[Cbe-oss-dev] Fw: Spidernet Problems on Rev 2 CPBW w/2.6.18-arnd7

Linas Vepstas linas at austin.ibm.com
Tue Feb 27 07:52:54 EST 2007


Hi Owen,

Owen Stampflee <ostampflee at terrasoftsolutions.com>  wrote:
> 
> After boot, dmesg gets filled with the following messages:
> 
> Got error interrupt on eth1, GHIINT0STS = 0x00000060, GHIINT1STS =
> 0x00010000, GHIINT20
> Got error interrupt on eth0, GHIINT0STS = 0x00000060, GHIINT1STS =
> 0x00010000, GHIINT20
> 
> Of course, neither eth0/eth1 are non-functional.
> 
> And after rmmod'ing spidernet, the following messages is repeated 20-50
> times:
> 
> eth0: forcing end of tx descriptor with status xa0000000
> eth1: forcing end of tx descriptor with status xa0000000
> 
> I havent tried 2.6.18.0, but 2.6.18.6 works fine so I'm guessing that
> something in arnd's patch series (but not any of the patches with
> "spider" in their name) was introduced that has broken it. Arnd or
> anyone else have any idea what patch may have broken this?

When you write "neither eth0/eth1 are non-functional.", I assume 
this double negative means "both eth0/eth1 are functional".

I am not sure what the current set of patches are that you are 
working with.  There was a raft of spider problems fixed, and
I do remember that there were several versions that spewed messages
like this. 

The spidernet patches in the mainline kernel should not have this 
problem. If the messages aren't too annoying, then I suggest simply 
ignoring them; they should go away in some later version.

--linas



More information about the cbe-oss-dev mailing list