List Info

Thread: Al's Xserve G4




Al's Xserve G4
user name
2007-06-20 13:01:27
So a kernel with the following configuration changes booted
for Al on
his G4 Xserve (based on GENERIC):

Added config for netbsd-wd0 root on wd0a type ffs

Commented out pdcsata* (hmm..  Meant to comment out
pdcide*)
Commented out radeonfb* and options RADEONFB_MMAP_BARS.
Commented out fwohci*, ieee1394if*, fwip*, sbp*

I also slightly patched pdcide.c:

Index: pdcide.c
============================================================
=======
RCS file: /cvsroot/src/sys/dev/pci/pdcide.c,v
retrieving revision 1.24
diff -p -u -r1.24 pdcide.c
--- pdcide.c	16 Nov 2006 01:33:10 -0000	1.24
+++ pdcide.c	20 Jun 2007 17:59:36 -0000
 -557,8
+557,13  pdc20265_pci_intr(void *arg)
 			continue;
 		crv = wdcintr(wdc_cp);
 		if (crv == 0)
-			printf("%s:%d: bogus intrn",
-			    sc->sc_wdcdev.sc_atac.atac_dev.dv_xname, i);
+{ static int akbcnt=0;
+  if (akbcnt++ < 10) {
+			printf("%s:%d: bogus intr (dmastat = %x)n",
+			    sc->sc_wdcdev.sc_atac.atac_dev.dv_xname, i,
+dmastat);
+  }
+}
 		else
 			rv = 1;
 	}

It did print the bogus intr message 10 times, but continued
through
the boot.

Anyone have time and resources to look into these things?

-allen

-- 
Allen Briggs  |  http://www.ninthw
onder.com/~briggs/  |  briggsninthwonder.com

Re: Al's Xserve G4
country flaguser name
United States
2007-06-21 17:45:51
Hi,

I wanted to thank everyone who has helped me with this.
Although it
complains when it is booting, it seems stable. I was able to
build tcsh,
pico, and the source tree without any real problems (just
user error).

I do have some concerns about the error at boot up and I was
wondering if
it would be safe to put this server into production this
way. Also, I need
to put some other options into the kernel so that it doesn't
run out of
file descriptors or network buffers. I also need to raise
the amount of
memory that it can use per process. The server works, so I
will be able to
do this, but I need to know how to apply this patch.

Thanks,
Al







On Wed, 20 Jun 2007, Allen Briggs wrote:

> Date: Wed, 20 Jun 2007 14:01:27 -0400
> From: Allen Briggs <briggsnetbsd.org>
> To: port-macppcnetbsd.org
> Subject: Al's Xserve G4
>
> So a kernel with the following configuration changes
booted for Al on
> his G4 Xserve (based on GENERIC):
>
> Added config for netbsd-wd0 root on wd0a type ffs
>
> Commented out pdcsata* (hmm..  Meant to comment out
pdcide*)
> Commented out radeonfb* and options
RADEONFB_MMAP_BARS.
> Commented out fwohci*, ieee1394if*, fwip*, sbp*
>
> I also slightly patched pdcide.c:
>
> Index: pdcide.c
>
============================================================
=======
> RCS file: /cvsroot/src/sys/dev/pci/pdcide.c,v
> retrieving revision 1.24
> diff -p -u -r1.24 pdcide.c
> --- pdcide.c	16 Nov 2006 01:33:10 -0000	1.24
> +++ pdcide.c	20 Jun 2007 17:59:36 -0000
>  -557,8 +557,13  pdc20265_pci_intr(void *arg)
>  			continue;
>  		crv = wdcintr(wdc_cp);
>  		if (crv == 0)
> -			printf("%s:%d: bogus intrn",
> -			    sc->sc_wdcdev.sc_atac.atac_dev.dv_xname,
i);
> +{ static int akbcnt=0;
> +  if (akbcnt++ < 10) {
> +			printf("%s:%d: bogus intr (dmastat =
%x)n",
> +			    sc->sc_wdcdev.sc_atac.atac_dev.dv_xname, i,
> +dmastat);
> +  }
> +}
>  		else
>  			rv = 1;
>  	}
>
> It did print the bogus intr message 10 times, but
continued through
> the boot.
>
> Anyone have time and resources to look into these
things?
>
> -allen
>
> --
> Allen Briggs  |  http://www.ninthw
onder.com/~briggs/  |  briggsninthwonder.com
>




Re: Al's Xserve G4
user name
2007-06-21 18:56:47
On Thu, Jun 21, 2007 at 03:45:51PM -0700, DataZap wrote:
> I do have some concerns about the error at boot up and
I was wondering if
> it would be safe to put this server into production
this way.

If you mean the "bogus intr", you're probably OK
with that.  I suspect
that there's something funny with the interrupt handling, so
we're
seeing two interrupts when we should be seeing one--the
second gets
the bogus intr message.

I don't have the time to go digging for this, though.

If you look back at the 2.1 dmesg, you had one of those
messages there.

> but I need to know how to apply this patch.

In general, apply patches by cd'ing to the appropriate
directory
and running "patch < patchfile".

For this one, you can just make the code at line 559 of
src/sys/dev/pci/pdcide.c look like:
============================================================
============
                if (crv == 0) 
{
#if 0
                        printf("%s:%d: bogus
intrn",
                           
sc->sc_wdcdev.sc_atac.atac_dev.dv_xname, i);
#endif
}
                else                
                        rv = 1;
============================================================
============

The counter was just to see if we hit it a lot or not.

-allen

-- 
Allen Briggs  |  http://www.ninthw
onder.com/~briggs/  |  briggsninthwonder.com

[1-3]

about | contact  Other archives ( Real Estate discussion Medical topics )