That's true - Windows 2000 also identifies the Promise FastTrack 387 controller as a RAID SCSI device. Thus I don't see any problem with Linux treating this SATA/PATA device as SCSI, too.


OK - here goes... but first let's recap on what's available on this motherboard in terms of storage capabilities:

The ASUS P4P800-E Deluxe has two storage controllers, not one:
IMHO, the Promise driver treats its conroller as SCSI in order to avoid complications or contention between the Intel controller and the Promise one - especially when all drives connected are using the legacy IDE cable...

Now my topology: In order to utilize all what this board offers, I connected a single 160GB HDD to the Promise controller (master), a DVDRW (master) + CDRW (slave) to the 1st IDE channel on the Intel controller, and a DVDR drive (master) to the 2nd IDE channel of the Intel controller. So the picture should look like this:

Quote:

/dev/sda - 160GB HDD (master in the single channel in the Promise 20387 controller). /dev/hda - DVDRW (master in 1st IDE channel on Intel's ICH5R) /dev/hdb - CDRW (slave in 1st IDE channel on Intel's ICH5R) /dev/hdc - DVDR (master in 2nd IDE channel on Intel's ICH5R)

Is this what you were asking for?

Thanks for any help you can provide!
Alex

Unbelievable, but I think that I found a thread of hope that could solve this problem (I know about quite a few more users who are eagerly waiting for a solution to this problem).

So... before I loose the URL... Here is a great work done by someone here named Erik_FL.

It will take me some time to digest all of this and implement it myself in my system, but I believe that I will prevail.

Thanks everybody!
Alex

Interesting.
I have a 3 different PCs with ASUS P4P800 variants - never had a problem.
Only one still has Windoze (XP SP2) on it as a boot option - the otheres are Linux only. Various distros, including Dapper and Edgy (different boxes) work fine.
Maybe you just got a dud.

Two questions:

1. Is one of the variants a P4P800-E Deluxe (please note: the "-E" and the "Deluxe" here are critical for proper identifaction.
2. If so, is it configured the way I configured it (P-ATA HDD and 3 CD/DVD drives)?

If it were a dud, how come Windows 2000 works so great on it? And please note that even Linux works - just not as smoothly as Windoze.

BTW, I just checked my dmesg again, to see if Erik's patch is already included there, and it is!

Notice the "sata_promise PATA port found" in there (this is how I figured out that Erik's patch is there):

Quote:

[17179572.384000] sata_promise PATA port found [17179572.404000] ata1: SATA max UDMA/133 cmd 0xF8866200 ctl 0xF8866238 bmdma 0x0 irq 169 [17179572.404000] ata2: SATA max UDMA/133 cmd 0xF8866280 ctl 0xF88662B8 bmdma 0x0 irq 169 [17179572.404000] ata3: PATA max UDMA/133 cmd 0xF8866300 ctl 0xF8866338 bmdma 0x0 irq 169 [17179572.608000] ata1: no device found (phy stat 00000000) [17179572.608000] scsi0 : sata_promise [17179572.812000] ata2: no device found (phy stat 00000000) [17179572.812000] scsi1 : sata_promise [17179572.976000] ata3: dev 0 cfg 00:045a 49:2f00 82:346b 83:7f01 84:4003 85:3c69 86:3c01 87:4003 88:407f 93:600b [17179572.976000] ata3: dev 0 ATA-7, max UDMA/133, 312581808 sectors: LBA48 [17179572.976000] ata3: dev 0 configured for UDMA/133 [17179572.976000] sata_get_dev_handle: SATA dev addr=0x40000, handle=0x00000000 [17179572.976000] scsi2 : sata_promise [17179572.976000] Vendor: ATA Model: SAMSUNG SP1614N Rev: TM10 [17179572.976000] Type: Direct-Access ANSI SCSI revision: 05 [17179572.980000] Driver 'sd' needs updating - please use bus_type methods [17179572.984000] SCSI device sda: 312581808 512-byte hdwr sectors (160042 MB) [17179572.984000] SCSI device sda: drive cache: write back [17179572.984000] SCSI device sda: 312581808 512-byte hdwr sectors (160042 MB) [17179572.984000] SCSI device sda: drive cache: write back [17179572.984000] sda: sda2 sda3 sda4 < sda5 sda6 > [17179573.024000] sd 2:0:0:0: Attached scsi disk sda

So... perhaps the problem is not in the Promise 20387 controller, but somewhere else in the way Linux supports the motherboard? Perhaps it is in the ACPI IRQ routing?

Thanks,
Alex

Sorry, No ...
Used to have one - my main machine actually. The gum tree across the road took a lightning strike, and it fried that motherboard along with a couple of hard disks (yes, I had a UPS).
I have a P4P800S-X in that box now.

Maybe I was lucky to lose that one ...

When I started linux ~1 year ago I researched mobo available,
and chose an ASUS A7N8X-E deluxe
because everything worked out of the box.
This is only one sample, one person.

Sorry I cannot provide technical input into that irq/bios whatever problem

As said before, no need to PM the HCL thread starter for your mobo,
you could post to that thread directly asking for more details,
like bios version used, whether sata drive were used, etc

I regret not doing the same (although how could have I known that the P4P800-E Deluxe was made for Windows 2000/XP only?).

What is "to PM the HCL thread starter"?

Thanks,
Alex

Have you tried any other distro? I found a few postings with people running fedora with that hardware.

Yes. At the end of my original posting (that started this thread) I wrote:
Note the subtlety: It's not that it won't boot at all, it's just it won't run without errors (pretty frustrating when the main reason I am trying to move from Windows to Linux is the desire to run in a safer and more stable environment).

I submitted a bug report to the Ubuntu developers (although IMHO it's in the kernel, shared by all distros).

Alex

Sorry for not keeping up... Can you provide a brief summary of your current situation (and what's going wrong).

yor tried PM=personal message
instead just post to the hcl thread for the bios details etc
even maybe to put a link to this thread

Maybe you're pisssed off at the world, but this is patently not true.
I ran Gentoo on that mobo fine for ages - across several (2.6) kernels. And if there was an issue with performance not being up to scratch I would have known.

No, I am not pissed off at all.

I am just trying to find a solution to this problem - a solution that will satisfy my high standards.

Perhaps I am naive, but I believe that there must be a solution out there. How do I know? Because I am running Windows 2000 perfectly on it. And I trust Linux to be able to provide everything that Windows can provide - and more.

All solutions that I have received so far suggest (in different phrasings):
1. Dump this mobo - or...
2. Dump Linux on this mobo - or...
3. Disable feature X, Y and/or Z on this mobo.

I understand that you were able to run linux on this mobo without following any of the above 3 "solutions". Can you show me how?

Thanks!
Alex

Fom memory I did nothing out of the ordinary.
H/T for the chip I had (2.6 GHz P4 ???) wasn't supported so I flashed it as soon as I got it.
I got the board specifically because it supported both P-ATA and S-ATA, and I had 2 of each in it. Ran like clockwork.

I was able to help xp_newbie solve the problem. I believe he intends to post details about the solution. Basically, the problem has to do with the possible settings for the Intel ICH5 IDE/SATA controller chip.

The ICH5 can operate in a "Compatible" mode that makes it appear to be a Primary and Secondary IDE controller, or it can operate in an "Enhanced" mode that appears to be one PCI device with Three disk controllers. The problem is that in "Enhanced" mode all three disk controllers share an interrupt request. If one has a mixture of Parallel ATA (IDE) drives and SATA drives on the ICH5 then two drivers will be loaded by Linux, "ide" and "sata_piix". Since the two drivers can't share an interrupt request, the "Enhanced" mode won't work. Using the "Compatible" mode will work because the Primary and Secondary controllers are assigned separate interrupt requests 14 and 15. Each driver will report that one of the possible controllers is in use, since the other driver will have claimed that controller. That causes no harm, and each driver uses the interrupt request for the controller that it claims.

In the "Compatible" mode the ICH5 only supports four disk drives in the following configurations.

Primary (IRQ 14) - PATA Master / PATA Slave
Secondary (IRQ 15) - First SATA port / Second SATA port

- or -

Primary (IRQ 14) - First SATA port / Second SATA port
Secondary (IRQ 15) - PATA Master / PATA Slave

- or -

Primary (IRQ 14) - PATA Master / PATA Slave
Secondary (IRQ 15) - PATA Master / PATA Slave

In the "Enhanced Mode" (unsupported on Linux) the ICH5 supports six disk drives.

Controller 1 (PCI INTA#) - PATA Master / PATA Slave
Controller 2 (PCI INTA#) - PATA Master / PATA Slave
Controller 3 (PCI INTA#) - First SATA port / Second SATA port

There are other possible configurations that may be supported by Linux.

Primary (IRQ 14) - PATA Master / PATA Slave
Secondary (IRQ 15) - PATA Master / PATA Slave
Controller 3 (PCI INTA#) - First SATA port / Second SATA port

Controller 3 can be either of the IDE channels or the SATA devices if the BIOS supports those configurations. In theory they will put Controller 3 on a PCI interrupt request separate from the Primary and Secondary IDE controllers. I haven't tested that configuration so I don't know for a fact that it will work.

Intel's documentation for the ICH5 does a reasonably good job of explaining these possible configurations.