Unix True 64 - PWS 500au does not boot Tru64 from CDROM

Author

PWS 500au does not boot Tru64 from CDROM

Stefaan A Eeckels

2004-09-02, 6:49 pm

Hi,

I'm trying to install Tru64 5.1A on a PWS500au. It has
768MB RAM, 2 SCSI disks (one 4GB, the other 8GB). The
CD-ROM is a 32x TEAC ATAPI device. The SCSI controller
is a Qlogic ISP 1040B/V2 (as indicated by the boot process).
The graphics card is a PowerStorm or suchlike, it has a
little rotary switch set to position '5'.
I've successfully upgraded the SRM to V7.2_1.

When I boot from the CDROM, I get a message "init_rootdev:
boot device translation failed" and the system halts. During
the process leading up to this message, I get a couple of
spurious interrupt messages.

I've tried to update the I/O adapter firmware, but this
doesn't work. As soon as I select the [alpha500]as500_pci_v7_2.exe
file to load, the machine enters a reboot loop, the ATAPI
bus seems to reset every second or so (clicks from the CDROM
like during the boot process, just continuous) and I have to
power-cycle to stop the behaviour.

I've also tried to remove the graphics card, and use the serial
console. This works OK until I try to boot from the CDROM. It
says "Jumping to bootstrap image", and then nothing appears on
the serial console (I did a "set console serial"), even though
the system is apparently booting.

Finally, when the PWS500au starts up, the CDROM drive seems
to be reset several times, but the system does boot into SRM.
When I got the machine it booted into Windows 2000 Advanced
Server, and it boots and runs OK with this OS.

Thanks for any pointers on how to solve this.

Take care,

--
Stefaan
--
"What is stated clearly conceives easily." -- Inspired sales droid

Mario Stargard

2004-09-02, 6:49 pm

Stefaan A Eeckels wrote:
> Hi,
>
> I'm trying to install Tru64 5.1A on a PWS500au. It has
> 768MB RAM, 2 SCSI disks (one 4GB, the other 8GB). The
> CD-ROM is a 32x TEAC ATAPI device. The SCSI controller
> is a Qlogic ISP 1040B/V2 (as indicated by the boot process).
> The graphics card is a PowerStorm or suchlike, it has a
> little rotary switch set to position '5'.
> I've successfully upgraded the SRM to V7.2_1.
>
> When I boot from the CDROM, I get a message "init_rootdev:
> boot device translation failed" and the system halts. During
> the process leading up to this message, I get a couple of
> spurious interrupt messages.
>
> I've tried to update the I/O adapter firmware, but this
> doesn't work. As soon as I select the [alpha500]as500_pci_v7_2.exe
> file to load, the machine enters a reboot loop, the ATAPI
> bus seems to reset every second or so (clicks from the CDROM
> like during the boot process, just continuous) and I have to
> power-cycle to stop the behaviour.
>
> I've also tried to remove the graphics card, and use the serial
> console. This works OK until I try to boot from the CDROM. It
> says "Jumping to bootstrap image", and then nothing appears on
> the serial console (I did a "set console serial"), even though
> the system is apparently booting.
>
> Finally, when the PWS500au starts up, the CDROM drive seems
> to be reset several times, but the system does boot into SRM.
> When I got the machine it booted into Windows 2000 Advanced
> Server, and it boots and runs OK with this OS.
>
> Thanks for any pointers on how to solve this.
>
> Take care,
>
I would imagine that 5.1a doesn't like the atapi drive. Have you tried
5.1b?

I managed to "shoe-horn" 5.1a onto a SmartArray by installing it with
patches and the new hardware disk onto an older system with a simple
scsi disk, then dumping and restoring onto the new system. This
requires deleting and recreating the hardware database that is in /etc,
but it might be an option to get the OS installed on your system.

Another option would be to try and find a scsi cdrom drive.

Win2K server ran on Alpha?

Mario

Stefaan A Eeckels

2004-09-02, 6:49 pm

On Mon, 30 Aug 2004 22:49:40 -0400
Mario Stargard <mstargard@myprivacy.ca> wrote:

Thanks for the quick reply. It's much appreciated.

> I would imagine that 5.1a doesn't like the atapi drive. Have you tried
> 5.1b?

I only have 5.1a (the hobby license I bought a few years ago).

> I managed to "shoe-horn" 5.1a onto a SmartArray by installing it with
> patches and the new hardware disk onto an older system with a simple
> scsi disk, then dumping and restoring onto the new system. This
> requires deleting and recreating the hardware database that is in /etc,
> but it might be an option to get the OS installed on your system.
>
> Another option would be to try and find a scsi cdrom drive.

I thought of that (I've a lot of Plextors lying around
gathering dust). But then I happened to stumble onto
a google cached article (the original is no longer) by
Albert Chin stating that Tru64 can only boot from the
first CDROM on the first IDE interface. My CDROM was
set to "slave" mode (dka100), and lo and behold, after
changing it to "master" the system came up without a
hitch. It's now happily running Tru64 5.1a, and I am
a happy (if somewhat tired) Tru64 user.

> Win2K server ran on Alpha?

Indeed it did - this was a beta version, the previous owner
told me he used the box for beta testing MS Alpha versions.
I don't know if he still has the W2K CD, if you're interesed
I can send you his email address.

Thanks again!

--
Stefaan
--
"What is stated clearly conceives easily." -- Inspired sales droid

www.quincey.com

2004-09-22, 9:24 pm

Stefaan A Eeckels <tengo@DELETEMEecc.lu> wrote in message news:<20040831100636.39bccd56.tengo@DELETEMEecc.lu>...
> On Mon, 30 Aug 2004 22:49:40 -0400
> Mario Stargard <mstargard@myprivacy.ca> wrote:
>
> Thanks for the quick reply. It's much appreciated.
>
>
> I only have 5.1a (the hobby license I bought a few years ago).
>
>
> I thought of that (I've a lot of Plextors lying around
> gathering dust). But then I happened to stumble onto
> a google cached article (the original is no longer) by
> Albert Chin stating that Tru64 can only boot from the
> first CDROM on the first IDE interface. My CDROM was
> set to "slave" mode (dka100), and lo and behold, after
> changing it to "master" the system came up without a
> hitch. It's now happily running Tru64 5.1a, and I am
> a happy (if somewhat tired) Tru64 user.
>
>
> Indeed it did - this was a beta version, the previous owner
> told me he used the box for beta testing MS Alpha versions.
> I don't know if he still has the W2K CD, if you're interesed
> I can send you his email address.
>
> Thanks again!
>
> --
> Stefaan

Stefaan,

The problem is not the CDROM or Systems in ability to support Tru64 UNIX.
The problem is a misconfigured console variable.

boodef_dev or
boot_osflags or
os_type

The code that generated the message is:
/*
* We have the console name of the boot device in bootdevice. Now
* let's translate that into some Digital UNIX information that we
* need. We call device_translate for this.
*/
dt_req.dt_version = 1;
dt_req.dt_flags = 0;
dt_req.dt_opcode = DT_GET_INFO;
dt_req.dt_console_str = bootdevice;
/* We don't want the dsf name */
dt_req.dt_dsf_name = NULL;
/*
* If the call to translate the boot device fails, we will print
* an error message and panic.
*/
if (device_translate(&dt_req) == FALSE) {
printf("init_rootdev: failed to translate boot device %s.\n",
bootdevice);
panic("init_rootdev: boot device translation failed");
}

init_rootdev() failed to translate the boot device.
*/

Good luck,

Quincey

Stefaan A Eeckels

2004-09-22, 9:24 pm

On 21 Sep 2004 21:25:42 -0700
quinceyc@hotmail.com (www.quincey.com) wrote:

> The problem is not the CDROM or Systems in ability to support Tru64
> UNIX. The problem is a misconfigured console variable.
>
> bootdef_dev or
> boot_osflags or
> os_type

Well, the CD-ROM was configured as a slave device on the
IDE bus, and what was happening is that while the boot loader
had no problems loading the OS from the CD-ROM, the kernel
apparently could not access it to finish the boot process.
[vbcol=seagreen]

As soon as I changed the CD-ROM to master, the boot process
completed without a hitch. Neither boot_osflags nor os_type
were changed, but bootdef_dev changed from the IDE slave CD-ROM
device to the IDE master CD-ROM device, and suddenly
device_translate() managed to do its job.

I think that it's fair to conclude that Tru64 5.1A cannot be
installed from a slave IDE CD-ROM, unless you know of some
SRM settings that specifically address this issue.

Take care,

--
Stefaan
--
"What is stated clearly conceives easily." -- Inspired sales droid

www.quincey.com

2004-09-22, 9:24 pm

Stefaan A Eeckels <tengo@DELETEMEecc.lu> wrote in message news:<20040922094454.712097f3.tengo@DELETEMEecc.lu>...
> On 21 Sep 2004 21:25:42 -0700
> quinceyc@hotmail.com (www.quincey.com) wrote:
>
>
> Well, the CD-ROM was configured as a slave device on the
> IDE bus, and what was happening is that while the boot loader
> had no problems loading the OS from the CD-ROM, the kernel
> apparently could not access it to finish the boot process.
>
>
> As soon as I changed the CD-ROM to master, the boot process
> completed without a hitch. Neither boot_osflags nor os_type
> were changed, but bootdef_dev changed from the IDE slave CD-ROM
> device to the IDE master CD-ROM device, and suddenly
> device_translate() managed to do its job.
>
> I think that it's fair to conclude that Tru64 5.1A cannot be
> installed from a slave IDE CD-ROM, unless you know of some
> SRM settings that specifically address this issue.
>
> Take care,
>
> --
> Stefaan

Stefaan,

You should see somthng like the following:
from
p00>> show config

Bus 00 Slot 07: Cypress PCI Peripheral Controller

Bus 00 Slot 07: Function 1: PCI IDE

Bus 00 Slot 07: Function 2: PCI IDE
dqb0.0.0.207.0 TOSHIBA CD-ROM XM-6302B

If you had 4 CDROM they would look similar to:

P00>>>show device
....
dqa0.0.0.207.x DQA0 TOSHIBA CD-ROM XM-6302B 1017
dqc0.0.0.207.x DQC0 TOSHIBA CD-ROM XM-6302B 1017
dqd0.0.0.207.x DQD0 TOSHIBA CD-ROM XM-6302B 1017
dqe0.0.0.207.x DQE0 TOSHIBA CD-ROM XM-6302B 1017
....

take a look at:

Table 7–4 Device Naming Conventions
Category Description
dq Driver ID Two-letter designator of port or class driver
dk SCSI drive or CD ew Ethernet port
dq IDE CD-ROM fw FDDI device
dr RAID set device mk SCSI tape
ci DSSI disk ci DSSI tape
ei Ethernet port pk SCSI port

It breaks down like this:
dqxx.x.x.x.x

a Storage adapter ID One-letter designator of storage adapter
(a, b, c…).
0 Device unit number Unique number (MSCP unit number). SCSI unit numbers
are forced to 100 X node ID.
0 Bus node number Bus node ID.
0 Channel number Used for multi-channel devices.
15 Logical slot number The slot number assigned by the firmware.
0 Hose number The hose number assigned by the firmware.

All this information is available in the user guides

Without seeing what you have as a boot command I'm not sure what going wrong.
It should but any properly configured device.

Quincey

Stefaan A Eeckels

2004-09-23, 9:13 am

On 22 Sep 2004 15:14:37 -0700
quinceyc@hotmail.com (www.quincey.com) wrote:

> Without seeing what you have as a boot command I'm not sure what going
> wrong. It should boot any properly configured device.

Mind you, nothing is going wrong :-). I had problems installing
from an ATAPI (IDE) CD-ROM connected as a slave on a machine
with freshly upgraded firmware.
When I tried to install from the ATAPI slave device, it booted
correctly (locating and loading the kernel, recognising devices,
etc.), but when it came to mounting the root device (which should
be the CD-ROM as the disks were empty) it failed.
I suppose that the kernel on the CD-ROM somehow doesn't know how
to deal with an ATAPI slave drive. Something with the SCSI
emulation maybe.

Anyhow, it doesn't matter anymore (unless as an exercise to
understand what happened) because a simple operation (setting
the jumper on the CD-ROM to "master") fixed the problem.

Thanks and take care,

--
Stefaan
--
"What is stated clearly conceives easily." -- Inspired sales droid