You may be right but I have certain expatiation about the way that setsockopt and getsockopt work on other system and that not realized under linux. I don't mind that linux works differently as long as I don't have to spend time working my way through the networking code to figure out what is doing. Docmunet the that it is diffrent! At 09:54 PM 4/2/01 +0200, Andi Kleen wrote: >On Mon, Apr 02, 2001 at 01:32:57PM -0600, William L. Jones wrote: > > SIGH, > > > > I expect getsockopt to return a value close setsockopt, except for some > > rounding > > down that some kernels do. > >That's an invalid assumption, because it assumes that there is no metadata >in kernel per packet, which is a wrong view of the world because there is. > >Splitting it in userview to two counters would unfortunately break all of >userlands that expects to work with buffer sizes, so we cannot do that. > > > > > It returning the 2*setsockpt size instead which is why I though > setsockeopt was > > broken. May getscokept should be change to return 1/2 its current value. > >No, that breaks applications that expect BSD behaviour. > > >-Andi >- >: send the line "unsubscribe linux-net" in >the body of a message to majordomo@vger.kernel.org - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org
