Subject: Re: CVS commit: src/sys/netinet
To: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
From: Matt Thomas <matt@3am-software.com>
List: tech-net
Date: 03/11/2005 08:49:23
On Mar 11, 2005, at 8:35 AM, YAMAMOTO Takashi wrote:

> hi,
>
>> Module Name:	src
>> Committed By:	matt
>> Date:		Wed Mar  9 03:39:28 UTC 2005
>>
>> Modified Files:
>> 	src/sys/netinet: ip_output.c
>>
>> Log Message:
>> Move all the hardware-assisted checksum/segment offload code together.
>>
>>
>> To generate a diff of this commit:
>> cvs rdiff -r1.146 -r1.147 src/sys/netinet/ip_output.c
>
>> 			/*
>> 			 * The len field is meaningless at this point.
>> 			 * Clear it now, since some HW requires this.
>> 			 */
>> 			ip->ip_len = 0;
>
> what's "some HW"?  wm(4)?
>
> i think it's better to move this into the driver of the "some HW"
> because at least rtl8169 seems to need this "meaningless" value and
> it's easier to clear the field than calculating it.

Does the rtl8169 support tcp segment offload?  If not, then the 
assignment doesn't matter.  If so, *what* would be a sensible value in 
this situation?

-- 
Matt Thomas                     email: matt@3am-software.com
3am Software Foundry              www: http://3am-software.com/bio/matt/
Cupertino, CA              disclaimer: I avow all knowledge of this 
message.