Re: Issues booting kernels in 7.0_RC1 (201507190150Z build)

[Adrian Christiansen <adrian.christiansen%gmail.com@localhost>]

Hello guys!,

I've been trying different configurations and kernels. Current as of
yesterday crashes as well. It doesn't seem to be an issue with obio.
But somehow the obio fault went away, maybe that was just a red
herring?

It took me every evening of this week to figure it out. But I've
learned something, I think :)

As of 201507240640Z, when uninorth tries to set the configure for AGP
for radeonfb, it will crash, not at once, after a few seconds. It
almost seemed like it would upset the next device who was trying to
write a uninorth_conf_write, which would end up catching the trap and
make the kernel panic. Somehow fwohci or gem?

When I was testing with sprinkled printf("%s: %d\n", __func__,
__LINE__); around in the uninorth.c source, going one step back trying
to find a common denominator. It almost seemed like the next device to
write a uninorth_conf_write would catch the trap and make the kernel
panic. Since the OpenFirmware output is so slow, it would crash before
the faster framebuffer had loaded and fwohci and gem was starting to
be initialized.

So if I disable radeonfb0 it will boot just fine!, it seems to be an
issue with uninorth and the agp initilazation. I'm now at the state
were I can run the system with the latest snapshot. How can I test
this further?

I'm going to transcribe the dmesg output I got and errors I found,
tomorrow morning. Just wanted to send an update on the progress I
finally made just now :)



Cheers!,
Adrian.

2015-07-21 20:50 GMT+02:00 Michael <macallan%netbsd.org@localhost>:
> Hello,
>
> On Tue, 21 Jul 2015 20:32:40 +0200
> Adrian Christiansen <adrian.christiansen%gmail.com@localhost> wrote:
>
>> obio0 at pci1 dev 23 function 08 addr 0x80000000
>> obio0: enabling KeyLargo internal modem
>> trap: pid 0.1 (system): kernel\240MCHK trap @ 0x101c04 (SRR1=0x141030)
>> panic: trap
>> Stopped in pid 0.1 (system) at  netbsd:vpanic+0x13c:    or      r3, r29, r29
>> 0x00aa37e0: at panic+0x4c
>> 0x00aa3820: at trap+0x454
>> 0x00aa38a0: kernel MCHK trap by bsr4rb+0x4: srr1=0x141030
>>             r1=0xaa3970 cr=0x24028084 xer=0x20000000 ctr=0x101c00
>> 0x00aa3970: at obio_attach+0x3bc
>
> Can you sprinkle something like
> printf("%s: %d\n", __func__, __LINE__);
> over arch/macppc/dev/obio.c::obio_attach() in order to pin down where
> exactly it panics?
>
> have fun
> Michael