From owner-cvs-sbin Mon Oct 14 14:42:51 1996 Return-Path: owner-cvs-sbin Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id OAA27589 for cvs-sbin-outgoing; Mon, 14 Oct 1996 14:42:51 -0700 (PDT) Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id OAA27582; Mon, 14 Oct 1996 14:42:45 -0700 (PDT) Received: (from bde@localhost) by godzilla.zeta.org.au (8.7.6/8.6.9) id HAA13903; Tue, 15 Oct 1996 07:38:13 +1000 Date: Tue, 15 Oct 1996 07:38:13 +1000 From: Bruce Evans Message-Id: <199610142138.HAA13903@godzilla.zeta.org.au> To: fenner@parc.xerox.com, j@uriah.heep.sax.de Subject: Re: cvs commit: src/sbin/savecore savecore.c Cc: cvs-all@freefall.freebsd.org, CVS-committers@freefall.freebsd.org, cvs-sbin@freefall.freebsd.org, joerg@freefall.freebsd.org Sender: owner-cvs-sbin@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk >> Sorry, I thought I had made sure that everything was in K. Unless I >> missed something, the patch in the PR causes minfree to be specified in >> bytes, which is not what the man page says. > >Yep, the patch in the PR broke with the tradition. My own (not >committed) version also used byte values for everything (with >`dumpsize' also enlarged to off_t), to avoid the implicit truncation I think the committed version has some rounding problems from not using bytes for everything and rounding up as late as possible. However, writing the last K is unlikely to work anyway because statfs can't report the exact amount free in advance (it can't know how many indirect blocks there will be in advance). Thus setting minfree to 0 is bogus. Bruce