Re: Version 0.9 of PA-RISC Linux Released
Thanks for the info - see in-lined comments below:
On Sun, 03 Jun 2001, Matthew Wilcox wrote:
> On Sun, Jun 03, 2001 at 09:42:11PM -0500, Michael S. Zick wrote:
> > Hi Folks -
> > Sub-Title this one: Adventures in running V0.9 on a HP9000/720 box:
> Well, 720 is explicitly mentioned as not yet being supported. There seems
> to be a bug in the ASP driver, so SCSI and LAN don't work.
Well, gee - I didn't know it wasn't supposed to work on a 720.
Would you swap me an e3000 for my hp9000/720?
Or should I try to fix this problem myself?
Or should I just wait?
SCSI and LAN - How about the EISA BA?
Note: The 9000/720 did not come standard with an EISA BA - it was an
option - but my box has the option (no card in slot thought).
> In addition,
> the trap handler needs another case to handle 720 & family (traps 26,
> 27 and 28 are reported as trap 18 on this cpu, and which one needs to be
> distinguished. there's a hack around which assumes this is a page-fault,
> but this will get you into an infinite loop in kernel space if user does
> unaligned data reference instead). Oh, and I hear a rumor that this
> revision of the CPU does not allow user-space reads of %cr27, so we need
> to write a handler for this machine, or use non-floating stacks in glibc.
That sounds like a well defined problem - where can I get an assembly language
guide and internal arch. description for the PA-RISC1.1(c I believe - could be
anything from a-z)?
> Sounds like a bug in the STI code too then... helge, did you want a 720
> or 730 to play with? :-)
I tried the 730_32STI kernel - it didn't work. I have BUILT kernels that do work
from the earlier sources (I am still waiting for the new ones to finish arriving).
If there is a shortage of 9000/720 test boxes running around - just let me know
when you need anything "model 720 specific" tested (like a bootable kernel).
> That's expected behavior. Richard, could we change the printk?
Good idea. Even reading both the code and the comments, I couldn't tell if this
was reporting a failure or just an alternate mode of operation.
> Don't know. Can't say I've tried the parallel port myself on any machine.
I'll let you know - I may have to use the printk hook to write to the parallel port
during boot-up before I get this thing running.
> I'm still xcompiling my kernels :-) I know it's a stable build
> environment, and up until jsm & alan modra found the most recent bug,
> it had no known bugs...
I did read the warnings that it might not work on anything older than what was
available for testing. So I am not complaining, just asking for help and/or
> I can't say that I know those systems... these are old X terminals based on
> the i960 processor? If so, I'm not aware of any work being done on these.
That's right - i960 processor based X terminals. One model is disk-less, the
other one can support both internal floppy and hard drives.
I had to pay 100 US$ for this system - I would hate to have to throw away
two thirds of it.
Oh - I am not a HP-9000 expert (less than two weeks experience) - the prior
owner kept all of the manuals, invoices, customer training books, sales documents,
etc. I am just quoting from what I have read in the old papers.