[Prism54-devel] Re: TxDescriptors -> 1024 default. Please not for every NIC!

Pekka Pietikainen pp@ee.oulu.fi
Sat, 22 May 2004 22:08:20 +0300


On Thu, May 20, 2004 at 09:38:11AM -0700, Jean Tourrilhes wrote:
> > Considering 802.11b's peak is at 11Mbit and standard 802.11g is at 54Mbit
> > (some manufacturers are using two channels and getting 108Mbit now) I'd
> > think we should stick at 100, as the patch proposes. Jean?
> 	I personally would stick with 100. The IrDA stack runs
> perfectly fine with 15 buffers at 4 Mb/s. If 100 is not enough, I
> think the problem is not the number of buffers, but somewhere else.
> For example, we might want to think about explicit socket callbacks
> (like I did in IrDA).
[removed netdev from Cc:]

Ok, after some more debugging it seems like WEP (104-bit key) is the problem 
and not txqueuelen == 1000 since WEP+txqueue == 100 breaks too

eth1: timeout waiting for mgmt response 1000, trigging device
eth1: timeout waiting for mgmt response
eth1: timeout waiting for mgmt response 1000, trigging device
eth1: timeout waiting for mgmt response
eth1: timeout waiting for mgmt response 1000, trigging device
eth1: timeout waiting for mgmt response

and without it it seems fine for days, with it and it sometimes hangs
within hours...

Any hints on where to start debugging? printk displaying what exactly
it's waiting for etc?

That code path seems to have: 
/* TODO: we should reset the device here */

which probably would clear this up whenever it happens :-)

Card is a 

02:00.0 Network controller: 3Com Corporation 3com 3CRWE154G72 [Office Connect Wireless LAN Adapter] (rev 01)
using the latest firmware.