Promise 20267 "working" but no UDMA

Promise 20267 "working" but no UDMA

[Andrew Chan]

Anybody manage to get UDMA 66/100 working with an on-board Promise 20267
chip?

Hardware: Tyan Tiger LE (with ServerWorks OSB4 _and_ Promise 20267 on-board)

Kernel: 2.4.3 with ide.2.4.3-p8.all.03242001.patch by Andre Hedrick (or
stock 2.4.3 with more or less same results)

FastTrack config: only 1 drive, configured as a SPAN volume consisting of 1
drive

I don't quite care about the Promise RAID features. I will use Linux
software RAID. The problem is that I cannot seem to be able to get the
controller into UDMA 4 (66 Mhz) mode!

I have enabled all the relevant DMA related kernel options. I have also
checked using the Seagate disk utility to make sure that the drive is
recognized (and configured) as UDMA 66 capable.

The following is from dmesg:

PCI: Found IRQ 10 for device 00:03.0
PDC20267: chipset revision 2
PDC20267: not 100% native mode: will probe irqs later
PDC20267: ROM enabled at 0xfeae0000
PDC20267: (U)DMA Burst Bit ENABLED Primary MASTER Mode Secondary MASTER
Mode.
PDC20267: neither IDE port enabled (BIOS)

I wonder what the last line meant by saying neither port is enabled?

# /sbin/hdparm -Tt /dev/hde

/dev/hde:
 Timing buffer-cache reads:   128 MB in  0.74 seconds =172.97 MB/sec
 Timing buffered disk reads:  64 MB in 12.82 seconds =  4.99 MB/sec [should
be much much faster here]

# cat /proc/ide/pdc202xx

                                PDC20267 Chipset.
------------------------------- General
Status ---------------------------------
Burst Mode                           : enabled
Host Mode                            : Normal
Bus Clocking                         : 33 PCI Internal
IO pad select                        : 10 mA
Status Polling Period                : 0
Interrupt Check Status Polling Delay : 0
--------------- Primary Channel ---------------- Secondary
Channel -------------
                enabled                          enabled
66 Clocking     enabled                          disabled
           Mode MASTER                      Mode MASTER
                FIFO Empty                       FIFO Empty
--------------- drive0 --------- drive1 -------- drive0 ----------
drive1 ------
DMA enabled:    no               no              yes               no
DMA Mode:       UDMA 4           NOTSET          NOTSET            NOTSET
PIO Mode:       PIO 4            NOTSET           NOTSET            NOTSET

# hdparm -d1 /dev/hde

/dev/hde:
 setting using_dma to 1 (on)
 HDIO_SET_DMA failed: Operation not permitted
 using_dma    =  0 (off)

# hdparm -X68 /dev/hde

[resulted in the following message in /var/log/messages]

Apr  1 19:03:21 promise kernel: ide2: Speed warnings UDMA 3/4/5 is not
functional.

# hdparm -i /dev/hde

/dev/hde:

 Model=ST36421A, FwRev=6.01, SerialNo=5BE064AN
 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
 RawCHS=13330/15/63, TrkSize=0, SectSize=0, ECCbytes=4
 BuffType=unknown, BuffSize=256kB, MaxMultSect=16, MultSect=off
 CurCHS=13330/15/63, CurSects=913440960, LBA=yes, LBAsects=12596850
 IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes: pio0 pio1 pio2 pio3 pio4
 DMA modes: mdma0 mdma1 *mdma2 udma0 udma1 udma2 udma3 *udma4

Many thanks for pointers!

Andrew Chan

Re: Promise 20267 "working" but no UDMA

[Andre Hedrick]

On Sun, 1 Apr 2001, Andrew Chan wrote:

> Anybody manage to get UDMA 66/100 working with an on-board Promise 20267
> chip?
> 
> Hardware: Tyan Tiger LE (with ServerWorks OSB4 _and_ Promise 20267 on-board)
> 
> Kernel: 2.4.3 with ide.2.4.3-p8.all.03242001.patch by Andre Hedrick (or
> stock 2.4.3 with more or less same results)
> 
> FastTrack config: only 1 drive, configured as a SPAN volume consisting of 1
> drive

No RAIDing allowed in the FTTK Bios.

> I don't quite care about the Promise RAID features. I will use Linux
> software RAID. The problem is that I cannot seem to be able to get the
> controller into UDMA 4 (66 Mhz) mode!
> 
> I have enabled all the relevant DMA related kernel options. I have also
> checked using the Seagate disk utility to make sure that the drive is
> recognized (and configured) as UDMA 66 capable.
> 
> The following is from dmesg:
> 
> PCI: Found IRQ 10 for device 00:03.0
> PDC20267: chipset revision 2
> PDC20267: not 100% native mode: will probe irqs later
> PDC20267: ROM enabled at 0xfeae0000
> PDC20267: (U)DMA Burst Bit ENABLED Primary MASTER Mode Secondary MASTER Mode.

This is a RAID mode BIOS and that is specified to disable under linux
unless a 0x55AA valid signature is present

> PDC20267: neither IDE port enabled (BIOS)

And it does!

> I wonder what the last line meant by saying neither port is enabled?
> 
> # /sbin/hdparm -Tt /dev/hde
> 
> /dev/hde:
>  Timing buffer-cache reads:   128 MB in  0.74 seconds =172.97 MB/sec
>  Timing buffered disk reads:  64 MB in 12.82 seconds =  4.99 MB/sec [should
> be much much faster here]
> 
> # cat /proc/ide/pdc202xx
> 
>                                 PDC20267 Chipset.
> ------------------------------- General
> Status ---------------------------------
> Burst Mode                           : enabled
> Host Mode                            : Normal
> Bus Clocking                         : 33 PCI Internal
> IO pad select                        : 10 mA
> Status Polling Period                : 0
> Interrupt Check Status Polling Delay : 0
> --------------- Primary Channel ---------------- Secondary
> Channel -------------
>                 enabled                          enabled
> 66 Clocking     enabled                          disabled
>            Mode MASTER                      Mode MASTER
>                 FIFO Empty                       FIFO Empty
> --------------- drive0 --------- drive1 -------- drive0 ----------
> drive1 ------
> DMA enabled:    no               no              yes               no
> DMA Mode:       UDMA 4           NOTSET          NOTSET            NOTSET
> PIO Mode:       PIO 4            NOTSET           NOTSET            NOTSET
> 
> # hdparm -d1 /dev/hde
> 
> /dev/hde:
>  setting using_dma to 1 (on)
>  HDIO_SET_DMA failed: Operation not permitted
>  using_dma    =  0 (off)
> 
> # hdparm -X68 /dev/hde
> 
> [resulted in the following message in /var/log/messages]
> 
> Apr  1 19:03:21 promise kernel: ide2: Speed warnings UDMA 3/4/5 is not
> functional.
> 
> # hdparm -i /dev/hde
> 
> /dev/hde:
> 
>  Model=ST36421A, FwRev=6.01, SerialNo=5BE064AN
>  Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
>  RawCHS=13330/15/63, TrkSize=0, SectSize=0, ECCbytes=4
>  BuffType=unknown, BuffSize=256kB, MaxMultSect=16, MultSect=off
>  CurCHS=13330/15/63, CurSects=913440960, LBA=yes, LBAsects=12596850
>  IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}
>  PIO modes: pio0 pio1 pio2 pio3 pio4
>  DMA modes: mdma0 mdma1 *mdma2 udma0 udma1 udma2 udma3 *udma4
                          ^^^^^^                         ^^^^^^

You have not enable the chipset code properly

Andre Hedrick
Linux ATA Development
ASL Kernel Development
-----------------------------------------------------------------------------
ASL, Inc.                                     Toll free: 1-877-ASL-3535
1757 Houret Court                             Fax: 1-408-941-2071
Milpitas, CA 95035                            Web: www.aslab.com

Re: Promise 20267 "working" but no UDMA

[Andrew Chan]

Andre Hedrick says:

>> FastTrack config: only 1 drive, configured as a SPAN volume consisting of
1 drive

> No RAIDing allowed in the FTTK Bios.

But my motherboard hangs at boot time (while Fasttrack tests for arrays) if
there is no array defined! There is a message from the Fasttrack bios that
says something like "no array found, press some key to continue". But I need
to remotely reboot these servers!

>> The following is from dmesg:
>>
>> PCI: Found IRQ 10 for device 00:03.0
>> PDC20267: chipset revision 2
>> PDC20267: not 100% native mode: will probe irqs later
>> PDC20267: ROM enabled at 0xfeae0000
>> PDC20267: (U)DMA Burst Bit ENABLED Primary MASTER Mode Secondary MASTER
Mode.

> This is a RAID mode BIOS and that is specified to disable under linux
> unless a 0x55AA valid signature is present

>> PDC20267: neither IDE port enabled (BIOS)

> And it does!

Which means? Can the Linux IDE driver be made to not deal with the Fasttrack
"raid" code and just treat the interface as normal non-raid ATA100
interfaces?

It is a real pity that Promise (the company) somehow doesn't seem to work
well with the Linux community. It is fast hardware (and I have 25 of them)
that lies useless for me right now.

Many thanks....

Andrew Chan

Re: Promise 20267 "working" but no UDMA

[Andrew Chan]

Actually, upon further testing, I found that the MB will move on without an
Fasttrack array. However, it will proceed to boot from PXE because it now
thinks that there is no hard disk attached to the entire system (as reported
by the Promise RAID chip)!

So,

1. Promise driver doesn't support SMP or 2.4.x kernels

2. Linux IDE code won't support the chip in any Promise RAID config, not
even single disk span

3. If not under RAID config, the motherboard thinks it has no hard disk, so
booting from disk is not possible even though the kernel can run the disks
in ATA100 mode

The remaining alternative is to boot from floppy (to load the kernel) and
specify the hard disk as root disk... :-(

I am most disappointed with Promise (not so much with Tyan).

Andrew

> >> FastTrack config: only 1 drive, configured as a SPAN volume
> >> consisting of 1 drive
>
> > No RAIDing allowed in the FTTK Bios.
>
> But my motherboard hangs at boot time (while Fasttrack tests for arrays)
> if there is no array defined! There is a message from the Fasttrack bios
> that says something like "no array found, press some key to continue".
> But I need to remotely reboot these servers!

RE: Promise 20267 "working" but no UDMA

[Nerijus Baliunas]

> > PDC20267: (U)DMA Burst Bit ENABLED Primary MASTER Mode Secondary MASTER
Mode.

How does MASTER mode differ from PCI?

I have:
PDC20267: IDE controller on PCI bus 00 dev 50
PDC20267: chipset revision 2
PDC20267: not 100% native mode: will probe irqs later
PDC20267: ROM enabled at 0xe8000000
PDC20267: (U)DMA Burst Bit ENABLED Primary PCI Mode Secondary PCI Mode.

Another question - I have Promise Ultra100 and 2 disks:

hda: QUANTUM FIREBALL CX10.2A, 9787MB w/418kB Cache, CHS=19885/16/63, UDMA(33)
hdc: IBM-DTLA-305030, 29314MB w/380kB Cache, CHS=59560/16/63, UDMA(100)

Why Quantum is shown as UDMA 33 when Promise BIOS shows it as UDMA 66?

Why DMA Mode:       UDMA 4 in /proc/ide/pdc202xx for IBM disk?
Shouldn't it be UDMA 5?

Regards,
Nerijus

Re: Promise 20267 "working" but no UDMA

[Ruth Ivimey-Cook]

At 12:13 PM 4/1/01, you wrote:
>Anybody manage to get UDMA 66/100 working with an on-board Promise 20267
>chip?

I'm a bit confused by this thread. I have two PDC2067 chips in my 
fileserver, both on Ultra100 boards and connected to 4 x 30Gb IBM deskstar 
disks. I have it set up as one 15GB and one ~75Gb partition using Linux 
RAID 5, and it works fine (as far as I can tell) with this config.

I do admit I have a fifth disk that has /bin, /etc, /boot on it for the 
purposes of booting, so I can boot single user and maintain the RAID array 
when necessary.

Is the issue here about the Promise FastTrack BIOS messing things up? If 
so, why use it, rather than the Promise Ultra100 BIOS, which seems to be 
fine (again, AFAIK).

Yours,

Ruth

-- 

Ruth 
Ivimey-Cook                       ruthc@sharra.demon.co.uk
Technical 
Author, ARM Ltd              ruth.ivimey-cook@arm.com

Re: Promise 20267 "working" but no UDMA

[Andrew Chan]

> Is the issue here about the Promise FastTrack BIOS messing things up? If
> so, why use it, rather than the Promise Ultra100 BIOS, which seems to be
> fine (again, AFAIK).

Ultra100 is okay.

FastTrack is not.

I have no choice since the motherboard has the chip on-board and with
FastTrack BIOS.

I am hoping Andre will come up with something so that I can use this
motherboard.

I don't care much about the Fasttrack RAID part but I do want to utilize its
ATA100 capability.

Re: Promise 20267 "working" but no UDMA

[Ruth Ivimey-Cook]

At 04:57 PM 4/3/01, you wrote:
>I have no choice since the motherboard has the chip on-board and with
>FastTrack BIOS.

Ahh. I understand.

I didn't know these MBs had the FastTrak BIOS built in; I was assuming you 
were using the PCI card.

Ruth
-- 

Ruth 
Ivimey-Cook                       ruthc@sharra.demon.co.uk
Technical 
Author, ARM Ltd              ruth.ivimey-cook@arm.com

Re: Promise 20267 "working" but no UDMA

[Juhani Rautiainen]

On Tue, 3 Apr 2001, Ruth Ivimey-Cook wrote:

> At 04:57 PM 4/3/01, you wrote:
> >I have no choice since the motherboard has the chip on-board and with
> >FastTrack BIOS.
> 
> Ahh. I understand.
> 
> I didn't know these MBs had the FastTrak BIOS built in; I was assuming you 
> were using the PCI card.

There are projects on the net which have modified Asus A7V BIOS so that
20265 works with FastTrack BIOS (it's same chip as 20267). Maybe that could
be done in reverse. For example check
http://www.tweakhardware.com/guide/a7v-raid/ 

Juhani 
-- 
Juhani Rautiainen			jrauti@iki.fi

RE: Promise 20267 "working" but no UDMA

[Andre Hedrick]

On Mon, 2 Apr 2001, Nerijus Baliunas wrote:

> > > PDC20267: (U)DMA Burst Bit ENABLED Primary MASTER Mode Secondary MASTER
> Mode.
> 
> How does MASTER mode differ from PCI?

Master mode is in BIOS RAID, and PCI is BIOS non-RAID
What I have not thought about doing is forcing all "MASTER" modes into
"PCI" modes.  In theory it may work because it changes the properties of
the chips behavior.

> I have:
> PDC20267: IDE controller on PCI bus 00 dev 50
> PDC20267: chipset revision 2
> PDC20267: not 100% native mode: will probe irqs later
> PDC20267: ROM enabled at 0xe8000000
> PDC20267: (U)DMA Burst Bit ENABLED Primary PCI Mode Secondary PCI Mode.
> 
> Another question - I have Promise Ultra100 and 2 disks:
> 
> hda: QUANTUM FIREBALL CX10.2A, 9787MB w/418kB Cache, CHS=19885/16/63, UDMA(33)
> hdc: IBM-DTLA-305030, 29314MB w/380kB Cache, CHS=59560/16/63, UDMA(100)
> 
> Why Quantum is shown as UDMA 33 when Promise BIOS shows it as UDMA 66?
> 
> Why DMA Mode:       UDMA 4 in /proc/ide/pdc202xx for IBM disk?
> Shouldn't it be UDMA 5?

This is what I love about Promise.
Mode 4 and Mode 5 have the exact same timings to the BUS; however, issuing
a setfeatures command to jump from Mode 4 to Mode 5, the card does voodoo
sensing.  It does internal tracking of the drive betweem these two modes
and auto-sets things in the chip.

If they can put that level of logic in the chips, it would be nice to have
it auto timing based on any setfeatures call to set the transfer rates.

Cheers,

Andre Hedrick
Linux ATA Development
ASL Kernel Development
-----------------------------------------------------------------------------
ASL, Inc.                                     Toll free: 1-877-ASL-3535
1757 Houret Court                             Fax: 1-408-941-2071
Milpitas, CA 95035                            Web: www.aslab.com