Re: Sendmail, lynx, Netscape, sshd, Linux kernel (twice)

[davem () DM COBALTMICRO COM (David S. Miller)]

   From: Andi Kleen <ak () muc de>
   Date: 10 Nov 1998 21:02:38 +0100

   I'm sorry that is not the full story. In some cases Linux 2.1 TCP
   accept may still return network errors. This patch against 2.1.127
   should fix it (and also fixes another bug). If an asynchronous error
   occurs now after the 3way handshake completion it will be returned on
   the first socket operation after accept.

I have two problems with this patch:

1) This changes the behavior of errors between 2.0.x and 2.1.x
   If this is the desired behavior, then we should agree to
   put the same change into 2.0.x as well else we will have
   developers coding their applications in these situations in
   two incompatible ways.  Or worse, they will make their code
   work on one, and thus not in the other.

2) The unhash's added are spurious, in the newly accepted case
   the receive and write buffers should be empty, therefore the
   destroy_sock() calls will unhash the socket for you.  I don't
   want these added since they are not necessary.  If there is a case
   where the immediate unhash will not happen in destroy_sock(),
   please tell me when that is.

Later,
David S. Miller
davem () dm cobaltmicro com