From owner-freebsd-hackers Mon Feb 5 20:53:50 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id UAA27791 for hackers-outgoing; Mon, 5 Feb 1996 20:53:50 -0800 (PST) Received: from tiny.mcs.usu.edu (tiny.mcs.usu.edu [129.123.15.8]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id UAA27786 for ; Mon, 5 Feb 1996 20:53:47 -0800 (PST) Received: (from kurto@localhost) by tiny.mcs.usu.edu (8.6.11/8.6.9) id VAA05705 for freebsd-hackers@freefall.cdrom.com; Mon, 5 Feb 1996 21:53:42 -0700 Date: Mon, 5 Feb 1996 21:53:42 -0700 From: Kurt Olsen Message-Id: <199602060453.VAA05705@tiny.mcs.usu.edu> To: freebsd-hackers@freefall.FreeBSD.org Subject: silo overflows Sender: owner-hackers@FreeBSD.ORG Precedence: bulk Hello, I recently installed a machine with a BOCA 2016 multiport serial card in it and have been seeing the occasional (several times a day) silo overflow. The board has 16 16550A's on it (as reported during boot up.) Looking in a manual I have here, it indicates that it's possible to set the receive trigger level to 1, 4, 8, or 14. Would there be any chance of clearing up the problem by moving down to 8 or 4? I'm not really concerned about too much cpu load, but data loss is kind of a problem since it leaves the modems in a bit of a state. If this seems reasonable what would be the best place to adjust it at? Obviously in sio.c, I suspect in the probe routine after determining the UART type, and setting ftl_init to FIFO_TRIGGER_8 or _4. Is that correct? Thanks a bunch, Kurt Olsen kurto@tiny.mcs.usu.edu