Re: PR/39203 CVS commit: src/sys/net

[christos%zoulas.com@localhost (Christos Zoulas)]

On Jul 25,  5:35pm, martin%duskware.de@localhost (Martin Husemann) wrote:
-- Subject: Re: PR/39203 CVS commit: src/sys/net

| On Fri, Jul 25, 2008 at 03:15:03PM +0000, Christos Zoulas wrote:
| >  Modified Files:
| >     src/sys/net: if_ether.h
| >  
| >  Log Message:
| >  PR/39203: Paul Ripke: PPPoE issues with broken MTU/MRU implementations
| >  Allow larger frames for systems that don't negotiate MTU/MRU properly.
| 
| I don't object this change, but it does look kinda wrong.
| 
| I wondered if the whole ETHER_MAX_FRAME() check in ether_input is not a bogus
| workaround for some old (maybe even long fixed) bugs in chipset drivers.
| Shouldn't we leave the decision "I did receive this frame OK" to the driver
| (modulo fcs if not checked in hardware)?

Yes, but it is also a cheap belt and suspenders test for ethernet health...
 
| If we did receive the frame ok, why should we drop it?

Are there parts in the system that make assumptions about maximum packet size?

christos