Hi Linus,
It seems that the TCP code has morphed again, breaking the RPC over
tcp write_space() calls. Basically it seems that the current version
of the tcp code no longer allows us to specify that we'd like to wait
until a specific number of bytes are free in the buffer (udp is
unchanged).
The following patch allows us to work with the current TCP
write_space() API by instead reenabling the sending of partial
requests.
Cheers,
Trond
--- linux/net/sunrpc/xprt.c.orig Tue Jul 18 21:39:29 2000
+++ linux/net/sunrpc/xprt.c Wed Sep 6 17:11:03 2000
@@ -1037,7 +1037,7 @@
return;
/* Wait until we have enough socket memory */
- if (sock_wspace(sk) < min(sk->sndbuf,XPRT_MIN_WRITE_SPACE))
+ if (!sock_writeable(sk))
return;
spin_lock_bh(&xprt_sock_lock);
@@ -1212,9 +1212,6 @@
*/
while (1) {
xprt->write_space = 0;
- status = -ENOMEM;
- if (sock_wspace(xprt->inet) < req->rq_slen + SOCK_MIN_WRITE_SPACE)
- break;
status = xprt_sendmsg(xprt, req);
if (status < 0)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 21:00:18 EST