PALinux on HP9000/735-125
I installed PALinux 0.93 on my HP9000/735 running hat 125Mhz. I use a SCSI Disk connected to the narrow SCSI canel and the CRX48z graphics system.
Im using kernel version 2.14.16-pa24 and updated the debian packages like mentioned in the 0.9.3. errata section today.
Some questions:
- I tried execessing the framebufferdevice. 'fbset' report '/dev/fb0: no such device' though the STI console works fine as a text-console. Whats the reasion
for that error?
Can I currently use the framebuffer device on the HP9000/735?
- has someone been able to make the 100mbit EISA-ethernet-card 3c597 work in the HP9000/735 under Linux?
- can I enable a german keymap for the HIL keyboard (trying to do this during the installation made the keyboard totaly unusable so I had to reboot )?
- is the HIL mouse still unsupported or is there a patch or so to make it work?
But inspite of all these problems:
Nice work. It's great to see my PARISC Box running a free operating system instead the obsolute HPUX 10.20.
Sven
Hi Sven,
Quoted 7 lines · Sven Cronenberg
I installed PALinux 0.93 on my HP9000/735 running hat 125Mhz. I use a SCSI Disk connected to the narrow SCSI canel and the CRX48z graphics system.
Im using kernel version 2.14.16-pa24 and updated the debian packages like mentioned in the 0.9.3. errata section today.
Quoted 6 lines
Some questions:
- I tried execessing the framebufferdevice. 'fbset' report '/dev/fb0: no such device'
though the STI console works fine as a text-console. Whats the reasion
for that error?
Can I currently use the framebuffer device on the HP9000/735?
Try a newer kernel. It might work starting with kernel 2.4.17-pa2 and above.
Quoted 2 lines
- has someone been able to make the 100mbit EISA-ethernet-card 3c597 work in the HP9000/735 under Linux?
Sorry, no idea.
Quoted 3 lines
- can I enable a german keymap for the HIL keyboard (trying to do this during the
installation made the keyboard totaly unusable so I had to reboot )?
You might want to write a script with setkeycodes. Search the archives, I think I once sent such a script.
Quoted 2 lines
- is the HIL mouse still unsupported or is there a patch or so to make it work?
Still unsupported.
Helge
[parisc-linux] Use the X-Server from HPUX und PALinux?
On many PARISC Systems there is still no X-Windows support or X-Windows has to access the graphics card via the framebuffer device.
I wonder if there could be a chance to get the X-Server from HPUX working under PALinux. Would the HPUX compatiblity be good enough to have a chance whith that?
Perhaps the suggestion sounds a bit unconventional but for example many people running NETBSD on the 680x0 based HP-Workstations have made good experiences with that. For example I am running a HP9000/425 und NETBSD 1.5 but because the HP300 port of XFree is slow and buggy, I use the Xserver from HPUX 9.
SVen
Re: [parisc-linux] Use the X-Server from HPUX und PALinux?
Quoted 4 lines · Sven Cronbenberg
I wonder if there could be a chance to get the X-Server from HPUX working under PALinux. Would the HPUX compatiblity be good enough to have a chance whith that?
Well, originally that was the plan John Marvin had but I don't think he's had time to work on the missing kernel support (ioctl calls for gfx card drivers). If someone wants to try it and report what's missing, we might find time to get it working peice meal. In general, HPUX 32-bit SOM binary support is in place though no one is regularly testing it.
grant
Re: [parisc-linux] Use the X-Server from HPUX und PALinux?
In reply to Sven CronbenbergSven Cronbenberg writes...
Quoted 3 lines
On many PARISC Systems there is still no X-Windows support or X-Windows has to access the graphics card via the framebuffer device.
You are aware of the STI framebuffer driver right? You can use that to run X on most HP graphics cards. The HPUX Xserver could be much faster but sti-fb is currently better than nothing.
Signature · 2 lines
Matt Taggart Linux Development Lab [email protected] HP Linux Systems Operation
Quoted 4 lines · Sven Cronenberg
- I tried execessing the framebufferdevice. 'fbset' report '
/dev/fb0: no such device'
though the STI console works fine as a text-console. Whats the reasion
for that error?
crx48z is untested, so we don't try to use it in framebuffer mode. There is a check in drivers/video/stifb.c which refuses untested cards (search for Hack). Just remove that check and see what happens.
Quoted 1 line
Can I currently use the framebuffer device on the HP9000/735?
if you build your own kernel with the check removed, it might work.
Quoted 2 lines
- has someone been able to make the 100mbit EISA-ethernet-card 3c597 work in the HP9000/735 under Linux?
IMHO no EISA card will work at the moment. During the weekend I tried to get an Adapatec 1742 card going in my B132, and the problem is, that the pdc doesn't initialize EISA cards. Looks like that is done by HPUX. So before any EISA cards will work, we need something like that for Linux, too.
Thomas.
Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessary a good idea. [ Alexander Viro on linux-kernel ]
Quoted 2 lines
Quoted 2 lines
- has someone been able to make the 100mbit EISA-ethernet-card 3c597
work> > in the HP9000/735 under Linux?
Quoted 7 lines
IMHO no EISA card will work at the moment. During the weekend I tried to get an Adapatec 1742 card going in my B132, and the problem is, that the pdc doesn't initialize EISA cards. Looks like that is done by HPUX. So before any EISA cards will work, we need something like that for Linux, too.
I was trying to get EISA cards working a while ago. The basic problem is, as Thomas indicated, that the EISA cards need to be configured by a configuration utility before you can use them. HPUX has a utility called eisa_config, but there is no equivalent in Linux. Seems all other EISA platforms (i386, alpha, mips) can be configured by booting a diskette, so nobody bothered with a linux version yet.
I think it should work to boot HPUX, get the correct EISA .cfg file and use the eisa_config utility. The configuration is actually stored in an eeprom, so it's not tied to the OS. I can't test this though as the only HPUX CDs I have are too old for the machine.
Another option may be to run the utility on Linux, tried that once but it complained it could not access /dev/eeprom. A bit scary too, if something goes wrong one could end up with a corrupt eeprom....
Christian
Quoted 2 lines · Christian Suder
I think it should work to boot HPUX, get the correct EISA .cfg file and use the
eisa_configutility. The configuration is actually stored in an
but that's not enough. I did that and setup the card with HPUX, but it's still unconfigured, when Linux finds it. The .cfg describes how to configure the different card parameters, eisa_config puts the selected params to a nvram, and some other part does the real card poking during kernel startup. Other platforms have that card poking stuff included into the firmware.
It isn't a big deal to do the card setup itself, because .cfg includes everything needed. It's more a problem of getting that information during the kernel startup. If we know the nvram setup and how stuff gets put there, we could read that and do the card setup from within the Linux kernel. But I haven't found anything about that, yet.
Thomas.
Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessary a good idea. [ Alexander Viro on linux-kernel ]
Just as a mostly offtopic note you may well be able to get some help with this (or loan out some help) to the mips port. Some SGI systems (SGI I2 comes to mind) have EISA slots, and I suspect suffer from the same problem, it's also possible that SGI has come up with a workaround for it, so you'd have to get in touch with someone in the know.
Nick
Quoted 28 lines · Thomas Bogendoerfer
Quoted 2 lines · Christian Suder
I think it should work to boot HPUX, get the correct EISA .cfg file and use the
eisa_configutility. The configuration is actually stored in anbut that's not enough. I did that and setup the card with HPUX, but it's still unconfigured, when Linux finds it. The .cfg describes how to configure the different card parameters,
eisa_configputs the selected params to a nvram, and some other part does the real card poking during kernel startup. Other platforms have that card poking stuff included into the firmware.It isn't a big deal to do the card setup itself, because .cfg includes everything needed. It's more a problem of getting that information during the kernel startup. If we know the nvram setup and how stuff gets put there, we could read that and do the card setup from within the Linux kernel. But I haven't found anything about that, yet.
Thomas.
Signature · 7 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessary a good idea. [ Alexander Viro on linux-kernel ] _______________________________________________ parisc-linux mailing list [email protected] http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
Quoted 2 lines · On 2002.01.06 23:29 Thomas Bogendoerfer wrote:
but that's not enough. I did that and setup the [EISA-] card with HPUX, but it's still unconfigured, when Linux finds it.
I think that the PDC only initialize a card that is used for booting. I also think that Matthew Wilcox had a HP 100MB Ethernet interface up and running a while ago. (The PDC might have initialzed that board when it searched for boot devices.)
Another problem with EISA boards currently, is that most of them are bus-masters and the current EISA support lacks the busmatering and DMA bits.
Quoted 5 lines
The .cfg describes how to
configure the different card parameters,eisa_configputs the selected params to a nvram, and some other part does the real card poking during kernel startup. Other platforms have that card poking stuff included into the firmware.
I have started an eeprom driver a while ago, and I started to try to decode the EISA parts of the EEPROM. I have about half of it figured out I think, but the other half seems to be somewhat hard.
I looked a doing a parser for the CFG files, but it turns out that there are a lot of options and stuff. Even binary code overlays that are supposed to be loaded and executed by the ECU (EISA configuration utility) exists. Adaptec uses one for the AHA-274x. Not suppoered by HPUX, though.
/Daneil
Quoted 4 lines · Daniel Engstrom
I think that the PDC only initialize a card that is used for booting. I also think that Matthew Wilcox had a HP 100MB Ethernet interface up and running a while ago. (The PDC might have initialzed that board when it searched for boot devices.)
hmm, I had a quick glance over the hp100.c and it looks like the driver does the eisa config itself. So that would explain why it worked.
Quoted 2 lines
Another problem with EISA boards currently, is that most of them are
bus-masters and the current EISA support lacks the busmatering and DMA bits.
that's why I was looking into the problem. I wanted to improve EISA support.
Quoted 3 lines
I have started an eeprom driver a while ago, and I started to try to decode the EISA parts of the EEPROM. I have about half of it figured out I think, but the other half seems to be somewhat hard.
that's a start. Do you have a patch for your eeprom driver somewhere ?
Quoted 4 lines
I looked a doing a parser for the CFG files, but it turns out that there are a lot of options and stuff. Even binary code overlays that are supposed to be loaded and executed by the ECU (EISA configuration utility) exists. Adaptec uses one for the AHA-274x. Not suppoered by HPUX, though.
I saw that overlay thing also for the AHA-1742. I just removed it from the .cfg to get eisa_config running with it.
Thomas.
Signature · 2 lines
Crap can work. Given enough thrust pigs will fly, but it's not necessary a good idea. [ Alexander Viro on linux-kernel ]