Date: Wed, 2 Jan 2008 07:31:31 +0100 (MET) From: Peter B <pb@ludd.ltu.se> To: pyunyh@gmail.com Cc: freebsd-hackers@freebsd.org Subject: Re: 6.2R Patch to allow broken frames Message-ID: <200801020631.m026VV3s018890@brother.ludd.ltu.se> In-Reply-To: <20080102042210.GC27551@cdnetworks.co.kr> from "Pyun YongHyeon" at Jan 02, 2008 01:22:10 PM
next in thread | previous in thread | raw e-mail | index | archive | help
>Some ethernet controllers allow receving broken frames but I'm not >sure whether it really helps on production box execept debugging aid. >Receving broken frames would waste PCI bandwidth as well as systems >resources as it would be dropped anyway in upper layer. It's intended as an aid to debug malfunctions. And this patch makes it possible to enable/disable it at will. For fpga + phy, it helps to see exactly what's sent over the wire, and what the phy likely see at it's input. This is especially true for the crc32, which took some effort to get it right in verilog. By enableing broken frames I could see that the rest of the packet was ok and what crc the code generated. Things like last step xor etc.. Is much easier to spot. >I guess you can see increasing input errors(if_ierrors) in case of >receving broken frames. Did the counter is not enough? Just seing that there is an error won't help much. One needs to see exactly what's sent, ie nibble by nibble. Ofcourse the intended tool to get the results is tcpdump(1). Higher levels arn't likely to benefit much. Except for verifying error resilence. Involved files: /usr/src/sbin/ifconfig/ifconfig.c /usr/src/sbin/ifconfig/ifconfig.h /usr/src/sys/dev/tx/if_tx.c /usr/src/sys/net/if.h
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200801020631.m026VV3s018890>