netdev
[Top] [All Lists]

Re: zerocopy changes in 3c59x.c

To: "David S. Miller" <davem@xxxxxxxxxx>
Subject: Re: zerocopy changes in 3c59x.c
From: Andrew Morton <andrewm@xxxxxxxxxx>
Date: Fri, 26 Jan 2001 22:49:27 +1100
Cc: Alexey Kuznetsov <kuznet@xxxxxxxxxxxxx>, "netdev@xxxxxxxxxxx" <netdev@xxxxxxxxxxx>
References: <3A714788.82C064BD@uow.edu.au>, <3A714788.82C064BD@uow.edu.au> <14961.24733.869800.77633@pizda.ninka.net>
Sender: owner-netdev@xxxxxxxxxxx
"David S. Miller" wrote:
> 
> Andrew Morton writes:
>  > And why do we do this:
>  >
>  >             vp->stats.rx_compressed++;
>  >
>  > each time we send a frame with hardware checksum?  This is the
>  > transmit path, not the receive path.  It seems we're abusing
>  > some ppp-related stats here.  What's up?
> 
> Every zerocopy driver does this, rx_compressed counts HW csummed
> transmit packets, and tx_compressed counts transmit packets
> containing more than one buffer.

Thanks.  I can see that :)

But why do this, rather than create new accounting fields?  Let
me guess: short-term thing, intended to be removed, didn't want
to hack the userspace tools?

<Prev in Thread] Current Thread [Next in Thread>