From: barmar@alum.mit.edu   
      
   In article ,   
    Rick Jones  wrote:   
      
   > David Schwartz  wrote:   
   > > On Mar 1, 9:57?am, Rick Jones  wrote:   
   >   
   > > > Isn't the issue with the bad UDP checksum a result of Linux's   
   > > > decision to defer the checksum to the copy to user space? ?That is   
   > > > to say, they tried to do something clever that didn't quite fit   
   > > > with the implied (via say age-old BSD) semantics of a socket?   
   >   
   > > Yes to the first question, sort of to the second.   
   >   
   > > First, nothing was wrong with Linux's decision. Validating a UDP   
   > > checksum is optional, and a UDP datagram can be discarded at any   
   > > time.   
   >   
   > I do recall that *generating* a UDP checksum is optional, I don't   
   > recall the specification saying that validating it was optional.   
      
   That didn't stop BSD from treating it as so.  For many versions there   
   was a single kernel option that controlled both generation and   
   validation of UDP checksums.  I don't think this was cleaned up until   
   some time in the 90's.   
      
   --   
   Barry Margolin, barmar@alum.mit.edu   
   Arlington, MA   
   *** PLEASE don't copy me on replies, I'll read them in the group ***   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)