Re: Need IDE Taskfile Access

Re: Need IDE Taskfile Access

[Martin Dalecki]

Użytkownik Garet Cammer napisał:
> For some time now we have been writing user applications that send ATAPI 
> commands to the IDE bus to initialize and configure our hardware RAID 1 
> controllers. This has been working well, thanks to Andre's patch that 
> gave us taskfile access through the ioctl API. We were counting on it to 
> be a permanent part of the 2.5/2.6 kernel, since there is a lot of 
> hardware in the field using these apps.
> Imagine our surprise when we discovered that taskfile access was being 
> abandoned completely!
> Although we understand that the kernel may need to filter some commands, 
> why can't applications access at least the Smart commands? Help!
> Regards,
> Garet Cammer
> Software Development
> Arco Computer Products
> (954) 925-2688

ATAPI is no problem. You can use the whole SCSI layer on top of
ide-scsi.

Re: Need IDE Taskfile Access

[Andre Hedrick]

Garet,

You are wasting electons, the interface is gone and the API to the
transport is wrecked.  I will need to compose a loadable module to renable
the support.  Clearly 2.5/2.6 is not friendly with the needs of the
industry and it will never be at this rate.

In the end, I will end up writing a closed ATA binary driver for sale as a
replacement.  I have had several requests to consider the option.  As much
as I do not like the idea, it is less offensive than the current
direction.

Andre Hedrick
LAD Storage Consulting Group


On Tue, 18 Jun 2002, Garet Cammer wrote:

> For some time now we have been writing user applications that send ATAPI commands to the IDE bus to initialize and configure our hardware RAID 1 controllers. This has been working well, thanks to Andre's patch that gave us taskfile access through the ioctl API. We were counting on it to be a permanent part of the 2.5/2.6 kernel, since there is a lot of hardware in the field using these apps.
> Imagine our surprise when we discovered that taskfile access was being abandoned completely!
> Although we understand that the kernel may need to filter some commands, why can't applications access at least the Smart commands? Help!
> Regards,
> Garet Cammer
> Software Development
> Arco Computer Products
> (954) 925-2688
> 

Re: Need IDE Taskfile Access

[Bartlomiej Zolnierkiewicz]

Do not worry Garet, I will reimplement it later in 2.5.
I need some free time to do it, (maybe next month?).

On Tue, 18 Jun 2002, Andre Hedrick wrote:

>
> Garet,
>
> You are wasting electons, the interface is gone and the API to the
> transport is wrecked.  I will need to compose a loadable module to renable
> the support.  Clearly 2.5/2.6 is not friendly with the needs of the
> industry and it will never be at this rate.

Will be...

>
> In the end, I will end up writing a closed ATA binary driver for sale as a
> replacement.  I have had several requests to consider the option.  As much
> as I do not like the idea, it is less offensive than the current
> direction.

Ugh, please dont...

>
> Andre Hedrick
> LAD Storage Consulting Group
>
>
> On Tue, 18 Jun 2002, Garet Cammer wrote:
>
> > For some time now we have been writing user applications that send ATAPI commands to the IDE bus to initialize and configure our hardware RAID 1 controllers. This has been working well, thanks to Andre's patch that gave us taskfile access through the ioctl API. We were counting on it to be a permanent part of the 2.5/2.6 kernel, since there is a lot of hardware in the field using these apps.
> > Imagine our surprise when we discovered that taskfile access was being abandoned completely!
> > Although we understand that the kernel may need to filter some commands, why can't applications access at least the Smart commands? Help!
> > Regards,
> > Garet Cammer
> > Software Development
> > Arco Computer Products
> > (954) 925-2688
> >
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

Re: Need IDE Taskfile Access

[Martin Dalecki]

Użytkownik Bartlomiej Zolnierkiewicz napisał:
> Do not worry Garet, I will reimplement it later in 2.5.
> I need some free time to do it, (maybe next month?).

First the main things have to stabilize.
And please note that using the SMART commands *is
possible* in 2.5 (modulo bugs and mistakes like for
example 92... argh...). I'm checking regularly
whatever ide-smart still works.

> On Tue, 18 Jun 2002, Andre Hedrick wrote:

>>You are wasting electons, the interface is gone and the API to the
>>transport is wrecked.  I will need to compose a loadable module to renable
>>the support.  Clearly 2.5/2.6 is not friendly with the needs of the
>>industry and it will never be at this rate.

If the "industry" asks - I'm responsive for certain.
Unless not - I don't.

> Will be...
> 
>>In the end, I will end up writing a closed ATA binary driver for sale as a
>>replacement.  I have had several requests to consider the option.  As much
>>as I do not like the idea, it is less offensive than the current
>>direction.
> 
> 
> Ugh, please dont...

Let him do. Why not? Sounds lake a sane business plan.

RE: Need IDE Taskfile Access

[Scott Tillman]

> >>You are wasting electons, the interface is gone and the API to the
> >>transport is wrecked.  I will need to compose a loadable module
> to renable
> >>the support.  Clearly 2.5/2.6 is not friendly with the needs of the
> >>industry and it will never be at this rate.
>
> If the "industry" asks - I'm responsive for certain.
> Unless not - I don't.

Well, I'm not "the industry" but I think I know of another need for these
capabilities as well.

I'm working with a group of people in an effort to get Linux running on the
XBox.  The XBox uses a set of security PIO commands to restrict access to
the IDE drive, requiring a 32 byte password to be delivered before sector
access is allowed.  As far as I can tell from my investigations and from
earlier discussions with Andre there is currently no way to issue this
command.  If I'm wrong in my estimation just let me know how, otherwise I
simply wish add my voice to the (albeit small) outcry for supporting the
entire ATA spec.

Another comment/question (related to XBox support):
As part of this effort the xbox-linux team has coded support for the XBox's
proprietary partitioning and it's new filesystem.  This code (and any
further kernel support code) has been developed for the 2.4.18 kernel, and
we have no desire to port it to 2.5.x unless there is some hope of it's
adoption.  Could I get an official decision on whether this code might be
adopted if made available to the 2.5.x kernel?

-Scott Tillman aka SpeedBump

PS: flames about why we are supporting the XBox (a design of the Evil
Empire) will be summarily ignored.  I can only point you to it's HDTV, NTSC,
PAL, and possibly VGA outputs, it's dvd/cd drive, and it's $199 USD price
tag.

Re: Need IDE Taskfile Access

[Dave Jones]

On Wed, Jun 19, 2002 at 06:43:57PM -0400, Scott Tillman wrote:
 > I'm working with a group of people in an effort to get Linux running on the
 > XBox.  The XBox uses a set of security PIO commands to restrict access to
 > the IDE drive, requiring a 32 byte password to be delivered before sector
 > access is allowed.
 > 
 > Another comment/question (related to XBox support):
 > As part of this effort the xbox-linux team has coded support for the XBox's
 > proprietary partitioning and it's new filesystem.  This code (and any
 > further kernel support code) has been developed for the 2.4.18 kernel, and
 > we have no desire to port it to 2.5.x unless there is some hope of it's
 > adoption.  Could I get an official decision on whether this code might be
 > adopted if made available to the 2.5.x kernel?

I wonder about the legality of including such a port in the mainline kernel.
The IDE restriction sounds like it definitly comes under the
'circumventing an access control' clause of the DMCA.

        Dave

-- 
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

RE: Need IDE Taskfile Access

[Scott Tillman]

> I wonder about the legality of including such a port in the
> mainline kernel.
> The IDE restriction sounds like it definitly comes under the
> 'circumventing an access control' clause of the DMCA.

Well, IANAL, but if you are referring to the IDE's security mode commands,
they are just the access method...you still need the actual key.  If you
want to count that then things like a DMA controller, or a mouse could be
considered circumvention devices.  I think of it like the lock on a door.
The lock commands are the lock, the code to access the commands is the
*user* of the key.  You still need to have the actual key to unlock
anything.  What the DMCA is geared to prevent is the "opening" of the lock
by removing the hinges.

circumvention: To go around; bypass;
(to circumvent != to use) in fact they are basically antonyms.

The IDE security commands are no more a "circumvention device" than you are.
You use keys to provide access to things behind locks.  It uses keys to
provide access to things behind locks.  So, if it is illegal for me to make
code which can, when provided a key, unlock my hard drive then...Hey, I've
got a great idea...let's arrest all pregnant women for manufacturing a
circumvention device.

Sorry...It just irks me to see this knee-jerk reaction in fear of the DMCA,
whether it's called for or not.

-Scott Tillman aka SpeedBump

Re: Need IDE Taskfile Access

[H. Peter Anvin]

Followup to:  <CBELJEJGBEIGHCIMEDHNCEPBCIAA.tillman@viewcast.com>
By author:    "Scott Tillman" <tillman@viewcast.com>
In newsgroup: linux.dev.kernel
> 
> PS: flames about why we are supporting the XBox (a design of the Evil
> Empire) will be summarily ignored.  I can only point you to it's HDTV, NTSC,
> PAL, and possibly VGA outputs, it's dvd/cd drive, and it's $199 USD price
> tag.
> 

Not to mention M$ takes a loss for every hardware unit sold.

	-hpa
-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt	<amsp@zytor.com>

Re: Need IDE Taskfile Access

[Alan Cox]

> I'm working with a group of people in an effort to get Linux running on the
> XBox.  The XBox uses a set of security PIO commands to restrict access to
> the IDE drive, requiring a 32 byte password to be delivered before sector
> access is allowed.  As far as I can tell from my investigations and from
> earlier discussions with Andre there is currently no way to issue this
> command.  If I'm wrong in my estimation just let me know how, otherwise I
> simply wish add my voice to the (albeit small) outcry for supporting the
> entire ATA spec.

That would I suspect be something for the kerneli patch.

Re: Need IDE Taskfile Access

[Andries Brouwer]

On Fri, Jun 21, 2002 at 11:40:40PM +0100, Alan Cox wrote:

> > I'm working with a group of people in an effort to get Linux running on the
> > XBox.  The XBox uses a set of security PIO commands to restrict access to
> > the IDE drive, requiring a 32 byte password to be delivered before sector
> > access is allowed.  As far as I can tell from my investigations and from
> > earlier discussions with Andre there is currently no way to issue this
> > command.  If I'm wrong in my estimation just let me know how, otherwise I
> > simply wish add my voice to the (albeit small) outcry for supporting the
> > entire ATA spec.
> 
> That would I suspect be something for the kerneli patch.

No. Don't be misled by the "password". Translated into driver language
the question just is: I have a vendor-unique command that must transport
data to the disk. How? And if it doesnt work today the possibility must
be added.

(Long ago I had various small programs that did vendor-unique things
to ata floppies, zip drives, large ide drives and the like.
Commands for doorlocking, for switching between removable disk / big floppy
mode, for reading and setting the native max address, etc.
Unfortunately some random patch that accompanied one of these programs
got into the official tree, so that we now have the HDIO_DRIVE_CMD ioctl.
It does do_cmd_ioctl(). But that was special purpose for setmax,
not general purpose, and it does not transport data to the disk.
We need a slightly more general interface. Maybe we do have it already,
I have not watched this closely.)

Andries

Re: Need IDE Taskfile Access

[Alan Cox]

> Imagine our surprise when we discovered that taskfile access was being abandoned completely!
> Although we understand that the kernel may need to filter some commands, why can't applications access at least the Smart commands? Help!
> Regards,
> Garet Cammer
> Software Development
> Arco Computer Products
> > (954) 925-2688
> 

The 2.5 code is rather in flux, various things are changing including
redoing generalised ide/scsi generic command access to be a lot more
consistent that ioctl back doors.

I wouldnt worry about it just yet