From owner-freebsd-hackers Wed Jan 10 05:28:20 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id FAA06098 for hackers-outgoing; Wed, 10 Jan 1996 05:28:20 -0800 (PST) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id FAA06077 for ; Wed, 10 Jan 1996 05:28:00 -0800 (PST) Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id AAA16547; Thu, 11 Jan 1996 00:15:09 +1100 Date: Thu, 11 Jan 1996 00:15:09 +1100 From: Bruce Evans Message-Id: <199601101315.AAA16547@godzilla.zeta.org.au> To: frank@fwi.uva.nl, freebsd-hackers@FreeBSD.ORG Subject: Re: NFS3 query Sender: owner-hackers@FreeBSD.ORG Precedence: bulk >> 3. Making of vnode_if.[ch] for the kernel on the client with /usr NFSv3- >> mounted gives corrupted files (with small pieces missing or something like >> that). >So I tried it and, sure enough, it failed. It only happens when v3 is used, >without NQNFS extensions. It seems that the last buffer of the files is not >written to the server, creating truncated files. >It isn't one of the 2 bugs reported in the patches that Rick put up for ftp, >I checked that. >My question is: did you guys already fix this one (I guess so, since it >was reported 4 months ago), and how did you fix it? It's not fixed. Random pieces get lost. One ktrace run showed the second 512-byte write being completely lost. The i/o is nothing special: ... read(0x0,0x2c100,0x200) break(0x2f000) fstat(0x1,...) break(0x30000) ... more breaks write(0x1,0x2f000,0x200) succeeds, but not present in final file ... no lseeks on fd 1 Bruce