Re: pegasosII crash at boot with -current

[Frank Wille <frank%phoenix.owl.de@localhost>]

Matthew Green wrote:

> i finally got around to trying my pegasosII with -current and it fails
> to boot.

How current is your current? Mine is 6 days old, and a generic ofppc kernel
boots on my Pegasos2/G3 with 512MB.


>  i get this:
>
>>> NetBSD/ofppc OpenFirmware Boot, Revision 1.8

It probably isn't important, but the recent revision is 1.11.
Did you boot the kernel from disk or over the net (100MBit or GBit-port?)?


> 0x0059adb0: kernel PERF trap by pool_cache_bootstrap+0xe0: srr1=0x1032
>             r1=0x59ae70 cr=0x24002042 xer=0x20000000 ctr=0

A performance monitor exception!? How could that happen?
Unforunately ddb doesn't show the performance monitor registers. The
contents of MMCR0 would have been interesting.

The only location where the perfmon is used in the kernel is while
calculating the clock frequency. And this is during configure().

The G3/G4 manual says that MMCR0 "must be cleared at power up", which could
mean its contents is undefined, and with some bad luck we get an exception
as soon as we enable interrupts? But I cannot believe that...


> anyone have any ideas what is up?

Not really. The only difference between our systems is that you have a G4
and I have a G3.

It worked for you in the past?
Did you already test a kernel after the switch from pegasospci to gtpci?

Since the last months I get kernel panics when an USB device is plugged in.
I don't know if there was a modification in some USB code or whether it is
caused by a problem with gtpci:

0xa9f26bd0: at panic+0x25c
0xa9f26c20: at trap+0x128
0xa9f26cb0: kernel DSI read trap @ 0xba938101 by usb_allocmem+0xb0:
srr1=0x9032
            r1=0xa9f26d70 cr=0x84000088 xer=0 ctr=0x3fe620 dsisr=0x40000000
0xa9f26d70: at usbd_do_request_flags_pipe+0xe8
0xa9f26da0: at uhci_open+0x5f4
0xa9f26de0: at usbd_setup_pipe+0xbc
0xa9f26e10: at usbd_new_device+0x480
0xa9f26eb0: at uhub_explore+0x1f4
0xa9f26f00: at usb_discover+0x60
0xa9f26f20: at usb_event_thread+0x38
0xa9f26f40: at emptyidlespin+0x10

Kiyohara-san wanted to have a look at it... ;)

-- 
Frank Wille