On Tue, 9 Jan 2001, Paul Cassella wrote:
> and mss_now seems to be less than skb->len when the printk happens. My
> copy of K&R is at work; could that comparison be being done unsigned
> because of skb->len? I wouldn't think so, but the alternative seems
> somewhat worse...
That'll teach me to post about integral promotions ...
> + printk(KERN_ERR "%s:%d:%s: err is unexpectedly %d.\n", file, line, func, ret);
... and hand-edit patches before breakfast.
I'm not familiar enough with the tcp code to know if this patch (against
-ac6) is a solution, band-aid, or, in fact, wrong, but I've run with it
(on -ac3) and haven't seen the errors for over twelve hours, which is
three times longer than it had been able to go without it coming up.
--- tcp.c.orig Thu Jan 11 08:54:50 2001
+++ tcp.c Thu Jan 11 08:56:42 2001
@@ -954,7 +954,7 @@
*/
skb = sk->write_queue.prev;
if (tp->send_head &&
- (mss_now - skb->len) > 0) {
+ (signed int)(mss_now - skb->len) > 0) {
copy = skb->len;
if (skb_tailroom(skb) > 0) {
int last_byte_was_odd = (copy % 4);
Or would this be better?
+ (unsigned int)mss_now > skb->len) {
Or making mss_now unsigned in the first place?
--
Paul Cassella
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
