Indy R4600 on Linux 2.6 -- possible cache issue?

Indy R4600 on Linux 2.6 -- possible cache issue?

[Stuart Longland]

[-- Attachment #1.1: Type: text/plain, Size: 2880 bytes --]

Hi All,

I've been tinkering with my Indy, trying to coax the contankerous
machine into booting Linux 2.6, until last night, without success.

For the record, the Indy's specs are as follows:
CPU:	R4600 SC rev2.0 @ 133MHz
RAM:	256MB
HDD:	9.0GB IBM SCSI
Video:	8-bit Newport (XL)

It successfully boots Linux 2.4 without incident... even in 64-bit mode.
---------------------------------8<-------------------------------------
> Linux indy 2.4.29-mipscvs-20050130 #2 Tue Feb 8 19:44:48 EST 2005 mips64 R4600 V2.0  FPU V2.0 SGI Indy GNU/Linux
>   09:36:06 up  8:02,  0 users,  load average: 0.02, 0.01, 0.00
--------------------------------->8-------------------------------------

The tests below have been conducted using the stock ip22_defconfig with
debugging options enabled and CPU set to R4x00.

Previously, it would either not boot at all, or it would just get past
starting userland before hard locking (as in, no SysRq joy, and no
response from pressing the power button).

My current kernel, made from a CVS HEAD snapshot last Sunday (4th
September) with cache disabled is the first one I've made which boots
consistently, although it's _very_ slow.

I'm wondering if we could be hitting a caching issue with the R4600 CPU?
 I also suspect the wd33c93 SCSI driver has a role to play in the
instability, as I'm getting a lot of warnings from it, they all look
something like the following

---------------------------------8<-------------------------------------
> [timestamp] drivers/scsi/wd33c93.c:77? spin_is_locked on uninitialised spinlock 893ce340.
--------------------------------->8-------------------------------------

The exact line number scrolls too quickly to read. (I'm not using serial
console presently -- I might try that later though)

As I say, the system boots, and in fact, I had the machine successfully
running e2fsck on its root partition (8GB) last night with this kernel.
 After 2 hours or so, it managed to get 75% though the filesystem check
-- I don't know beyond that, as it was approaching 1:00AM and I needed
sleep. ;-)  It's running extremely slow though, so I'd like to discover
what's causing these crashes.

The crashes happen on most boots with cache enabled.  Occasionally, I'll
get lucky, and the machine will boot on its own, but it's like 1 in
every 5 boots or so.

I'll see if I can get a dump out of this box, but as I say, it gets
through a file system check and activating swap, that's further than
I've ever got this machine running with cache enabled.
-- 
 ____                   _             Stuart Longland (a.k.a Redhatter)
/  _ \   ___    ___  __| |__  __   __ Gentoo Linux/MIPS Cobalt and Docs
- (_) \ /   \  ;   \(__   __)/  \ /  \                        Developer
 \    //  O _| / /\ \  | |  | /\ | /\ |
 /   / \   /__| /  \ \ | |  | \/ | \/ |
(___/   \____/|_;  |_| \_/   \__/ \__/ http://dev.gentoo.org/~redhatter

[-- Attachment #1.2: debug.config.gz --]
[-- Type: application/gzip, Size: 5746 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]