C180: no real progress :-(
Hi friends !
In the past days I learned a lot about HP workstations and the Linux port to hppa/parisc.
After I could exchange my first used C180 (256 MB, 2*4,5 GB FWSCSI HDD, GRAPHICS(3) accelerated graphics) by a newer model of the same series (C180) I tried to get palinux-0.9.2 up-n-running on that box.
Few weeks ago I posted some thoughts I had in this mailing list because palinux-0.9.2 was crashing on the older box all the time during installation.
Some kernel messages led my in the false direction. I thought DINO (h/w v2.1; or better the software glue of DINO) was the root cause of "my" problems. This box had firmware v2.7.
Installation via network/NFS did not work because of problems; With the older box i tried many things (100Mbps, 10Mbps, Hub, cross-over cable) but no success ...
After I got the replacement on Friday (DINO v3.1 [which should be ok. according the kernel messages ...], firmware v5.1) the problems are still present.
BTW: Grant, you are right ! But newer firmware seems not to solve the problem ...
I tried serial console device as well as STI console [PS/2 keyboard/mouse, GRAPHICS(0) with PC-style video monitor via HP adaptor connector].
The C180 boots palinux-0.9.2 from CD (additional drive at SESCSI). The system crashes either during kernel/driver installation (1st stage) or during "software" installation (2nd stage).
Only one time I could get the system installed from serial console. As I tried to switch to STI console the keyboard port problem hit me; The only chance was to reboot the box. As a result the root filesystem was corrupted so hard that fsck had problems tpo repair the corrupted filesystem ... :-(
At this point I have some (foe me) important questions:
Has anybody a running C180 under palinux-0.9.2 ?
Has anybody similar problems with other HP w/s and could solve those problems ? How solved ?
Some hints to the kernel guys:
[1] Try to fix the keyboard port problem ! I don't know how many reboots I had top handle this weekend but un-/replugging the keyboard is very boring ... :-)
[2] Try to get full support of any key (at least german keyboard support lacks on essential things ...) eg. no pipe/vertical bar; no support of capslock (keyboard unusable after accidently hit capslock; nothing helped; un-/replug, too)
[3] Sometimes I could see a message like
"ioctl: LOOP_CLR_FD: No such device or address"
short time before the C180 crashed. Any idea ?
Some hints to the system guys:
[1] Do real disabling of "dhcp-client" if it is set to "no" during installation. (Those kernel message caused by dhcp-client are boring; try it with open editor; great ;-)
[2] Try to include a date check/set into the installation routine; user should be informed about hardware clock setting)
I'm still trying to get that beast (= my C180) up-n-running !
Any help would be great !
Reinhard.
Quoted 4 lines · Reinhard Gimbel
The C180 boots palinux-0.9.2 from CD (additional drive at SESCSI). The system crashes either during kernel/driver installation (1st stage) or during "software" installation (2nd stage).
I have heard of a C160 with what could be similar problems. The C160 has internal diff scsi drives, and it seems that using that interface causes disk corruption. The only way he managed an install was to disconnect all devices from the diff scsi interface, and install to disks on the external scsi. Is that something you can try easily? Note, he had to physically disconnect the diff disks, it wasn't sufficient to simply not use them.
What scsi interfaces does that box have anyway? Boot up msgs should mention chip numbers - 53c710, 53c720, etc.
Quoted 3 lines
[1] Try to fix the keyboard port problem ! I don't know how many reboots I had top handle this weekend but un-/replugging the keyboard is very boring ... :-)
That problem is being actively worked on.
Quoted 4 lines
[2] Try to get full support of any key (at least german keyboard support lacks on essential things ...) eg. no pipe/vertical bar; no support of capslock (keyboard unusable after accidently hit capslock; nothing helped; un-/replug, too)
Hmm, yeah; I had a UK keyboard where pipe key didn't work. I got it with <alt>+124 on the numeric pad, or with <alt-gr> and the key to the left of '1' (back-quote, etc). Obviously needs fixing though.
Quoted 3 lines
[3] Sometimes I could see a message like
"ioctl:LOOP_CLR_FD: No such device or address"
short time before the C180 crashed. Any idea ?
That's a feature of the installer ;) Happens on all archs and is harmless.
Quoted 5 lines
Some hints to the system guys:
[1] Do real disabling of "dhcp-client" if it is set to "no" during installation. (Those kernel message caused by dhcp-client are boring; try it with open editor; great ;-)
That is a problem with the dhcp-client package, again common to all archs. The solution is "rm /etc/rc?.d/*dhcp-client". It has been discussed on the debian-boot list, and hopefully someone will fix it. It has been suggested that we switch back to using pump. There is a new version of dhcp-client in experimental, but it is much bigger, which causes problems for people that boot from floppies, at least.
Quoted 2 lines
[2] Try to include a date check/set into the installation routine; user should be informed about hardware clock setting)
I agree, telling what the time is in GMT and in your timezone without giving you a chance to do anything about it is a little pointless. You can use the firmware to set it correctly before you start, but if you are like me you always forget until it is too late.
Richard
[parisc-linux] J series question
Can J282 CPU be put into J200, or J210? Is there any corss-reference available for CPU boards and what machines they can fit into?
Tony
Re: [parisc-linux] J series question
Hi Tony,
According to
http://hwdb.parisc-linux.org/view.php3?type=machine&name=J282 , the model is a 780 Firhawk with a PCXU cpu. As far as I know, it is onee of the model in 64 bits. And the J200/210 are 770 models with PCXT'. In my opinion, you should not do that but I may be mistaken.
Quoted 2 lines
Can J282 CPU be put into J200, or J210? Is there any corss-reference available for CPU boards and what machines they can fit into?
Thomas.
ESIEE Team
Quoted 8 lines · On Tue, Aug 07, 2001 at 00:13 Richard Hirst wrote:
I have heard of a C160 with what could be similar problems. The C160 has internal diff scsi drives, and it seems that using that interface causes disk corruption. The only way he managed an install was to disconnect all devices from the diff scsi interface, and install to disks on the external scsi. Is that something you can try easily? Note, he had to physically disconnect the diff disks, it wasn't sufficient to simply not use them.
Any idea how one could connect the drives without those expensive Fast/Wide SCSI cables (The box I got is from a local 2nd hand computer dealer for approx USD 50 ...)
What do you mean with "external" SCSI ? If I connect the drives to the connector on the backplane isn't it the same FWSCSI channel as the internal one (which might be buggy) ? Or is it a seperate channel ?
Should I use one of the PCI slots to install a seperate SCSI host adaptor ? Could be an idea but actually I don't have an extra FWSCSI PCI adaptor on hand (I would like to use the internal 4.5 GB FWSCSI drives ...)
I also don't have a seperate external FWSCSI box where I could put the C180 internal drives in. (Again, I'm doing this as a hobby and my wife will propably kill me if I buy boxes and cables for lets say USD 150 ... ;-)
Isn't there another way to go ?
On that box was HP-UX 10.20 installed and the hardware is still the same. So there must/could be a software solution, right ?
BTW: Is that a problem of C1xx only ? Or is that problem known with other HP PA-Risc workstations, too ?
Quoted 2 lines
What scsi interfaces does that box have anyway? Boot up msgs should mention chip numbers - 53c710, 53c720, etc.
According to the boot message its a 53c710 for the SESCSI channel and a 53c720 for the FWSCSI channel
Quoted 10 lines
Quoted 5 lines
[2] Try to get full support of any key (at least german keyboard support lacks on essential things ...) eg. no pipe/vertical bar; no support of capslock (keyboard unusable after accidently hit capslock; nothing helped; un-/replug, too)
Hmm, yeah; I had a UK keyboard where pipe key didn't work. I got it with <alt>+124 on the numeric pad, or with <alt-gr> and the key to the left of '1' (back-quote, etc). Obviously needs fixing though.
Thanks for that hint with <alt>-xzy ! Seems to be the same as with good old DOS boxes, right ? Why I did not try that ?
Quoted 6 lines
Quoted 3 lines
[3] Sometimes I could see a message like
"ioctl:LOOP_CLR_FD: No such device or address" short time before the C180 crashed. Any idea ?That's a feature of the installer ;) Happens on all archs and is harmless.
The problem is that one isn't sure if that could be the core reason or not, you know ...
I'm not used to use Debian Linux. It's the first time to use Debian so I couldn't have much experience in that ...
Quoted 11 lines
Quoted 3 lines
[1] Do real disabling of "dhcp-client" if it is set to "no" during installation. (Those kernel message caused by dhcp-client are boring; try it with open editor; great ;-)
That is a problem with the dhcp-client package, again common to all archs. The solution is "rm /etc/rc?.d/*dhcp-client". It has been discussed on the debian-boot list, and hopefully someone will fix it. It has been suggested that we switch back to using pump. There is a new version of dhcp-client in experimental, but it is much bigger, which causes problems for people that boot from floppies, at least.
I know the SuSE Linux approach of init. I have to get familiar with the Debian/Redhat way to do it ...
Now some philosophy:
On the other hand one has to fight on many different frontiers if one is using uncommon things :-)
Of course it's easier to go the known way. But why not to go the unknown way ? There are many things to discover ...
Thanks a lot ! There's hope :-) There's light on the end of the dark tunnel :-)
Reinhard.
Quoted 9 lines
Quoted 9 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
"Reinhard" == Reinhard Gimbel <[email protected]> writes:
I have heard of a C160 with what could be similar problems. The C160 has internal diff scsi drives, and it seems that using that interface causes disk corruption. The only way he managed an install was to disconnect all devices from the diff scsi interface, and install to disks on the external scsi. Is that something you can try easily? Note, he had to physically disconnect the diff disks, it wasn't sufficient to simply not use them.
The `external' SCSI Richard is referring to in my C160 is a Symbios 53C875 PCI board.
If you shop for one you have to be careful. Most PC branded controllers only support 5V PCI. You need a controller with 3.3V PCI support for the C180. So I ended up buying a couple of Symbios branded ones. They support both voltages.
If you want to use the differential drives you got with the box, you should buy the differential version of the controller.
And you need an external enclosure for the disks as it is close to impossible to do internal cable routing in the box. The 53C720 pins go directly to the I/O bay edge connector.
Another alternative is to shop around for a dead C200/C240/C360 I/O daughterboard which has an 875 instead of a 720. That way you can use the internal disk bay. But you need Ultra Wide SCSI disks and not differential ones in that case.
And finally - if you don't feel like spending money - keep bugging Richard until he fixes the 710/720 problems ;)
Signature · 2 lines
Martin K. Petersen Cereal Bowl Engineer, Linuxcare, Inc. http://mkp.net/ SGI XFS, Linux/PA-RISC, GNOME
Quoted 3 lines · Reinhard Gimbel
Few weeks ago I posted some thoughts I had in this mailing list because palinux-0.9.2 was crashing on the older box all the time during installation.
I haven't seen any commits from rbrad yet.
I know he's still working on fixing the ccio driver bugs.
Quoted 4 lines
Some kernel messages led my in the false direction. I thought DINO (h/w v2.1; or better the software glue of DINO) was the root cause of "my" problems. This box had firmware v2.7.
please please please update the firmware on this box. See the parisc-linux FAQ for details on how to upgrade firmware. You don't want to chase data corruption problems due to the old dino. Upgrading the firmware will prevent the scenario. The box is useful but one can't use PCI devices (except graphics) in it.
Quoted 3 lines
Installation via network/NFS did not work because of problems; With the older box i tried many things (100Mbps, 10Mbps, Hub, cross-over cable) but no success ...
This should work. Make sure link lights on the hub/switch and C180 are on. Shouldn't need a cross-over cable if you have a hub. Othwerwise, review the palo parameters you are using.
Quoted 7 lines
After I got the replacement on Friday (DINO v3.1 [which should be ok. according the kernel messages ...], firmware v5.1) the problems are still present.
BTW: Grant, you are right ! But newer firmware seems not to solve the problem ...
No. But it avoids some other ones.
grant
Grant Grundler
parisc-linux {PCI|IOMMU|SMP} hacker
+1.408.447.7253