[PATCH] Promise FastTrack metadata detection

Hi,

we noticed that dmraid has not been able to detect the metadata for the
following card:
  Model: "Promise FastTrak100 TX2"
  Vendor: pci 0x105a "Promise Technology, Inc."
  Device: pci 0x6268 "PDC20270 (FastTrak100 LP/TX2/TX4)"
  SubVendor: pci 0x105a "Promise Technology, Inc."
  SubDevice: pci 0x4d68 "FastTrak100 TX2"

with a disk size > 128 GB.

The reason is that the BIOS setup tool only detects up to 268435377 sectors
(while the actual size of the tested device is 488397168 sectors).
So the metadata is written at sector 268435377, where it is not detected
by dmraid, because it has a list of offsets from the end of the device.

I am not sure if this limitation applies to more Promise cards. A patch which
works with our test machine is below. Of course it does not allow to use more
than the limit, but at least the metadata is detected and the device can be set
up.

Matthias

_______________________________________________
Ataraid-list mailing list
Ataraid-list <at> redhat.com

Re: Ali/Uli 528(8|9) BIOSRAID?

How to retrive metadata format specs?
Can I help with my mainboard ?
I'm going to read some previous posts.

   On Wed, Mar 28, 2007 at 05:08:38PM -0400, James McOrmond wrote:
   >    > Does anybody know if there's planned support for this BIOSRAID 
in an    > upcoming dmraid release?

   Should be me 

   Not yet, because I don't have metadata format specs for this.
   Any pointers to those or an OSS implementation would surely help this.

   >    > --    > James A. McOrmond (jamesm xandros com)
   > Hardware QA Lead & Network Administrator
   > Xandros Corporation, Ottawa, Canada.
   > Morpheus: ...after a century of war I remember that which matters 
most:
   >  *We are still HERE!*
   >    > _______________________________________________
   > Ataraid-list mailing list
   > Ataraid-list redhat com
   > https://www.redhat.com/mailman/listinfo/ataraid-list

   --
   Regards,
   Heinz    -- The LVM Guy --

   *** Software bugs are stupid.
       Nevertheless it needs not so stupid people to solve them ***

Re: [PATCH] Promise FastTrack metadata detection

On Wed, Sep 12, 2007 at 03:27:20PM +0200, Matthias Koenig wrote:

Hi Matthias,

can I have the metadata samples created with your patch applied
via "dmraid -rd" (tar/brzip2) ?

We got to do more than support recovery, ie. the mapping needs to work around
the strange metadata location.

Heinz

w
> Index: 1.0.0.rc14/lib/format/ataraid/pdc.c
> ===================================================================
> --- 1.0.0.rc14.orig/lib/format/ataraid/pdc.c
> +++ 1.0.0.rc14/lib/format/ataraid/pdc.c
>  <at>  <at>  -155,6 +155,21  <at>  <at>  static void *pdc_read_metadata(struct li
>  			}
>  		}
>  
> +		/* Some pdc card detect only PDC_MAX_SECTOR sectors, even
> +		 * if the disk is bigger, this prevents dmraid from discovering
> +		 * the metadata. This quirk searches for metadata at the
> +		 * sector PDC_MAX_SECTOR
> +		 */
> +		if (di->sectors >= PDC_MAX_SECTOR) {
> +			if (read_file(lc, handler, di->path, ret, sizeof(*ret),
> +				      PDC_MAX_SECTOR << 9) &&
> +			    !strncmp((const char*) ret->promise_id, PDC_MAGIC,

Re: Ali/Uli 528(8|9) BIOSRAID?

On Wed, Sep 12, 2007 at 05:33:33PM +0200, Cosimo Guglielmucci wrote:
> How to retrive metadata format specs?
> Can I help with my mainboard ?
> I'm going to read some previous posts.

One costly approach would be to try retrieving the last Megabyte off of
the end of each device in your RAID set plus the size in sectors of each
involved device.  BIOS properties for the device woud help to.

But this would be just a metadata sample involving painfull reengineering
rather than basing a format handler on vendor specs which is the preferred
way to go. That's what my quote below was aiming at.

Heinz

> 
>   On Wed, Mar 28, 2007 at 05:08:38PM -0400, James McOrmond wrote:
>   >    > Does anybody know if there's planned support for this BIOSRAID 
> in an    > upcoming dmraid release?
> 
>   Should be me 
> 
>   Not yet, because I don't have metadata format specs for this.
>   Any pointers to those or an OSS implementation would surely help this.
> 
>   >    > --    > James A. McOrmond (jamesm xandros com)
>   > Hardware QA Lead & Network Administrator
>   > Xandros Corporation, Ottawa, Canada.
>   > Morpheus: ...after a century of war I remember that which matters 
> most:

Re: Ali/Uli 528(8|9) BIOSRAID?

I was going to a simple dmraid -n deva devb
>> One costly approach would be to try retrieving the last Megabyte off of
>> the end of each device in your RAID set plus the size in sectors of each
>> involved device.  BIOS properties for the device woud help to.
>>     
I remember something from my school about ioctl() that can help me.
Have you a aimed tool?
Device is a Raptor 74GB, I'm going to search how many bios-es there are 
for it.

>> But this would be just a metadata sample involving painfull reengineering
>> rather than basing a format handler on vendor specs which is the preferred
>> way to go. That's what my quote below was aiming at.
>>
>> Heinz
>>     

ALI is just an nVidia branch now. Maybe it's possible to have specs if 
NVraid's one were released.
Thank you.
Cosimo

Ubuntu Gutsy Tribe5 compatibility problems

I'm trying to get dmraid working with the latest Ubuntu Gutsy
prerelease CD Tribe5. The first time I ran the Live CD everything worked 
fine, but after installing to the fakeraid it wouldn't boot (couldn't 
find /dev/mapper/pdc_bhifjacba7) and now even running from the Tribe5 
Live CD won't find the volumes either. Basically it finds the raid set 
but none of the volumes on it and there is nothing in /dev/mapper. If I 
stick in the Feisty Live DVD and boot, it finds everything no problem.

I've created a bug report with more details...
https://bugs.launchpad.net/ubuntu/+source/devmapper/+bug/139736

This looks very familiar to:
https://www.redhat.com/archives/ataraid-list/2007-July/msg00032.html
https://bugzilla.redhat.com/show_bug.cgi?id=253695

Re: Ubuntu Gutsy Tribe5 compatibility problems

On Mon, Sep 17, 2007 at 12:38:25AM -0400, David wrote:
> I'm trying to get dmraid working with the latest Ubuntu Gutsy
> prerelease CD Tribe5. The first time I ran the Live CD everything worked 
> fine, but after installing to the fakeraid it wouldn't boot (couldn't 
> find /dev/mapper/pdc_bhifjacba7) and now even running from the Tribe5 
> Live CD won't find the volumes either. Basically it finds the raid set 
> but none of the volumes on it and there is nothing in /dev/mapper. If I 
> stick in the Feisty Live DVD and boot, it finds everything no problem.

Not familiar with Ubuntu.
Looks like an initrd flaw not activating the volumes...

Any Ubuntu wizards ?

Heinz

> 
> I've created a bug report with more details...
> https://bugs.launchpad.net/ubuntu/+source/devmapper/+bug/139736
> 
> This looks very familiar to:
> https://www.redhat.com/archives/ataraid-list/2007-July/msg00032.html
> https://bugzilla.redhat.com/show_bug.cgi?id=253695
> 
> _______________________________________________
> Ataraid-list mailing list
> Ataraid-list <at> redhat.com
> https://www.redhat.com/mailman/listinfo/ataraid-list

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: Ubuntu Gutsy Tribe5 compatibility problems

AFAIK initrd involves installing to a volume, not running from a Live CD.

Right now if I just boot with the Tribe5 Live CD and install dmraid 
through synaptic, it will find the metadata no problem but won't mount 
the partitions to /dev/mapper.

If I boot with the Feisty Live DVD, and install dmraid, it will find 
both the metadata and create the volumes in /dev/mapper.

Thanks for your help!

Heinz Mauelshagen wrote:
On Mon, Sep 17, 2007 at 12:38:25AM -0400, David wrote:
 > I'm trying to get dmraid working with the latest Ubuntu Gutsy
 > prerelease CD Tribe5. The first time I ran the Live CD everything worked
 > fine, but after installing to the fakeraid it wouldn't boot (couldn't
 > find /dev/mapper/pdc_bhifjacba7) and now even running from the Tribe5
 > Live CD won't find the volumes either. Basically it finds the raid set
 > but none of the volumes on it and there is nothing in /dev/mapper. If I
 > stick in the Feisty Live DVD and boot, it finds everything no problem.

Not familiar with Ubuntu.
Looks like an initrd flaw not activating the volumes...

Any Ubuntu wizards ?

Heinz

 >
 > I've created a bug report with more details...

Re: Ubuntu Gutsy Tribe5 compatibility problems

_______________________________________________
Ataraid-list mailing list
Ataraid-list <at> redhat.com
https://www.redhat.com/mailman/listinfo/ataraid-list

Re: Ubuntu Gutsy Tribe5 compatibility problems

Well, that's the problem... When I run the Gutsy Tribe5 liveCD and 
install dmraid, the devices don't show up in /dev/mapper.

If I use the Feisty Live DVD, the devices show up no problem.

so... I think its pretty clear that there's a bug here in the Gutsy 
dmraid pre-release...

dmraid also works if I delete my linux partitions, but then... I have no 
linux partitions.

I've posted the output from running dmraid in debug mode, but not sure 
what to do next.

I thought I should try Gutsy and report a problem before the release...

Richard Bailey wrote:
When installing Ubuntu on dmraid you install dmraid in the live CD 
environment to get access to the devices under /dev/mapper, but before 
you reboot you should chroot into your newly installed system and 
install dmraid into the system so that the initrd that grub uses when 
you reboot has dmraid installed.

- Richard Bailey

On 9/17/07, ataraid-list-request <at> redhat.com < 
ataraid-list-request <at> redhat.com> wrote:

     Date: Mon, 17 Sep 2007 15:17:23 +0200
     From: Heinz Mauelshagen <mauelshagen <at> redhat.com>
     Subject: Re: Ubuntu Gutsy Tribe5 compatibility problems
     To: "ATARAID (eg, Promise Fasttrak,     Highpoint 370) related
             discussions" <ataraid-list <at> redhat.com>
     Message-ID: <20070917131723.GC7463 <at> redhat.com>
     Content-Type: text/plain; charset=us-ascii

     On Mon, Sep 17, 2007 at 12:38:25AM -0400, David wrote:
     > I'm trying to get dmraid working with the latest Ubuntu Gutsy
     > prerelease CD Tribe5. The first time I ran the Live CD everything 
worked
     > fine, but after installing to the fakeraid it wouldn't boot (couldn't
     > find /dev/mapper/pdc_bhifjacba7) and now even running from the Tribe5
     > Live CD won't find the volumes either. Basically it finds the 
raid set
     > but none of the volumes on it and there is nothing in 
/dev/mapper. If I
     > stick in the Feisty Live DVD and boot, it finds everything no 
problem.

     Not familiar with Ubuntu.
     Looks like an initrd flaw not activating the volumes...

     Any Ubuntu wizards ?

     Heinz

     >
     > I've created a bug report with more details...
     > https://bugs.launchpad.net/ubuntu/+source/devmapper/+bug/139736
     >
     > This looks very familiar to:
     > https://www.redhat.com/archives/ataraid-list/2007-July/msg00032.html
     > https://bugzilla.redhat.com/show_bug.cgi?id=253695
     >
     > _______________________________________________
     > Ataraid-list mailing list
     > Ataraid-list <at> redhat.com
     > https://www.redhat.com/mailman/listinfo/ataraid-list

 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

     Heinz Mauelshagen                                 Red Hat GmbH
     Consulting Development Engineer                   Am Sonnenhang 11
     Storage Development                               56242 Marienrachdorf
                                                       Germany
     Mauelshagen <at> RedHat.com                            PHONE +49  171 
7803392
                                                       FAX   +49 2626 924446

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

--

-- 
Reporter: What is your opinion on the obesity problem?
King: I prefer it to the famine problem
- Wizard of ID

_______________________________________________
Ataraid-list mailing list
Ataraid-list <at> redhat.com
https://www.redhat.com/mailman/listinfo/ataraid-list