From owner-freebsd-current@FreeBSD.ORG Sun Feb 22 22:24:46 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id A2F5B16A4CE for ; Sun, 22 Feb 2004 22:24:46 -0800 (PST) Received: from omoikane.mb.skyweb.ca (omoikane.mb.skyweb.ca [64.42.246.20]) by mx1.FreeBSD.org (Postfix) with ESMTP id 586E043D1F for ; Sun, 22 Feb 2004 22:24:46 -0800 (PST) (envelope-from mark@skyweb.ca) Received: by omoikane.mb.skyweb.ca (Postfix, from userid 1001) id 161FD631B0; Mon, 23 Feb 2004 00:25:03 -0600 (CST) Date: Mon, 23 Feb 2004 00:25:03 -0600 From: Mark Johnston To: current@freebsd.org Message-ID: <20040223062502.GA717@omoikane.mb.skyweb.ca> Mail-Followup-To: current@freebsd.org Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.4.1i Subject: cvs-src summary for 22/02/04 X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 23 Feb 2004 06:24:46 -0000 My apologies to those whose e-mails I haven't replied to this weekend - hopefully I'll have a chance to sit down with my e-mail next week. I've had quite a busy weekend, which also explains the lateness of the summary. Without further ado: FreeBSD cvs-src summary for 15/02/04 to 22/02/04 ++++++++++++++++++++++++++++++++++++++++++++++++ This is a regular weekly summary of FreeBSD's cutting-edge development. It is intended to help the FreeBSD community keep up with the fast-paced work going on in FreeBSD-CURRENT by distilling the deluge of data from the CVS mailing list into a (hopefully) easy-to-read newsletter. This newsletter is marked up in reStructuredText_, so any odd punctuation that you see is likely intended for the reST parser. .. _reStructuredText: http://docutils.sourceforge.net/rst.html You can get old summaries, and an HTML version of this one, at http://www.xl0.org/FreeBSD/. Please send any comments to Mark Johnston (mark at xl0.org). .. contents:: ============ New features ============ PThreads barrier support added ------------------------------ Mike Makonnen (mtm) added support for PThreads barriers and barrier attributes to the libthr threading library. In threading terms, a barrier is a way of synchronizing many threads together. Each thread that synchronizes with (calls) the barrier is suspended. When enough threads are suspended, the barrier lets them all run again at once. http://docs.freebsd.org/cgi/mid.cgi?200402191351.i1JDprSG065022 GEOM disk concatenation support added ------------------------------------- Pawel Jakub Dawidek (pjd) added support to GEOM for disk concatenation. Disk concatenation allows multiple physical storage devices to be used as though they were one big disk. He also committed a utility called gconcat, used to create, destroy, and label GEOM concatenated-disk devices. GEOM support: http://docs.freebsd.org/cgi/mid.cgi?200402191519.i1JFJoYQ084836 gconcat: http://docs.freebsd.org/cgi/mid.cgi?200402191602.i1JG29pe094944 New Sendmail version 8.12.11 merged to 4.x ------------------------------------------ Gregory Neil Shapiro (gshapiro) merged Sendmail 8.12.11 to FreeBSD 4.x. This is the same update applied to -CURRENT `last week`_. Please see the `Sendmail 8.12.11 release notes`_ for details on the minor changes that the new version involves. .. _`last week`: http://excel.xl0.org/FreeBSD/15-02-04.html#id8 .. _`Sendmail 8.12.11 release notes`: http://www.sendmail.org/8.12.11.html Syslog support added to nologin(8) ---------------------------------- Colin Percival (cperciva) committed an update to nologin, the program that displays a message to accounts that are not allowed to log in, to have it send a syslog message when a locked-out user tries to log in. This increaes the binary size by 100 kb, so he has added an option to disable it. http://docs.freebsd.org/cgi/mid.cgi?200402221003.i1MA3PW0024791 cleanworld target added to the system Makefile ---------------------------------------------- Garance A. Drosehn (gad) added a target to the system Makefile called "cleanworld". This makes the command "make cleanworld" in /usr/src that removes all the files that would have been created by "make buildworld" and "make buildkernel". http://docs.freebsd.org/cgi/mid.cgi?200402162018.i1GKIQK4029731 Basic large file support for ext2 filesystems --------------------------------------------- Tim J. Robbins (tjr) added support for large files on ext2 filesystems. The code works only on filesystems that are compatible with recent ext2 code, as well as needing a compatibility flag set. http://docs.freebsd.org/cgi/mid.cgi?200402181408.i1IE8QaL004568 ================= Committer changes ================= Alfred Perlstein (alfred) introduced Jim Rees (rees) as a new src committer. Jim will be working on the client and server for version 4 of the NFS (Network File System) file-sharing system. http://docs.freebsd.org/cgi/mid.cgi?200402170216.i1H2GuU4058858 Paul Saab (psaab) introduced Vinod Kashyap (vkashyap), a new src committer, who will be working on the drivers for 3ware's ATA RAID adapters. http://docs.freebsd.org/cgi/mid.cgi?200402202335.i1KNZxrO075548 ================= Discussion topics ================= Outputting line numbers when panicking, revisited ------------------------------------------------- As mentioned in the `January 23rd summary`_, Poul-Henning Kamp (phk) had made a commit that added filenames and line numbers to panic messages. The discussion surrounding that change was revived late last week as Bruce Evans (bde) pointed out that the discussion had been left open. Nate Lawson (njl) mentioned that he had already committed a patch by Ian Dowse (iedowse) that causes backtraces to be properly logged. Marcel Moolenaar (marcel) suggested that KASSERT(), the macro used for assertions, could be modified to print a file and line number. John Baldwin (jhb) suggested the MPASS() macro, which already outputs the line number, instead, and some discussion followed about renaming MPASS() to ASSERT(). .. _`January 23rd summary`: http://excel.xl0.org/FreeBSD/23-01-04.html#id5 Bruce's original message: http://docs.freebsd.org/cgi/mid.cgi?20040213035557.M24031 Nate's comment: http://docs.freebsd.org/cgi/mid.cgi?20040217083910.M29569 John's MPASS suggestion: http://docs.freebsd.org/cgi/mid.cgi?200402180839.09285.jhb M_WAITOK memory allocation flag modified ---------------------------------------- Dag-Erling Smorgrav (des) committed code to the memory subsystem allowing it to fail and return NULL when a request for memory with the M_WAITOK flag could not be satisfied. This started a discussion about whether the change in the meaning of M_WAITOK was good. The problem stems from wanting the ability to request arbitrary amounts of memory in kernel code without risking that the request will cause a panic. There were numerous requests to back out the change in favor of some other workaround to the problem. Maxime Henrion (mux) mentioned having written a patch that adds that ability; some discussion followed over whether the functionality is worthwhile. Dag-Erling backed out the commit on Monday. http://docs.freebsd.org/cgi/mid.cgi?200402161841.i1GIfwmv008267 Maxime's post: http://docs.freebsd.org/cgi/mid.cgi?20040216210503.GC35475 Dag-Erling's backout: http://docs.freebsd.org/cgi/mid.cgi?200402162136.i1GLaxWG053281 Properly removing vendor files from -CURRENT in CVS --------------------------------------------------- Andrey Chernov (ache) removed some files related to GNU getopt from the system. Ollivier Robert pointed out that the commit took the files off of the vendor CVS branch. There was some discussion about how files on the vendor branch should be removed when the need arises. For the full thread, which requires deep knowledge of the FreeBSD CVS management policy (or at least more than I have) to understand, please use the link below. http://docs.freebsd.org/cgi/mid.cgi?200402170343.i1H3hCot079054 Boot-time configuration with device.hints vs. loader.conf --------------------------------------------------------- Wes Peters (wes) committed code to the ACPI manual page explaining the difference between sysctl(8) variables and device.hints(5) tunables. This prompted discussion on the benefits of putting changes in loader.conf instead of device.hints. Neither file is automatically changed, but device.hints can be updated with new kernel versions. Nate Lawson (njl) also mentioned that loader.conf is empty by default, making it easier to read when troubleshooting. http://docs.freebsd.org/cgi/mid.cgi?200402172038.i1HKcnKY035565 Nate's comment: http://docs.freebsd.org/cgi/mid.cgi?20040217143034.B31176 Major modifications to device code ---------------------------------- Poul-Henning Kamp (phk) introduced significant upgrades to the system device handling code. The new code allows devices to be tracked much more precisely than they used to be, making bugs easier to track down and fix. It also frees most devices from using pre-allocated device numbers. The upgrade changes the API for drivers, so all third-party drivers will need to be updated to match. Poul-Henning has updated all of the system drivers for the new API. API change: http://docs.freebsd.org/cgi/mid.cgi?200402212110.i1LLAtEX008732 http://docs.freebsd.org/cgi/mid.cgi?200402212157.i1LLvQWp020145 =================== Important bug fixes =================== IPv6 packet filter now properly enabled --------------------------------------- Mike Makonnen (mtm) committed code to the startup scripts for the IPv6 version of ipfw, correcting a problem that caused the firewall not to be properly enabled after the rules were loaded. This closes `PR 61501`_, submitted by Roderick van Domburg. .. _`PR 61501`: http://www.freebsd.org/cgi/query-pr.cgi?pr=61501 http://docs.freebsd.org/cgi/mid.cgi?200402190653.i1J6rPlP057009 Coverity-reported bugs fixed ---------------------------- Ted Unangst of Coverity_ sent reports for a number of bugs that he had found with Coverity's automatic bug-detection software. The most serious bug fixed was in the RAIDframe RAID driver, where far too little memory was being allocated in some cases. These fixes were committed over the course of the week by Colin Percival (cperciva), Jim Rees (rees), and Soeren Schmidt (sos). Thanks to Colin Percival (cperciva) for filling me in on the Coverity story. .. _Coverity: http://www.coverity.com/ http://docs.freebsd.org/cgi/mid.cgi?200402220103.i1M13coA076447 Jailed processes moving around corrected ---------------------------------------- Jacques Vidrine (nectar) applied a fix to the jail code to prevent jailed processes from moving from one jail to another. http://docs.freebsd.org/cgi/mid.cgi?200402192103.i1JL3KhF076615 =============== Other bug fixes =============== Tim J. Robbins (tjr) modified the savecore(8) utility, which is used to save system crash dumps, so that it would properly save dumps larger than 2 GB. http://docs.freebsd.org/cgi/mid.cgi?200402201222.i1KCMqgJ004355 Bruce Evans (bde) committed a fix for a serious off-by-one error in the DOS filesystem code. This bug would have caused memory corruption under specific conditions when using DOS filesystems. http://docs.freebsd.org/cgi/mid.cgi?200402212247.i1LMlKJ5033386 Scott Long (scottl) fixed a recently-introduced panic in the aac driver for Adaptec SCSI controllers. http://docs.freebsd.org/cgi/mid.cgi?200402210532.i1L5WfBi068177 Andrey A. Chernov (ache) updated many system tools to use the system version of getopt_long(), which parses command-line options, instead of their own versions or the GNU version.