Yenta TI on MIPS

[RB]

I've been using ksymoops to decode these, but am enough of a newb on
this to not know whether my x86-only ksymoops is able to do so
properly, even with the modules/System.map/vmlinux/etc.

I'll start poking around toward linux-mips or the general list.  I
think it's a buggy implementation by Linksys of Yenta TI in general -
looking at the driver they provide in their GPL tarball for 2.4.20
(ftp://ftp.linksys.com/opensourcecode/wrt54g3g-st/2.00.9/WRT54G3G-ST_v2.00.9_US.tgz),
there are a lot of magic numbers and general ugliness in yenta.c.

The OpenWRT crew has done a reasonable job of cleaning up that code,
but there are still things like hard-coding the IRQ mask and memory
resources (since probing doesn't work) that are just ugly anyway.
Regardless, there's still a large chunk of 'cb_writeX' calls in the
Linksys-provided driver that (of course) has zero documentation and
doesn't seem to have been ported over.


RB

On 1/18/08, Peter Stuge <stuge-linux-pcmcia at cdy.org> wrote:
> On Fri, Jan 18, 2008 at 07:31:03PM -0700, RB wrote:
> > > What does the oops look like?
> > Here's one from the crash induced by reading I365_INTCTL;
>
> Symbol names would help. But I think you should find MIPS experts
> who can help you sort out the low level problem. Try linux-mips or
> simply lkml for a good mix. Once the bus is back in order we may be
> able to help. (Maybe)
>
>
> > I don't have the rest immediately available, but this thread has
> > most of what I've posted on this:
> > http://thread.gmane.org/gmane.comp.embedded.openwrt.devel/506
>
> I'll have a look.
>
>
> > Data bus error, epc == c00c4178, ra == c00c5c94
> > Cpu 0
> > $ 0   : 00000000 1000a800 c00aa000 80240000
> > $ 4   : 803ab000 00000001 00000001 00000000
> > $ 8   : 80239b00 80240000 80280000 80280000
> > $12   : 80280000 80e03bd2 80278d04 00000000
> > $16   : 80110000 00000066 803ab000 80110000
> > $20   : c00d0000 00001000 80023454 c00d0000
> > $24   : 00000000 80203c24
> > $28   : 80e02000 80e03ce0 c00d0000 c00c5c94
> > Hi    : 00000063
> > Lo    : 126fa800
> > epc   : c00c4178     Not tainted
> > Using defaults from ksymoops -t elf32-i386 -a i386
>
> Register dump is good, but..
>
>
> > Cause : 0000001c
> >         803ab000 80110000 c00c5c94 c00c5b1c 00000000 802c20ec 00001000
> 00000066
> >         00000066 80e03d48 00001066 c00d0000 00000000 c00c54b0 00000000
> c00d0000
> >         80108b28 80108aa0 802c2000 c00d0000 803ab000 c00c89ac 802c20b0
> c00c9018
> >         00000200 0000003e 00000028 802100c3 802c0340 1c000000 1c000fff
> 80e8ebc0
> > Call
> Trace:[<80110000>][<80023478>][<80110000>][<80110000>][<c00c5c94>][<c00c5b1c>][<c00c54b0>][<80108b28>][<80108aa0>][<c00c89ac>][<801287c8>][<8010ed2c>][<8012c2a4>][<801f7f5c>][<800bd85c>][<8012c714>][<8012af2c>][<800f9570>][<8012c5ac>][<8012b098>][<801287c8>][<8012b4b8>][<80066a34>][<8010ef44>][<80047898>][<8000bc60>][<8024b920>][<8024b920>]
> > Code: afb00018  8c82000c  00809021 <90500803> 3c11c00d  321000ff
> > 3c028002  3c05c00d  26249040
>
> This doesn't mean much.. Enable the option to include the symbol
> table in the kernel for oops printouts.
>
>
> //Peter
>
> _______________________________________________
> Linux PCMCIA reimplementation list
> http://lists.infradead.org/mailman/listinfo/linux-pcmcia
>