Date: Mon, 3 Mar 1997 06:43:06 -0800 (PST) From: rtm@eecs.harvard.edu To: freebsd-gnats-submit@freebsd.org Subject: kern/2857: DE500 board exhibits capture effect Message-ID: <199703031443.GAA04861@freefall.freebsd.org> Resent-Message-ID: <199703031450.GAA05492@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 2857 >Category: kern >Synopsis: DE500 board exhibits capture effect >Confidential: no >Severity: non-critical >Priority: low >Responsible: freebsd-bugs >State: open >Class: sw-bug >Submitter-Id: current-users >Arrival-Date: Mon Mar 3 06:50:01 PST 1997 >Last-Modified: >Originator: Robert Morris >Organization: Harvard University >Release: 2.1.5 >Environment: FreeBSD punish.uun.org 2.1.5-RELEASE FreeBSD 2.1.5-RELEASE #0: Tue Feb 4 16:42:28 EST 1997 rtm@punish.uun.org:/usr/src/sys/compile/PUNISH i386 >Description: I have some DE500 boards running at 10 megabits with FreeBSD 2.1.5. I believe I've run into the problem that the 21x40 capture effect resolution was meant to solve: if one host is doing a long file transfer, other hosts drop outgoing packets due to excessive re-transmits. When I turn on CAPTREFFCT and ENHCAPTEFFCT in if_de.c, the problem goes away (for one greedy host, anyway; multiple greedy hosts still cause problems). FreeBSD says this about my cards: de0 <Digital DC21140 Fast Ethernet> rev 18 int a irq 10 on pci0:12 de0: Digital DE500 DC21140 [10-100Mb/s] pass 1.2 Ethernet address 00:00:f8:02:36 :c5 >How-To-Repeat: Transfer a file between two DE500 hosts at 10 megabits. Ping between two other hosts on the same net. Observe ping times of tens or hundreds of milliseconds rather than the expected 1 or 2. >Fix: Turn on TULIP_CMD_CAPTREFFCT and TULIP_CMD_ENHCAPTEFFCT in tulip_dc21140_de500_media_select in if_de.c. I have done this and observed no ill effects. I have not tried this with other boards or at 100 megabits. >Audit-Trail: >Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199703031443.GAA04861>