Date: Sun, 12 Oct 1997 09:34:54 +0930 From: Greg Lehey <grog@lemis.com> To: parrothd@midwest.net Cc: freebsd-questions@FreeBSD.ORG Subject: Re: tty-level buffer overflows Message-ID: <19971012093454.31985@lemis.com> In-Reply-To: <3433AF01.1578@midwest.net>; from bla bla on Thu, Oct 02, 1997 at 09:26:09AM -0500 References: <3.0.3.32.19971001225109.006e88f0@midwest.net> <19971002142420.49578@lemis.com> <3433AF01.1578@midwest.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Oct 02, 1997 at 09:26:09AM -0500, bla bla wrote: > Greg Lehey wrote: >> >> On Wed, Oct 01, 1997 at 10:51:09PM -0500, Jonathan E. Lyons wrote: >>> Is this anything to be concerned about? I've got an X2 modem, with the port >>> speed set at 57600..... >>> >>> >>> Sep 28 16:42:11 cplkagan /kernel: sio0: 660 more tty-level buffer overflows >>> (total 660) >>> Sep 29 13:52:41 cplkagan /kernel: pid 18534 (ping), uid 0: exited on signal 3 >>> Sep 30 21:05:02 cplkagan /kernel: sio0: 100 more tty-level buffer overflows >>> (total 760) >>> Sep 30 22:00:02 cplkagan /kernel: sio0: 198 more tty-level buffer overflows >>> (total 958) >>> Sep 30 22:05:01 cplkagan /kernel: sio0: 199 more tty-level buffer overflows >>> (total 1157) >>> Sep 30 22:05:02 cplkagan /kernel: sio0: 1076 more tty-level buffer >>> overflows (total 2233) >>> Oct 1 05:30:02 cplkagan /kernel: sio0: 940 more tty-level buffer overflows >>> (total 3173) >> >> Yes. With PPP, each of these means a lost packet, which is expensive. >> This shouldn't happen. Is the machine slow or heavily loaded? >> Otherwise you might be losing interrupts. >> >> Greg > > Depends on your version of slow,:), it's a 486/66 8megs of ram, running > ppp -alias, for a small house LAN. It does however has other processes > running, but whenever I do a top -s 1 about %90 of the machine is idle, > until someone starts to dl, or hits the Web server from the local LAN. > Could it be the serial ports itself? The board it self has one built-in > serial port, but I didn't think it could handle the speed(it's an old > Packard Hell MB) so I threw in a multi I/O card, trying to avoid serial > problems... The CPU has enough power, but with 8 MB you're going to be doing a lot of swapping, during which I think the async interrupts are locked out. I'd consider this fits my description of "heavily loaded", even if the CPU is 90% idle. This could be your problem. Greg
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19971012093454.31985>