From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from userp2120.oracle.com ([156.151.31.85]:45834 "EHLO userp2120.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750795AbdLMGUV (ORCPT ); Wed, 13 Dec 2017 01:20:21 -0500 Date: Tue, 12 Dec 2017 22:18:25 -0800 From: "Darrick J. Wong" Subject: [ZOMG RFCRAP PATCH 0/2] xfs: horrifying eBPF hacks Message-ID: <20171213061825.GO19219@magnolia> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: linux-xfs@vger.kernel.org Cc: Richard Wareing , david@fromorbit.com, hch@infradead.org Heh. So here's a kernel patch that builds on Josef's eBPF return value override patch series to provide an eBPF-kprobe-overridable hook function so that administrators can program XFS to redirect a file to the rt device or the data device depending on their own funny constraints any time that the system is doing the first write into an empty file. The second patch is against Brendan Gregg's bcc repository; it adds a python script to compile and inject a sample eBPF program that behaves (roughly) the same as Richard's earlier patches. Soooo... rather than dumping a bunch of static code into XFS to support his particular usecase, we're building him an eBPF Bazooka and telling him to have fun. :P (More generally, it's a science fair project for letting people customize XFS behavior with eBPF in a controlled manner.) --D