On Thu, 23 Oct 2008 21:30:32 -0400 Jim Rees <rees@xxxxxxxxx> wrote: > Steve French wrote: > > If that is the case (ie that cifs and nfs never need to set these over > tcp - I am still having trouble reconciling that with the NFS guys' > comments that they must set rcvbuf (and Jim's comment below) > > If you have an application that wants to read a big chunk of data from a > socket, and won't remove any of that data from the socket until the entire > chunk has arrived, then the application must set the receive socket buffer > size big enough to hold the entire chunk. Otherwise the application will > stall. > > As far as I can tell, the corresponding situation does not hold for send > buffers, because the tcp layer will grow the send buffer to be big enough to > hold whatever the application wants to send. But I don't know if that's > always true, or if there is some limit, so to be safe our NFS patch > continues to set both the receive and send buffer sizes. Right, that's not actually the case for CIFS, so I don't think we need to have CIFS tune the receive buffer. cifs_demultiplex_thread just slurps data off of the socket as it comes without concern about whether we have a whole request or not. We may end up having to call kernel_recvmsg a few extra times, but I don't think we need to worry about it getting stuck. -- Jeff Layton <jlayton@xxxxxxxxxx> -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
