Kernel 2.4 test patchlets

[Pete Zaitcev]

Hi, David:

The current policy is that we have to create test branches, which produces
a small chilling effect. I noticed you really started to crank out those
kernels when you took over 2.4, just like in good old days. What do you
think, can you ship something like the attached in some "next" release?
It's for bz#116011.

Thanks,
-- Pete

--- linux-2.4.22-1.2176/drivers/usb/serial/usbserial.c	2004-03-11 20:53:43.000000000 -0800
+++ linux-2.4.26-pre2-nip/drivers/usb/serial/usbserial.c	2004-03-11 22:26:04.000000000 -0800
@@ -717,9 +717,14 @@
 		 * It's a ususal thing on serial to lose characters, isn't it?
 		 * Neener, neener! Actually, it's probably an echo loop anyway.
 		 * Only happens when getty starts talking to Visor.
+		 *
+		 * XXX Seems that ppp likes to do this as well.
+		 * Adding dump_stack to catch this.
 		 */
-		if (++rate % 1000 < 5)
+		if (++rate % 1000 < 3) {
 			err("too much data (%d)", count);
+			dump_stack();
+		}
 		job->len = POST_BSIZE;
 	}
 	memcpy(job->buff, buf, job->len);