On Sat, May 12, 2001 at 04:04:05PM -0700, David S. Miller wrote: > > Andi Kleen writes: > > It violates the definition of write() though if the TCP socket is > > accessed via write(2), because writes are guaranteed to be atomic. > > The later could be fixed by aquiring the inode semaphore in > > sock_write, but I'm not sure if it is worth the cycles. > > Hmmm, do we even know of any implementation that doesn't essentially > implement write() via send()/sendmsg()? I don't know any, but of course it's possible that sendmsg() is atomic on other sockets implementations. I guess it's not worth changing until someone reports an actual breakage in some program caused by it. -Andi - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org
