(interleaved chips) cfi_cmdset_0002 -> Kernel Panic

Hello!

In recent posts it is mentioned that there is a kernel panic like this:

/ # cat tmp/rootfs.jffs2.img > /dev/mtd0
Waiting for write to complete timed out in do_write_oneword.Scheduling in interrupt kernel BUG at sched.c:564!

Oops: Exception in kernel mode, sig: 4
NIP: C0012AD8 XER: 00000000 LR: C0012AD8 SP: C1F8BF10 REGS: c1f8be60 TRAP: 0700    Not tainted

MSR: 00089032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11

TASK = c1f8a000[23] 'cat' Last syscall: 4
last math 00000000 last altivec 00000000
GPR00: C0012AD8 C1F8BF10 C1F8A000 0000001B 00001032 00000001 00000000 C01D0000
GPR08: 00000DF8 00000000 00000000 C1F8BE30 44002008 00000000 7FFFFEB0 00000000
GPR16: 30005140 30005000 00000002 000000E4 00009032 01F8BF40 00000000 C0005E28
GPR24: C0005B80 30005154 00000004 00000001 00000001 7FFFEDF0 00000000 C1F8BF10 Call backtrace:

C0012AD8 C0005E44 00000488 10046E98 100451DC 1000B03C 10003C5C 10003808 30048EF4 10003590

Kernel panic: Aiee, killing interrupt handler!

In interrupt handler - not syncing
<0>Rebooting in 180 seconds..

It is happening because of the broken AMD command set for chips like mine (at least as far as I understand it from pretty recent posts to the list). I have the following startup with kernel.org 2.4.25:

physmap flash device: 300000 at ff000000
CFI: Found no Physically mapped flash device at location zero Search for id:(100 100) interleave(2) type(2)
Search for id:(100 100) interleave(2) type(2)
Search for id:(100 100) interleave(2) type(2)
Search for id:(00 02) interleave(2) type(1)
Search for id:(00 02) interleave(2) type(1)
Search for id:(00 02) interleave(2) type(1)
Search for id:(01 22da) interleave(4) type(2)
Found: AMD AM29LV800BT
Physically mapped flash: Found 4 x16 devices at 0x0 in 32-bit mod

It wouldn't work with a recent patch from the CVS snaphost downloads, because auto detection won't work. AFAIK it is broken.

How are the chances of fixing it? Is there a working version "out there" that supports my flash chip, auto detection and works?? Is there a quickfix I could apply (increasing some timeout value, copying over another implementation of that do_write_oneword function)???

Thanks in advance,
Oliver Korpilla

Hello

I patched a revision of MTD with version 1.93 (as suggested elsewhere) of this file (cfi_cmdset_0002.c) against the kernel. Configured fine, built fine (kernel.org 2.4.25), but did not autodetect the interleaved chip (as the newest snapshot fails to do as well).

So I guess if reverting would be an option, I still do not know where to revert to.

Any ideas?

Thanks in advance,
Oliver Korpilla

Hello!

Did you tried the last diff for cfi_cmdset_0002.c from the mailinglist?

greets dkey

No, but I guess I'm trying again with my reverted version (1.93 of the cfi_cmdset_0002.c).

I will work on a mapping for my board (was easier when it was configurable :) ) and trying out that first, taking the mbx860.c as an example.

Guess that was a stupid newbie mistake. :)

Thanks.

Oliver Korpilla