Scott,
I had a look at the upper layers...and it look like it goes into QOS and
fair queueing.
In any case, I see the queueing is stopped in e1000_xmit_frame( ) by
calling netif_stop_queue( ) . But I couldn't find to when does it set it
back...I mean, when does e1000 starts queueing again? I didn't find any
call to netif_start_queue( ) from e1000.
Thanks
"Feldman, Scott" wrote:
> I suppose the layer above the driver could keep track of drops. We
> return 1 or 0 in the xmit_frame routine, where 1 = packet not sent
> ("dropped"), 0 = packet was sent. You'd need to look at the code above
> the driver to see if they're keeping track of whether the driver returns
> 0 or 1.
> -scott
>
> > -----Original Message-----
> > From: Prashanth T [mailto:prasht@in.ibm.com]
> > Sent: Wednesday, July 09, 2003 11:30 AM
> > To: Feldman, Scott
> > Subject: Re: Info on 'tx_fifo_errors' for e1000
> >
> >
> > Thanks Scott for the reply.
> >
> > In e1000_main.c , I also see a comment "Tx Dropped needs to
> > be maintained
> > elsewhere". Can you please let me know what does this mean?
> > As far as I
> > understand from the code, the data is put into the tx ring
> > and the rest controller should take care. Please correct me
> > if I am wrong. Again in this case, I cannot find any
> > register corresponding to dropped tx packets.
> >
> > Prashanth
> >
> > "Feldman, Scott" wrote:
> >
> > > > Can anybody let me know which register of Gigabit ethernet
> > > > controller in drivers/net/e1000/e1000_hw.h [2.4.20 kernel]
> > > > corresponds to 'tx_fifo_errors' of net_device_stats
> > >
> > > No register corresponds to tx_fifo_errors because the controller
> > > doesn't start a DMA into the FIFO until there is room in the FIFO.
> > >
> > > -scott
> >
> >
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
