Re: [parisc-linux] Trouble booting a 710 (bushmaster)

Re: [parisc-linux] Trouble booting a 710 (bushmaster)

[Richard Hirst]

On Mon, Jan 22, 2001 at 10:02:46AM -0500, Jim Buttafuoco wrote:
> 	This also fails.  attached is the boot log.  

> SCSI subsystem driver Revision: 1.00
> sim700: Couldn't get consistent shared memory
> sim700: Configuring 53c700 (SCSI-ID 7) at f0825100, IRQ 86, options 1
> scsi0: Revision 0x0
> Post test1, istat 05, sstat0 00, dstat 84
> sim700: WARNING IRQ probe failed, (returned 0)
> scsi0: WARNING: target data areas are not dma coherent!
> scsi0: test 1 completed ok.

Well, scsi is happier now.  The messages above are expected, and
test 1 passing means the scsi chip can access memeory ok.

> scsi0 : LASI/Simple 53c7xx
> scsi0: Unable to abort command for target 0
> scsi0: Unable to send Bus Device Reset for target 0
> scsi0: Unable to do SCSI bus reset
> scsi0: >>>>>>>>>>>> Host reset <<<<<<<<<<<<
> scsi0: could not halt NCR chip, stage 0
> scsi0: Unable to abort command for target 1
> scsi0: Unable to send Bus Device Reset for target 1
> scsi0: Unable to do SCSI bus reset
> scsi0: >>>>>>>>>>>> Host reset <<<<<<<<<<<<

Those messages might be caused by an unterminated scsi bus, or by
a problem with interrupts.

The network problem still looks like interrupts from the 82596 are
disabled somehwere and not getting to the processor.

Still seems likely to me that the kernel is not enabling some interrupts
properly for the 710 bushmaster machines.

Anyone on the list got a similar machine working?  The h/w database
lists quite a few people as owning one of these boxes.

My 715 now works as far as scsi and network are concerned, but (a) I
have to disable nscd, (b) tar hangs, and (c) I have disabled parallel
port support.  nscd and tar both seem to go in to a loop calling
sched_yield() from libc.

Richard

Re: [parisc-linux] Trouble booting a 710 (bushmaster)

[Christoph Plattner]

I have the same SCSI message on my 720. The problem here
is definitly only one device, the TEAC floppy on the
SCSI bus. When I disconnect it, it runs perfect.

So this message also shows problems with special devices.
I don't know, if there are incaptibilities in the devices
(under HP sometimes this drive is seen as boot device, some
times not....).

Ot there is a problem in the routine probing the devices ....

With friendly regards
	Christoph P.


Richard Hirst wrote:
> 
> On Mon, Jan 22, 2001 at 10:02:46AM -0500, Jim Buttafuoco wrote:
> >       This also fails.  attached is the boot log.
> 
> > SCSI subsystem driver Revision: 1.00
> > sim700: Couldn't get consistent shared memory
> > sim700: Configuring 53c700 (SCSI-ID 7) at f0825100, IRQ 86, options 1
> > scsi0: Revision 0x0
> > Post test1, istat 05, sstat0 00, dstat 84
> > sim700: WARNING IRQ probe failed, (returned 0)
> > scsi0: WARNING: target data areas are not dma coherent!
> > scsi0: test 1 completed ok.
> 
> Well, scsi is happier now.  The messages above are expected, and
> test 1 passing means the scsi chip can access memeory ok.
> 
> > scsi0 : LASI/Simple 53c7xx
> > scsi0: Unable to abort command for target 0
> > scsi0: Unable to send Bus Device Reset for target 0
> > scsi0: Unable to do SCSI bus reset
> > scsi0: >>>>>>>>>>>> Host reset <<<<<<<<<<<<
> > scsi0: could not halt NCR chip, stage 0
> > scsi0: Unable to abort command for target 1
> > scsi0: Unable to send Bus Device Reset for target 1
> > scsi0: Unable to do SCSI bus reset
> > scsi0: >>>>>>>>>>>> Host reset <<<<<<<<<<<<
> 
> Those messages might be caused by an unterminated scsi bus, or by
> a problem with interrupts.
> 
> The network problem still looks like interrupts from the 82596 are
> disabled somehwere and not getting to the processor.
> 
> Still seems likely to me that the kernel is not enabling some interrupts
> properly for the 710 bushmaster machines.
> 
> Anyone on the list got a similar machine working?  The h/w database
> lists quite a few people as owning one of these boxes.
> 
> My 715 now works as far as scsi and network are concerned, but (a) I
> have to disable nscd, (b) tar hangs, and (c) I have disabled parallel
> port support.  nscd and tar both seem to go in to a loop calling
> sched_yield() from libc.
> 
> Richard
> 
> ---------------------------------------------------------------------------
> To unsubscribe: send e-mail to parisc-linux-request@thepuffingroup.com with
> `unsubscribe' as the subject.

-- 
-------------------------------------------------------------------------
private:	christoph.plattner@dot.at
company:	christoph.plattner@alcatel.at

[parisc-linux] strace ??? - is there any ???

[Christoph Plattner]

I often use strace to analyse the problems I have with
applications. Espacially I had success in analysing problems
around a bash which is directly called by the kernel
by editing the kernel (init/main.c) to directly start a 
statically linked strace to start the bash.

To analyse the problem on the 720 (handler_interrupt)
I want to repeat this sucessful step, but I miss the
strace. 

Is ther any instrumented for HP PA-RISC ??

With friendly regards

	Christoph



-------------------------------------------------------------------------
private:	christoph.plattner@dot.at
company:	christoph.plattner@alcatel.at

Re: [parisc-linux] strace ??? - is there any ???

[Richard Hirst]

On Mon, Jan 22, 2001 at 09:01:15PM +0100, Christoph Plattner wrote:
> I often use strace to analyse the problems I have with
> applications. Espacially I had success in analysing problems

strace source is in cvs on puffin.external.hp.com, but you need to
check it out and build it yourself at the moment.

Richard

Further (build ?): [parisc-linux] strace ??? - is there any ???

[Christoph Plattner]

Hallo Mr Hirst !

Now I have following problem. What is the clean
way to build such autoconfigured tool cross ?

Is it via `--target=hppa-linux', but here it will build for i386 !
Or is it `--host=hppa-linux --build=i386-linux' ??? 

Or have I to define a CC=/usr/parisc/bin/hppa-linux-gcc ?

Must I use CC= also on configure and make or only on make
or configure..... 

... etc....

How to do this in the correct way ?

Thanks

	Christoph


Richard Hirst wrote:
> 
> On Mon, Jan 22, 2001 at 09:01:15PM +0100, Christoph Plattner wrote:
> > I often use strace to analyse the problems I have with
> > applications. Espacially I had success in analysing problems
> 
> strace source is in cvs on puffin.external.hp.com, but you need to
> check it out and build it yourself at the moment.
> 
> Richard
> 
> ---------------------------------------------------------------------------
> To unsubscribe: send e-mail to parisc-linux-request@thepuffingroup.com with
> `unsubscribe' as the subject.

-- 
-------------------------------------------------------------------------
private:	christoph.plattner@dot.at
company:	christoph.plattner@alcatel.at

Re: Further (build ?): [parisc-linux] strace ??? - is there any ???

[Christoph Plattner]

Again:  The configure script cannot find `config.h.in'
Is this missed in the CVS version of this day ?
	
	Christoph


Christoph Plattner wrote:
> 
> Hallo Mr Hirst !
> 
> Now I have following problem. What is the clean
> way to build such autoconfigured tool cross ?
> 
> Is it via `--target=hppa-linux', but here it will build for i386 !
> Or is it `--host=hppa-linux --build=i386-linux' ???
> 
> Or have I to define a CC=/usr/parisc/bin/hppa-linux-gcc ?
> 
> Must I use CC= also on configure and make or only on make
> or configure.....
> 
> ... etc....
> 
> How to do this in the correct way ?
> 
> Thanks
> 
>         Christoph
> 
> Richard Hirst wrote:
> >
> > On Mon, Jan 22, 2001 at 09:01:15PM +0100, Christoph Plattner wrote:
> > > I often use strace to analyse the problems I have with
> > > applications. Espacially I had success in analysing problems
> >
> > strace source is in cvs on puffin.external.hp.com, but you need to
> > check it out and build it yourself at the moment.
> >
> > Richard
> >
> > ---------------------------------------------------------------------------
> > To unsubscribe: send e-mail to parisc-linux-request@thepuffingroup.com with
> > `unsubscribe' as the subject.
> 
> --
> -------------------------------------------------------------------------
> private:        christoph.plattner@dot.at
> company:        christoph.plattner@alcatel.at

-- 
-------------------------------------------------------------------------
private:	christoph.plattner@dot.at
company:	christoph.plattner@alcatel.at

Re: Further (build ?): [parisc-linux] strace ??? - is there any ???

[Richard Hirst]

On Mon, Jan 22, 2001 at 10:46:23PM +0100, Christoph Plattner wrote:
> Again:  The configure script cannot find `config.h.in'
> Is this missed in the CVS version of this day ?

autoheader will generate that for you.

I normally build strace natively on my B180, will try and recall how
to cross-compile.

Richard

Re: Further (build ?): [parisc-linux] strace ??? - is there any ???

[Richard Hirst]

On Mon, Jan 22, 2001 at 10:43:12PM +0100, Christoph Plattner wrote:
> Hallo Mr Hirst !
> 
> Now I have following problem. What is the clean
> way to build such autoconfigured tool cross ?

HOSTCC=gcc CC=hppa-linux-gcc ./configure --target=hppa-linux --host=hppa-linux
make


worked for me, except for a problem with redefinition of struct dirent64.
I hacked that out of /opt/palinux/hppa-linux/include/linux/dirent.h to
temporarily get round it.

I havn't checked the binary works.  You might have to remove a chunk
of code from strace/defs.h, depending on how new your glibc is.  With
a new glibc you need to remove the following block:

  /* Hack until I figure out why we don't call ptrace in glibc, rather
   * than the system call direct.
   */
  static inline long raw_ptrace(int req, pid_t pid, void *addr, long data)
  {
          long res, ret;
  
          if (req > 0 && req < 4)
                  data = (long)&ret;
          res = ptrace(req, pid, addr, data);
          if (res >= 0 && req > 0 && req < 4) {
                  errno = 0;
                  return ret;
          }
          return res;
  }
  
  #define ptrace raw_ptrace


Richard

[parisc-linux] Bug in palo/palo/error.c

[Christoph Plattner]

In the file `palo/palo/error.c' a bug is in the char array
`*errormessages[]'. After `#include "usage.h"' there must be
a comma (or at the end of `usage.h', but this is a bad solution,
as I cannot use this long text on another position).

I got the error "10" but the message of error "11" as message "0"
(usage) and message "1" are concanated together !!

By the way: The PALO documentation should include a hint, that
I have to produce a PC-(intel)-compatible disk partition table.
My SCSI disk, I prepared was connected to a SUN and I have forgotten,
that there are differences in the layout, although it is no
problem to build the code on a x86 machine and run plao on 
a sun. 

So, if palo (or HP itself) has the restriction, only to handle 
X86 (DOS) Partition table format, then please mention it....

With friendly regards

	Christoph


-------------------------------------------------------------------------
private:	christoph.plattner@dot.at
company:	christoph.plattner@alcatel.at