Bug 176249

Created attachment 83772 [details]
hwinfo.txt

Created attachment 83773 [details]
y2logs.tgz

Some more correct information:
The HDDs are but ATA HDDs and  NOT in RAID mode

They are detected by the kernel all right. Thomas, please have a look at the logs.

hwinfo only shows me data about /dev/sda to /dev/sdd (which seem to be USB card 
reader slots) as devices without media, no other disks.
libstorage itself also detects no additional disk.

Could you attach the content of the sysfs subdirectory for the disk in question.

sata_promise says:

  <7>sata_promise 0000:0a:02.0: version 1.03
  <6>GSI 18 sharing vector 0xC9 and IRQ 18
  <6>ACPI: PCI Interrupt 0000:0a:02.0[A] -> GSI 19 (level, low) -> IRQ 201
  <6>ata1: SATA max UDMA/133 cmd 0xFFFFC20000030200 ctl 0xFFFFC20000030238 bmdma 0x0 irq 201
  <6>ata2: SATA max UDMA/133 cmd 0xFFFFC20000030280 ctl 0xFFFFC200000302B8 bmdma 0x0 irq 201
  <6>ata1: SATA link down (SStatus 0)
  <6>scsi0 : sata_promise
  <6>ata2: SATA link down (SStatus 0)
  <6>scsi1 : sata_promise

@#5 : 
Yes there is an integrated USB card reader.
Could you explain where the sysfs is and of which disk?

Steffen, I guess he has no devices on the SATA ports but only on the PATA ports. Ridvan, please attach a full dmesg from a booted 9.3 system that worked for you.

Created attachment 84155 [details]
dmesg from SuSE 9.3 x86_64 live

Created attachment 84156 [details]
Yast screenshot showing the partitions

Ok, so you have two PATA devices attached to the parallel port and nothing on the two SATA ports. Somehow 10.1 misses the PATA port completely.

Ok, looked further into this. For 9.3, we had the libata devel branch included which we don't have in 10.1. The PATA on SATA controller support is still missing from mainline, so we don't have it.

It's not up to me to decide whether we should include this (after 10.1 is final).

Something for Thorsten to decide.  Jens, please explain the effects for SLE10.

SLES9 didn't support PATA-on-SATA either, so it's not a regression from SLES9 -> SLES10.

Was it in SLES9 SP3?

What impact will such a patch have on the rest?  I fear it's too late to add this now - and this change should really go upstream...

It was not in any SLES9, nor the later service packs.

The sata_promise change itself is pretty minor, however it depends upon other libata changes. I'd prefer not to do it, I'm afraid we can't easily support this for 10.1 at this late time. For 10.2, we could.

*** Bug 116364 has been marked as a duplicate of this bug. ***

*** Bug 148638 has been marked as a duplicate of this bug. ***

I clearly don't understand SUSE's bug handling procedure. The bug is closed with WONTFIX as I understood because "I'm afraid we can't easily support
this for 10.1 at this late time. For 10.2, we could". But as there is no open bug for this issue, you will forget it and when somebody reports the bug for 10.2, you say again that is too late for 10.2, maybe for a later version.
The bug was reported on 2005-09-10 and 2006-02-07. I posted a patch for 10.0 and updated to 10.1 while it was in beta stage. I use and work fine in 10.1 final. If you don't want to apply, fine, but than don't close the bug, as it is still present. I better update with each tested version if the fix went into the kernel or not, so the reports don't just dissappear in the hyperspace.

Let's move this to 10.2.

Lets change it to a feature request then, not a blocker.

Bring the libata Patch from Andras Mantia into a regulary YOU Kernel Update,
so it will be possible to solve the really blocking situation for installation
of SuSE10.1 on Motherboards with promise ATA/SATA controller!
There should also be generated a boot.iso image for network installation,
which contains this patch.
But after installing the SuSE 10.1 from the Installation DVD, there should be
also available a kernel update which contains this patch, because i fear, by continuing the installation process from DVD, the patched boot.iso kernel will be overwritten by an old one, which does not work after restarting the PC. So you have to do an You Kernel update before restarting the PC. 

*** Bug 187632 has been marked as a duplicate of this bug. ***

Jens, is the patch now in the upstream kernel so that we get it with our next kernel update?

Nope, it's not in 2.6.17 nor in 2.6.18-rc1.

How can we get it into vanilla 2.6.18/2.6.19?

Well someone needs to submit a patch for review... I don't have the hardware nor the time to pursue it, I'm afraid. I'd suggest one of the users here just sent the patch upstream to linux-ide@vger.kernel.org (and cc Jeff Garzik) and ask for inclusion, I'm sure he'd point out what needs changing or if he already has similar functionality in his git tree queued for inclusion.

There are three more bugs related to this issue:

Bug 170097 - Promise SATA 300-TX2 plus does not see any SATA disk

Bug 176249 - PDC20378 (FastTrak 378/SATA 378) can not be recognized during installation

Bug 206889 - Will not operate Promise Fasttrack 378 Controller

FYI:  Fedora Core 6 Test 3 can operate the controller either in RAID or PATA mode.

To my knowledge, this controller has been broken since 10.0, 9.3 worked.  I manually patched a vanilla kernel to get openSUSE 10 to work.

Created attachment 103107 [details]
Patch for 2.6.18 as in SUSE 10.2 Beta1

Patch for 2.6.18 that makes the PATA part of the controller work. Sincerely, the only change that is not so nice is in libata.h, because in theory it might affect other drivers as well, but a grep for _port_flags shows that only the sata_promise driver is using it, so I'd assume this patch is safe (not talking about the fact that I use since 10.0 came out).

Just talked to Hannes, he will take care of this.

As I see this is solved for Beta2. In that case you can close it. ;-)

Gladly.