linux-apus-devel

test10 boot problems

[Michel <mic...@ii...>]

Hi all,


I'm trying to get MOL working in 2.4.0-test10 (MOL doesn't support test11=
 or
later yet) by applying Roman's latest changes and adding a find_path_devi=
ce
function which always returns NULL.

But the kernel doesn't boot:

Starting kswapd v1.8
Machine check in kernel mode.
Caused by (from SRR1=3D49070): Transfer error ack signal
NIP: C013A930 XER: 00000000 LR: C0214798 REGS: c2f69dc0 TRAP: 0200
MSR: 00049070 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
TASK =3D c2f68000[1] 'swapper' Last syscall: 120=20
last math 00000000 last altivec 00000000
GPR00: 801F01A0 C2F69E70 C2F68000 C026CBFC C3810000 000003AD 00000000 000=
00010=20
GPR08: 00000C38 C3801000 C000D0C4 00000000 C01E0C38 7723EA94 770CABF0 3C8=
A4900=20
GPR16: 10101010 11111111 12121212 13131313 003FF000 15151515 16161616 171=
71717=20
GPR24: 18181818 19191919 1A1A1A1A 00000000 00000001 C0240000 C01FA644 C02=
6CBFC=20
Call backtrace:=20
C0240000 C0214798 C021283C C020A9CC C0208D48 C0207908 C0202814=20
C02028B4 C0003B00 C0009678=20
Kernel panic: machine check
Rebooting in 180 seconds..<<<<<<<<<<<<<<<<<<<<


Any idea what's wrong here?


Michel


--=20
Earthling Michel D=E4nzer (MrCooper)  \  CS student and free software ent=
husiast
Debian GNU/Linux (powerpc,i386) user \   member of XFree86 and The DRI Pr=
oject

Re: test10 boot problems

[Roman Z. <zi...@fh...>]

Hi,

On Fri, 29 Dec 2000, Michel D=E4nzer wrote:

> Any idea what's wrong here?

Without ksymoops output? Nope, sorry. :)

bye, Roman

Re: test10 boot problems

[Michel <mic...@ii...>]

Roman Zippel wrote:

> On Fri, 29 Dec 2000, Michel D=E4nzer wrote:
>=20
> > Any idea what's wrong here?
>=20
> Without ksymoops output? Nope, sorry. :)

Yeah, alright, alright :)

So it happened: I used ksymoops for the first time. Unbelievable, isn't i=
t?

The panic was in pm2fb_reset, so I found out that the generic PCI support=
 in
pm2fb doesn't work in 2.4 :(

I had to go back to test9 because get_module_symbol isn't even there in t=
he
APUS test10. I made some progress, it now at least tries to load the mol
module, but then it panics:

Kernel panic: kernel access of bad area pc c4877758 lr c4877708 address
C0003930 tsk insmod/199

Which ksymoops translates to:

>>NIP; c4877758 <END_OF_CODE+45ef908/????>   <=3D=3D=3D=3D=3D

Could this be because mol.o is built for test10? Or how could I debug thi=
s?


Anyway, I've now also created a mol-branch in the APUS 2.4 tree. So other=
s can
see if they get the same panic. :)


Michel


--=20
Earthling Michel D=E4nzer (MrCooper)  \  CS student and free software ent=
husiast
Debian GNU/Linux (powerpc,i386) user \   member of XFree86 and The DRI Pr=
oject

Re: test10 boot problems

[Roman Z. <zi...@fh...>]

Hi,

On Sat, 30 Dec 2000, Michel D=E4nzer wrote:

> So it happened: I used ksymoops for the first time. Unbelievable, isn't i=
t?

Usually I don't use it, too. I use gdb for that.

> Which ksymoops translates to:
>=20
> >>NIP; c4877758 <END_OF_CODE+45ef908/????>   <=3D=3D=3D=3D=3D

When you insmod the mol module, try -m, so you get a list of addresses /
symbols and it should be possible to find where it crashed.

bye, Roman

Re: test10 boot problems

[Michel <mic...@ii...>]

Roman Zippel wrote:

> > So it happened: I used ksymoops for the first time. Unbelievable, isn=
't
> > it?
>=20
> Usually I don't use it, too. I use gdb for that.

I also tried that, but it couldn't get the symbols.


> > Which ksymoops translates to:
> >
> > >>NIP; c4877758 <END_OF_CODE+45ef908/????>   <=3D=3D=3D=3D=3D
>=20
> When you insmod the mol module, try -m, so you get a list of addresses =
/
> symbols and it should be possible to find where it crashed.

Okay, I'll try that, thanks.


Happy New Year everybody,

Michel


--=20
Earthling Michel D=E4nzer (MrCooper)  \  CS student and free software ent=
husiast
Debian GNU/Linux (powerpc,i386) user \   member of XFree86 and The DRI Pr=
oject