From owner-gcc@vger.rutgers.edu Sun Jul 2 08:04:40 1995 Return-Path: owner-gcc@vger.rutgers.edu Received: from coli-gate.coli.uni-sb.de (root@coli-gate.coli.uni-sb.de [134.96.68.11]) by jurix.jura.uni-sb.de (8.6.12/8.6.9) with ESMTP id IAA19794 for ; Sun, 2 Jul 1995 08:04:37 +0200 Received: from vger.rutgers.edu (majordomo@vger.rutgers.edu [128.6.190.2]) by coli-gate.coli.uni-sb.de (8.6.10/8.6.4) with ESMTP id IAA10990 for ; Sun, 2 Jul 1995 08:04:11 +0200 Received: (from majordomo@localhost) by vger.rutgers.edu (8.6.12/8.6.12) id BAA27742 for linux-gcc-outgoing; Sun, 2 Jul 1995 01:19:57 -0400 X-Authentication-Warning: vger.rutgers.edu: Processed from queue /var/spool/mqueue_post From: hjl@nynexst.com (H.J. Lu) Message-Id: <9507020503.AA12810@texas.nynexst.com> Subject: Re: Kernel 1.3.3 and NFS To: temp@temptation.interlog.com (Temptation) Date: Sun, 2 Jul 95 1:03:15 EDT Cc: linux-gcc@vger.rutgers.edu (linuxgcc), bonn@eskimo.com (David Bonn), kjahds@kjahds.com (Kenneth Albanowski), lmfken@lmf.ericsson.se (Kenneth Osterberg), ian@lasermoon.co.uk, mat@ardi.com (Mat Hostetter), ecn@clark.net (Eric C. Newton), doughera@lafcol.lafayette.edu (Andy Dougherty), brian@mathworks.com (Brian Bourgault), jchristy@bga.com (John W. Christy), craig@metrolink.com (Craig Groeschel), imp@village.org (Warner Losh), robf@www.willows.com (Rob Farnum), Huw.Rogers@ska.com (Huw Rogers), meissner@cygnus.com (Michael Meissner) In-Reply-To: ; from "Temptation" at Jul 1, 95 11:33 pm X-Mailer: ELM [version 2.3 PL11] Sender: owner-gcc@vger.rutgers.edu Precedence: bulk Status: RO > > > Speaking about NFS, here is what I have found. > Running 1.3.6 and all HJ's ELF bins from /private/... > (my NFS mounts always worked before by the way) > I've tried compiling different version of NFS, new and old, and the bins > from Slackware-beta-ELF (Eg. nfs-server-2.2beta2.tar was the newest I've > tried) all of them would compile, and run, but when I tried to do an ls > on a mount that had files it would hang, if the dir had no files, it > worked. anyways it was rpc.nfsd. everytime I compile it as ELF, it would > hang ls. I didn't try compiling as a.out, but I have an old rpc.nfsd that > is a.out and it works when I run rpc.portmap and rpc.mountd as ELF > compiled. > > If you knew all this already, forget I said anything, I just didn't see a > patch or a fix for it yet. > Oooooops. Red faced :-(. Please try this patch on libc 5.2.0. Thanks for the bug report. If someone had made a small test case, I wouldn't have downloaded the NFSD just to check out this bug :-). H.J. =================================================================== RCS file: /home/cvs/linux/libc/sysdeps/linux/seekdir.c,v retrieving revision 1.3 diff -c -r1.3 seekdir.c *** 1.3 1995/06/29 03:13:17 --- seekdir.c 1995/07/02 04:55:30 *************** *** 20,39 **** return; } switch (dir->dd_getdents) { case no_getdents: case have_getdents: - dir->dd_nextloc = lseek(dir->dd_fd, offset, SEEK_SET); - dir->dd_size = dir->dd_nextloc = 0; - break; - - /* We don't have to do anything since no readdir () is called yet. */ case unknown: break; default: errno = EBADF; break; } } --- 20,41 ---- return; } + #if 0 switch (dir->dd_getdents) { case no_getdents: case have_getdents: case unknown: + dir->dd_nextoff = lseek(dir->dd_fd, offset, SEEK_SET); + dir->dd_size = dir->dd_nextloc = 0; break; default: errno = EBADF; break; } + #else + dir->dd_nextoff = lseek(dir->dd_fd, offset, SEEK_SET); + dir->dd_size = dir->dd_nextloc = 0; + #endif } =================================================================== RCS file: /home/cvs/linux/libc/sysdeps/linux/telldir.c,v retrieving revision 1.3 diff -c -r1.3 telldir.c *** 1.3 1995/06/29 03:13:21 --- telldir.c 1995/07/02 04:54:04 *************** *** 23,39 **** switch (dir->dd_getdents) { - case unknown: - /* readdir () is not called yet. */ - offset = 0; - break; - case no_getdents: /* We are running the old kernel. This is the starting offset of the next readdir(). */ offset = lseek(dir->dd_fd, 0, SEEK_CUR); break; case have_getdents: /* The next entry. */ offset = dir->dd_nextoff; --- 23,36 ---- switch (dir->dd_getdents) { case no_getdents: /* We are running the old kernel. This is the starting offset of the next readdir(). */ offset = lseek(dir->dd_fd, 0, SEEK_CUR); break; + case unknown: + /* readdir () is not called yet. but seekdir () may be called. */ case have_getdents: /* The next entry. */ offset = dir->dd_nextoff;