From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Subject: Re: [PATCH v5] fs: make dumpable=2 require fully qualified path Date: Mon, 25 Jun 2012 15:28:53 -0700 Message-ID: <20120625152853.1ec6a824.akpm@linux-foundation.org> References: <20120625180327.GA31931@www.outflux.net> <20120625151409.5e13702b.akpm@linux-foundation.org> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: linux-kernel@vger.kernel.org, Rob Landley , Alexander Viro , Alan Cox , "Eric W. Biederman" , Doug Ledford , Serge Hallyn , James Morris , Joe Korty , linux-doc@vger.kernel.org, linux-fsdevel@vger.kernel.org To: Kees Cook Return-path: In-Reply-To: Sender: linux-doc-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Mon, 25 Jun 2012 15:20:19 -0700 Kees Cook wrote: > On Mon, Jun 25, 2012 at 3:14 PM, Andrew Morton > wrote: > > On Mon, 25 Jun 2012 11:03:27 -0700 > > Kees Cook wrote: > > > >> Instead of introducing a suid_dumpable=3 mode and breaking all users > >> of mode 2, this only disables the unsafe portion of mode 2 (writing to > >> disk via relative path). > > > > It would be nicer to generate the warning at configuration time rather than > > at core-dumping time. __I have a feeling that's Hard To Do, but why? > > I actually feel that this is much more discoverable because anyone > trying to figure out why a crash isn't dumping core isn't going to be > looking back in time through their dmesg to find the warning from > setting the sysctl. Additionally, I think it's the same kind of thing > that the other existing "Skipping core dump" printk messages are > reporting. People who don't discover we changed their kernel until after they lost an important core dump will be sad. Do both?