From owner-freebsd-stable Sun Oct 26 01:12:11 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id BAA24392 for stable-outgoing; Sun, 26 Oct 1997 01:12:11 -0700 (PDT) (envelope-from owner-freebsd-stable) Received: from soul.irex.urc.ac.ru (soul.irex.urc.ac.ru [193.233.85.100]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id BAA24368 for ; Sun, 26 Oct 1997 01:11:42 -0700 (PDT) (envelope-from anton@urc.ac.ru) Received: from urc.ac.ru (localhost [127.0.0.1]) by soul.irex.urc.ac.ru (8.8.7/8.8.5) with ESMTP id OAA00804 for ; Sun, 26 Oct 1997 14:16:02 GMT Message-ID: <345350A1.ECE2AE98@urc.ac.ru> Date: Sun, 26 Oct 1997 14:16:01 +0000 From: Anton Voronin Organization: URC FREEnet X-Mailer: Mozilla 4.03b8 [en] (X11; I; FreeBSD 2.2-STABLE i386) MIME-Version: 1.0 To: stable@freebsd.org Subject: Kernel's nfs dramatically slows down a diskless workstation Content-Type: text/plain; charset=koi8-r Content-Transfer-Encoding: 7bit Sender: owner-freebsd-stable@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi, this problem appeared in stable a time ago, I was waiting for 2.2.5 issue with a hope it will be changed there, but it still exists. I have pseudo-diskless workstations (only swap, /tmp, /var/tmp and 2M ibm-dos partition for booting are on local disks; the rest is mounted from the server) each has only 8M RAM. When I compile and install a new kernel for one, it becomes very-very slow. Systat -vmstat shows that amount of interrupts take up to 78% of time, and up to 700 interrupts per second are produced by ed0 installed on IRQ10. The kernel compiled on 2.2-release works x*10 times faster (at least it makes me feel so). I tried to localize the changes between the modern stable and 2.2-release source code leading to this problem and found that if I take only nfs.h, nfs_bio.c and nfs_vnopts.c from 2.2-release (located in /sys/nfs/) and place them instead of the same -stable files, kernel after building works pretty enough. Because of nfs.h and nfs_vnopts.h differ in only function prototypes and calls, possibly the critical changes are located in nfs_bio.c. Can it be fixed or returned to previous version? Thanks, Anton -- Anton Voronin | Ural Regional Center of FREEnet, | Technical University of Chelyabinsk, Russia http://www.urc.ac.ru/~anton | Student / programmer / system administrator