 Re: New PCI QFE 501-5406 card in Ultra 30, for Don and others! :-)
On 2008-08-21, Barry L. Bond wrote:
>
> Hi DoN!
[ ... ]
>> Look on the back. You would either have a 13W3 connector (DB-25
>>size, but thee coax pins, and ten single conductor pins) in the first or
>>second slot under the audio slot, or a card with a VGA style 15-pin D
>>series connector in one of the bottom four slots. Whatever your monitor
>>plugs into. Sun calls these "framebuffers", while other just call them
>>"graphics cards". If you *have* one, and you're getting the error
>>message, it may have been damaged. You might try pulling it, as long as
>>you are using a plain ASCII terminal for the console instead of the Sun
>>keyboard and monitor. As long as you don't have a Sun keyboard plugged
>>in, you don't need the monitor anyway. :-)
>
> Yes. I have a 13W3 connector, in the "0" slot near the bottom with
> what looks like a monitor symbol next to it.
O.K. That is likely to be one of the Creator series or perhaps
an Elite-3D. Does it have a small DIN connector down close to the tab
of the card bracket? (Not that this is important for what we are
dealing with.)
> And, I have an adapter that
> allows use of a VGA (a "regular" PC) monitor), which is what I'm using.
O.K.
> I do have a monitor and a keyboard plugged into appropriate slots on
> the back of the Ultra 30. In fact, I had never connected a terminal to
> TTYA or TTYB until I did so, responding to one of your early replies to me
> after the lightning strike.
O.K. With a serial terminal, you don't need the monitor or
keyboard at all. The terminal's "break" key takes the place of the
"Stop-A" pair on the keyboard.
But that depends on the serial port working, and IIRC that also
is zapped.
>> You said something below about the monitor -- so you probably
>>do. As long as it works to display the system's status on boot with a
>>keyboard connected, you can perhaps ignore the error message.
>
> It displays the systems status on boot, and then provides a login.
> With the "disk" drive I'm using now, it is an ASCII-style login prompt. I
> can type root or barry and log into it, normally. (With my "disk1", the
> larger hard drive, after it boots it displays the Common Desktop graphical
> login. I didn't install the graphical stuff the first time I installed
> Solaris, because it was only a 4.2G disk.)
O.K.
> I am ignoring it -- at least right now! :-)
>
>>> I get:
>>>
>>><<>>
>>> 0 PCI 33 1 network-SUNW,hme
>>><<>>
>>
>> What about any qfe possible entries?
>
> Good question! :-D
>
>>>>So you should see five of these -- one for the hme0 in the system board,
>>>>and the other four from the qfe card.
>>>
>>> Nope. Only one!
>>>
>>> (The path_to_inst file had entries for them all, but not that
>>> command.)
>>
>> Interesting. What numbers did the path_to_inst file have? (It
>>is made more difficult by my not having a system running Solaris 8, or a
>>system with both Solaris and a QFE card installed to compare.)
>
> Here is an excerpt of six lines from my path_to_inst file:
>
><<>>
> "/pci@1f,4000/pci@2/SUNW,qfe@2,1" 2 "qfe"
> "/pci@1f,4000/pci@2/SUNW,qfe@3,1" 3 "qfe"
> "/pci@1f,4000/pci@2/SUNW,qfe@0,1" 0 "qfe"
> "/pci@1f,4000/pci@2/SUNW,qfe@1,1" 1 "qfe"
> "/pci@1f,4000/network1,1" 0 "hme"
> "/pci@1f,4000/SUNW,hme@2,1" 1 "hme"
><<>>
O.K. Interesting the order that these are in, but the instance
number matches the port number in the card's name.
So -- perhaps it should have hostname.qfe0 and hostname.qfe1 in
place of the hme versions? Remember that I'm using the qfe on an
OpenBSD system, which calls them *all* "hme?".
I tried to spot the spare qfe card which I have somewhere to
stuff it into the spare Sun Blade 1000 which I am currently
experimenting with to see what it shows up as.
>>> Yes. A "man qfe" shows a man page with "qfe - SUNW,qfe Quad
>>> Fast-Ethernet device driver" under NAME.
>>
>> Read through it, and see if it mentions anything special needed.
>>Perhaps it is something which should be in /opt/SUNWqfe which is only
>>installed if the card is detected during installation, just like the
>>/opt/SUNWleo for a particular framebuffer in the SPARCstation 5 which I
>>am still running.
>
> Reading through it...
>
> I see:
>
><<>>
> The qfe driver is a "style 2" data link service provider. All M_PROTO and
> M_PCPROTO type messages are interpreted as DLPI primitives. Valid DLPI
> primitives are defined in . Refer to dlpi(7P) for more
> information. An explicit DL_ATTACH_REQ message by the user is required to
> associate the opened stream with a particular device (ppa). The ppa ID is
> interpreted as an unsigned long data type and indicates the corresponding
> device instance (unit) number.
>
> Devices
>
> The driver returns an error (DL_ERROR_ACK) if the ppa field value does not
> correspond to a valid device instance number for this system. The device
> is initialized on first attach and de-initialized (stopped) at last
> detach.
><<>>
>
> I can't say that I see anything like "this is needed". I excerpted
> the above because the DL_ERROR_ACK are the errors I am getting.
O.K.
> It mentions the /dev/qfe "file", which I have.
Good.
[ ... ]
> Do you still have the install CDs or DVDs?
>
> Yes.
>
> So, look on it/them for SUNWqfe? I'm checking, now...
>
>
>
> At first, when I inserted the Solaris 8 Documentation CD, the drive
> drawer stayed closed. But, I couldn't see anything that happened. I
> checked "mount" and I couldn't see anything different.
[ ... lots of problems with CD-ROM drive (or is it really the
SCSI DVD-ROM which is in some Ultra-60 machines? ]
Anyway -- is vold running? Try:
ps -ae | grep 'vold'
If so, it is trying to mount the CD-ROM where it thinks it
belongs (and you need to have the proper mount point present). To turn
to off, try:
sh /etc/init.d/volmgt stop
and once that is done, edit the /etc/vfstab file to add the line:
================================================== ====================
/dev/sr0 /dev/rsr0 /cdrom hsfs - no ro
================================================== ====================
to the file, then make sure to create the directory /cdrom.
Then (as long as "vold" is not running you can mount the CD
using "mount /cdrom" and get rid of it using "umount /cdrom".
> Is there any way I can get the SUNWqfe that I guess may have been put
> in the /opt directory if I had had the QFE card installed when I last
> installed Solaris, other than on the CD?
Well ... not sure whether it should be there or not.
> The ONLY THING I need at this point is something else to delay me
> if I have to replace the CD drive!
>
> The light blinks, as normal, when the drawer is closed. But, it
> opens again. (Perhaps the CD is not actually "spinning up?")
>
> Now, the drawer is even opening again even if I close it with NO CD
> in the drive!
>
> Doing a find looking for qfe in file names, though, I see these:
>
><<>>
> /var/sadm/pkg/SUNWqfed
> /var/sadm/pkg/SUNWqfedx
> /var/sadm/pkg/SUNWqfedu
><<>>
O.K. Solaris 10 has only /var/sadm/pkg/SUNWqfed and
/var/sadm/pkg/SUNWqfedu. Not sure what the 'x' suffix version does. It
is also not present in the Solaris 2.6 version on a SS-5.
> Might that be them, just not in /opt?
That probably says that they have already been installed.
Look in: /kernel/drv/sparcv9/qfe
if it is there, it is installed -- for the 64-bit mode of the CPUs only.
In the Solaris 2.6, it is only:
/kernel/drv/qfe
BTW While testing these, have you had ethernet cables connecting
two of the QFE ports to the hub or router which you have? If so
do you see leds lit on both the hub and the QFE card?
I wonder whether the QFE card is bad?
>>> I got:
>>>
>>><<>>
>>> lo0: flags=1000849 mtu 8232 index 1
>>> inet 127.0.0.1 netmask ff000000
>>><<>>
>>
>> So *nothing* is working other than the loopback pseudo-device.
>>Are the entries in /etc/path_to_inst still edited to different numbers?
>
> No. I put it back EXACTLY THE WAY IT WAS before I edited anything!
> What I have excerpted above is how they currently look.
O.K. BTW -- before you change *any* file like this -- make a
copy (e.g. /etc/path_to_inst.ORIG) to make it easier to restore. But I
think that it is built fresh by a reconfigure boot.
Hmm ... does your system have a man page for path_to_inst? In
particular, does it say this?
================================================== ====================
Note that it is generally not necessary for the system
administrator to change this file, as the system will main-
tain it.
The system administrator can change the assignment of
instance numbers by editing this file and doing a reconfi-
guration reboot. However, any changes made in this file will
be lost if add_drv(1M) or devfsadm(1M) is run before the
system is rebooted.
================================================== ====================
And does Solaris 8 even *have* devfsadm? I know that it is not
in Solaris 2.6, and I thought that it originated in Solaris 9.
This suggests that maybe you want to totally rename the
path_to_inst file (perhaps to "PATH_TO_INST") and then do a shutdown and
issue a "boot -r" from the OBP prompt, since the system probably can't
see the "/reconfigure" file if you try that way to reboot. This should
recreate the path-to-inst file, using only what it finds in the system
at that time. And you can still use the other drive to recover with the
renamed file if that becomes necessary.)
>> Add the original ethernet card back into the system (leaving the
>>QFE installed and finding another slot for the original card) and do
>>the:
>>
>> touch /reconfigure
>> reboot
>>
>>cycle and see what happens. In particular, look at the output from
>>
>> ifconfig -a
>>
>>to see whether the card is added back.
>
> Okay. The QFE card is in PCI slot 2. (Nothing was in 1, 3 or 4.)
>
> I just put the original ethernet card I had in slot 4. (This was to
> leave more room between the cards, for cleaning or whatever occasionally,
> if by some wonderful chance they remain this way...)
>
> I did a "boot -r".
>
> An "ifconfig -a" shows:
>
><<>>
> lo0: flags=1000849 mtu 8232 index 1
> inet 127.0.0.1 netmask ff000000
> hme2: flags=1000843 mtu 1500 index 3
> inet 192.168.2.2 netmask ffffff00 broadcast 192.168.2.255
> ether 8:0:20:8f:d3:67
><<>>
Interesting -- it does not see hme0 (in the system board) or
hme1 (which supposedly is in the QFE). Another reason to get the system
to create a new path_to_inst file by doing a "boot -r" after the file
has been renamed.
But -- it *does* like that card -- in any slot.
>> Hmm ... another thing to check -- *which* slot is the QFE in?
>>according to the FEH the topmost PCI slot is a 3.3V 33/66 MHz slot,
>>while the other three are 5.0V 33 MHz and the QFE cards are 3.3/5V 33
>>MHz so I *think* that it should work in any slot, but try it in a
>>different slot anyway. Especially if it is in the topmost slot.
>
> It was and still is in PCI slot 2. It is not in the topmost slot.
O.K. Just for the fun of it -- move the QFE card to the topmost
slot and repeat your reconfigure boot -- then your "ifconfig -a" to see
whether it sees the ports again.
> I'll wait on this and see whether you still want me to try this, in
> light of what I've included in this article.
>
>> Is your original card one of 501-4359, 501-5019, or 540-3981? If
>>so, it should work in the same slots -- but who knows. Hmm ... is the
>>card's connector fully seated along the length of the system board's
>>connector?
>
> I don't know. Can I tell by looking at the card?
Yes -- usually a white or orange colored barcode label starting
with the five digits of one of those, without the '-'. The number will
be longer, with the rest of it making up the card's serial number.
On one of my Creator-3D cards, the barcode label is along the
back edge on the component (down) side of the card.
Again -- the number probably does not matter, because it works.
But you probably want to look on eBay for a match for the one you have,
so your search should be for the number both with and without the '-'.
> (Hmmm... looking
> at the "top" of the circuit board, I see a number 3697 on a little white
> sticker. That's the only thing I see, on the side I can see, other than
> "Sun, Assembled in USA" on another little white sticker.)
Nope -- too short. And the barcode label will probably be on
the bottom (component) side of the board.
> Let me guess, there is something on the other side that indicates it?
> (Now that I have in the computer!) :-D (I should have "read ahead"
> before I put it in the computer, on the bottom PCI slot!) :-D
:-)
> And, yes, it is fully seated. I did verify this when I put it in the
> system, before I booted the first time afterward. And, it's nice. If it
> IS fully seated, that screw on the end is just for a little "hold", but it
> is already fully against the metal frame where the screw goes in -- if it
> is "in" fully in the motherboard slot.
>
>> Or -- could the PCI bus have been damaged by the lighting?
>>Consider getting a replacement system board as the safer (and surer) way
>>to go.
>
> I don't *think* so. At least, when I got home on Tuesday, July 22
> (it will be ONE MONTH tomorrow!),
That hurts.
> the Linux was still connected to the
> Sun's port via the card I just placed back in. And, mutt was working.
> And, I had 27 new emails, which I read, deleted, switched folders, etc.
>
> At least, the PCI slot that contained this card (which was in PCI
> slot 2, where the QFE is now) still appeared to be working.
O.K. But maybe the QFE will work better in the top slot (slot
1). It is worth a try.
> I'll put what you had at this point in a "Non-Sun-Stuff" email. :-)
Good. It does not belong here.
> However, I'm going to go out, before it gets darker, even though it
> is currently raining hard, and remove the hurricane shutters. Fay is
> FINALLY moving again! And, none of the computer models show her getting
> too close to where I live. And, now that she is finally heading westward
> over Florida, she surely is not likely to strengthen. So, it may be up to
> 60-mph winds, but that's okay, and just continued lots of rain.
>
> My mother HATES the hurricane shutters up, because they block her
> ability to see out! :-D
Well -- she wants to *use* her fixed eye.
> I am not confident that we're not going to need
> them, so I'm going out to take them down.
That does not sound right. Didn't you mean to say that you
*were* confident that you would not need them?
> Tomorrow morning, I take my mother to the Filutowski Cataract and
> Laser Institute. They'll check her eye (just over one week since the
> surgery) and confirm for us that Wednesday of next week, she'll have her
> right eye done. (This was not "in stone" until they saw her again, this
> week.)
Good Luck,
DoN.
--
Email: | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
 |