From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id EF2A0927 for ; Tue, 9 Sep 2014 19:02:47 +0000 (UTC) Received: from mail-ob0-f172.google.com (mail-ob0-f172.google.com [209.85.214.172]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id B83171FB58 for ; Tue, 9 Sep 2014 19:02:46 +0000 (UTC) Received: by mail-ob0-f172.google.com with SMTP id wp18so1313878obc.17 for ; Tue, 09 Sep 2014 12:02:45 -0700 (PDT) MIME-Version: 1.0 Date: Tue, 9 Sep 2014 21:02:45 +0200 Message-ID: From: Andreas Ruprecht To: josh@joshtriplett.org Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Cc: ksummit-discuss@lists.linuxfoundation.org Subject: [Ksummit-discuss] Linux Kernel Tinification - Research project and Tools List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Hi guys, I just discovered the thread (reaching back to May) on the ksummit-discuss mailing list and your (Josh's) talk topic at LinuxCon Europe on Linux Kernel Tinification, and might have something interesting to share. I'm a member of the VAMOS/CADOS[0] project at the University of Erlangen-N=C3=BCrnberg, where we're working on different topics concerning the analysis of Linux, particularily the kernel configuration system. One of these projects is "Tailor"[1], where we developed a method of generating an (almost) minimal kernel configuration, basically by observing a "bigger" (might be distribution-provided) kernel and solving a formula of feature constraints for a solution. Results for this were published at various conferences (HotDep '12 [2], NDSS '13 [3]), and just next week, I will be at GPCE '14 in Vasteras, Sweden to present our most recent work [4]. Probably the most important difference to note is that in earlier versions, we used the ftrace function tracer to gain insights on the control flow - so we did not need to patch the kernel -, whereas we found that for embedded systems, we needed to patch tiny instructions into conditional blocks to create usable results (ftrace was just too much overhead). Unfortunately, the code for patching the kernel has not yet been integrated into our tool box, as the student developing it is just about to finish his master's thesis, but if you are interested, we could also make this available to you. If you would like to try it out, please go ahead (instructions at [1])! If there are problems, don't hesitate to contact me. I will also be at Linux Plumbers Conference in D=C3=BCsseldorf in October, if you'd like to discuss this further in person. Best regards, Andreas Ruprecht [0]: https://www4.cs.fau.de/Research/CADOS/ [1]: http://vamos.informatik.uni-erlangen.de/trac/undertaker/wiki/Undertake= rTailor [2]: https://www4.cs.fau.de/Publications/2012/tartler_12_hotdep.pdf [3]: https://www4.cs.fau.de/Publications/2013/kurmus_13_ndss.pdf [4]: https://www4.cs.fau.de/Publications/2014/ruprecht_14_gpce.pdf