>>> wrote:
>>>>>
>>>>> Mark Nelson writes:
>>>>>  > On Tue, Sep 2, 2008 at 9:19 PM, Jeff Garzik <jeff@garzik.org> wrote:
>>>>>  > > Mark Nelson wrote:
>>>>>  > >>
>>>>>  > >> Add an appropriate entry for the Promise PDC42819 controller. It
>>>>> has an
>>>>>  > >> AHCI mode and seems to work correctly with board_ahci.
>>>>>  > >>
>>>>>  > >> This chip is found on Promise's FastTrak TX2650 (2 port) and
>>>>> TX4650
>>>>> (4
>>>>>  > >> port)
>>>>>  > >> software-based RAID cards (for which there is a binary driver,
>>>>> t3sas) and
>>>>>  > >> can be found on some motherboards, for example the MSI K9A2
>>>>> Platinum,
>>>>>  > >> which calls the chip a Promise T3 controller.
>>>>>  > >>
>>>>>  > >> Signed-off-by: Mark Nelson <mdnelson8@gmail.com>
>>>>>  > >> ---
>>>>>  > >> Apparently this chip also supports SAS disks, will these work
>>>>> with
>>>>> the
>>>>>  > >> ahci driver (I've only tested with normal desktop SATA disks)?
>>>>>  > >
>>>>>  > > ahci is definitely SATA-only, so I wonder how they modified AHCI
>>>>> to
>>>>> support
>>>>>  > > SAS transport...
>>>>>  >
>>>>>  > That is a good question. From my understanding the controller is a
>>>>> SAS
>>>>>  > controller that can drive SATA disks (as I guess all SAS controllers
>>>>> can
>>>>>  > do...). Could it be that the controller can be setup to act as an
>>>>> AHCI
>>>>>  > SATA controller or if the right registers are poked a SAS
>>>>> controller?
>>>>>
>>>>> Based on how the sata_promise-supported chips work I'd guess that they
>>>>> have dual programming interfaces with AHCI for JBOD SATA and a native
>>>>> one (perhaps just additional control registers) for SAS and raid.
>>>>>
>>>> That's interesting; and makes sense. I'm guessing that we won't end up
>>>> with an open source driver that can do SAS on these chips then... But
>>>> for now at least we can use them as SATA controllers and if Promise
>>>> wants to come to the table with specs or datasheets they can in the
>>>> future. I'm just happy to be able to use my eSATA ports (and perhaps
>>>> the possibility of getting my hard disk off the SB600, although I should
>>>> benchmark both controllers before jumping to any conclusions).
>>>
>>> FWIW, I'll poke my Promise contacts, but they have gone mostly silent in
>>> the
>>> past year, so don't get your hopes up...
>>
>> Thanks Jeff!
>>
>> No need for hope - we have ahci now :)
>
> Well, I still think this PCI ID addition leaves open the possibility that
> plugging in a SAS disk could hang the system or similar misbehavior.

I finally managed to get my hands on a SAS disk to try with the T3
controller (it's an IBM-branded SAS disk borrowed from a bimini).
With the patch that adds the PCI ID to the ahci driver nothing bad
happens to the system when the SAS disk is connected - it just
times out:

ACPI: PCI Interrupt 0000:02:00.0[A] -> GSI 17 (level, low) -> IRQ 17
ahci 0000:02:00.0: AHCI 0001.0100 32 slots 4 ports 3 Gbps 0xf impl RAID mode
ahci 0000:02:00.0: flags: 64bit ncq sntf ilck pm led clo pmp pio slum part
PCI: Setting latency timer of device 0000:02:00.0 to 64
scsi4 : ahci
scsi5 : ahci
scsi6 : ahci
scsi7 : ahci
ata5: SATA max UDMA/133 abar m8192@0xfbefc000 port 0xfbefc100 irq 17
ata6: SATA max UDMA/133 abar m8192@0xfbefc000 port 0xfbefc180 irq 17
ata7: SATA max UDMA/133 abar m8192@0xfbefc000 port 0xfbefc200 irq 17
ata8: SATA max UDMA/133 abar m8192@0xfbefc000 port 0xfbefc280 irq 17
ata5: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
ata5: link online but device misclassified, retrying
ata5: reset failed (errno=-11), retrying in 10 secs
ata5: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
ata5: link online but device misclassified, retrying
ata5: reset failed (errno=-11), retrying in 10 secs
ata5: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
ata5: link online but device misclassified, retrying
ata5: reset failed (errno=-11), retrying in 35 secs
ata5: limiting SATA link speed to 1.5 Gbps
ata5: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
ata5: link online but device misclassified, device detection might fail
ata6: SATA link down (SStatus 0 SControl 300)
ata7: SATA link down (SStatus 0 SControl 300)
ata8: SATA link down (SStatus 0 SControl 300)

And even with the SAS disk connected but out of action the T3
can still use my eSATA disk perfectly:

ata7: exception Emask 0x10 SAct 0x0 SErr 0x4040000 action 0xe frozen
ata7: irq_stat 0x00000040, connection status changed
ata7: SError: { CommWake DevExch }
ata7: hard resetting link
ata7: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
ata7.00: ATA-7: HTS721060G9SA00, MC3IC10H, max UDMA/100
ata7.00: 117210240 sectors, multi 0: LBA48
ata7.00: configured for UDMA/100
ata7: EH complete
scsi 6:0:0:0: Direct-Access     ATA      HTS721060G9SA00  MC3I PQ: 0 ANSI: 5
sd 6:0:0:0: [sdb] 117210240 512-byte hardware sectors (60012 MB)
sd 6:0:0:0: [sdb] Write Protect is off
sd 6:0:0:0: [sdb] Mode Sense: 00 3a 00 00
sd 6:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't
support DPO or FUA
sd 6:0:0:0: [sdb] 117210240 512-byte hardware sectors (60012 MB)
sd 6:0:0:0: [sdb] Write Protect is off
sd 6:0:0:0: [sdb] Mode Sense: 00 3a 00 00
sd 6:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't
support DPO or FUA
 sdb: sdb1
sd 6:0:0:0: [sdb] Attached SCSI disk
sd 6:0:0:0: Attached scsi generic sg3 type 0
kjournald starting.  Commit interval 5 seconds
EXT3 FS on sdb1, internal journal
EXT3-fs: mounted filesystem with ordered data mode.


So, how about my patch then? :)

Mark.