From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from sc8-sf-mx2-b.sourceforge.net ([10.3.1.12] helo=sc8-sf-mx2.sourceforge.net) by sc8-sf-list1.sourceforge.net with esmtp (Exim 4.30) id 1C9q3i-0007rA-QN for user-mode-linux-devel@lists.sourceforge.net; Tue, 21 Sep 2004 12:12:14 -0700 Received: from marasystems.com ([213.150.153.194] helo=filer.marasystems.com) by sc8-sf-mx2.sourceforge.net with esmtp (Exim 4.41) id 1C9q3g-00061Q-S4 for user-mode-linux-devel@lists.sourceforge.net; Tue, 21 Sep 2004 12:12:14 -0700 From: Henrik Nordstrom Subject: Re: [uml-devel] Managing 2.6 merge - url for the patches sent to merge In-Reply-To: <200409212001.43965.blaisorblade_personal@yahoo.it> Message-ID: References: <200409181944.58072.blaisorblade_personal@yahoo.it> <200409202030.49422.blaisorblade_personal@yahoo.it> <200409210321.i8L3LgdR015667@ccure.user-mode-linux.org> <200409212001.43965.blaisorblade_personal@yahoo.it> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Sender: user-mode-linux-devel-admin@lists.sourceforge.net Errors-To: user-mode-linux-devel-admin@lists.sourceforge.net List-Unsubscribe: , List-Id: The user-mode Linux development list List-Post: List-Help: List-Subscribe: , List-Archive: Date: Tue, 21 Sep 2004 21:11:45 +0200 (CEST) To: Paolo Giarrusso Cc: user-mode-linux-devel@lists.sourceforge.net, Jeff Dike , Gerd Knorr On Tue, 21 Sep 2004, Paolo Giarrusso wrote: >> Mysteriously, setting that descriptor >> CLOEXEC doesn't seem to help. > No, this problem seems to go away on 2.6 hosts (at least, that's my random > experience). It seems, so, a 2.4 host bug. > > See this patch from Henrik Normstrod for a workaround: > > http://marasystems.com/download/uml/10-flock.patch When I looked into this problem I concluded the fcntl locks in the host kernel got completely messed up by the cloning or something similar. Using flock works just as good assuming the ubd files is local and is not plauged by the same host bug. Most likely the host gets confused by the close being done from another pid than who issued the lock. The host bug is not limited to rebooting only, the lock is permanent and will stay forever. In fact I suspect the CLOEXEC flag as such is working, it is just that the host does not realize the lock belongs to this fd. This Linux kernel issue can probably be utilized to play some interesting DOS games on Linux systems where you have a account or otherwise the ability to run code, but I have not cared to investigate the exact details closely. Regards Henrik ------------------------------------------------------- This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 Project Admins to receive an Apple iPod Mini FREE for your judgement on who ports your project to Linux PPC the best. Sponsored by IBM. Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php _______________________________________________ User-mode-linux-devel mailing list User-mode-linux-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel