From patchwork Wed Aug 29 17:24:04 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Axel Burri X-Patchwork-Id: 10580711 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 2EAB65A4 for ; Wed, 29 Aug 2018 17:31:47 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 14FA52B8D5 for ; Wed, 29 Aug 2018 17:31:47 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 11AFF2B914; Wed, 29 Aug 2018 17:31:47 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00,MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id DF7652B8D5 for ; Wed, 29 Aug 2018 17:31:44 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727605AbeH2V3j (ORCPT ); Wed, 29 Aug 2018 17:29:39 -0400 Received: from tty0.vserver.softronics.ch ([91.214.169.36]:53536 "EHLO fe1.digint.ch" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727460AbeH2V3j (ORCPT ); Wed, 29 Aug 2018 17:29:39 -0400 Received: by fe1.digint.ch (Postfix, from userid 1000) id E0E3D307E2; Wed, 29 Aug 2018 19:23:10 +0200 (CEST) Received: from mail.tty0.ch (hermes.tty0.ch [172.21.1.5]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by fe1.digint.ch (Postfix) with ESMTPS id 2A7B8307D2 for ; Wed, 29 Aug 2018 19:23:10 +0200 (CEST) Received: from ares.tty0.ch (ares.tty0.ch [10.0.1.10]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.tty0.ch (Postfix) with ESMTPS id 154CB1DA0EF; Wed, 29 Aug 2018 19:22:39 +0200 (CEST) Received: by ares.tty0.ch (sSMTP sendmail emulation); Wed, 29 Aug 2018 19:24:44 +0200 From: Axel Burri To: linux-btrfs@vger.kernel.org Cc: Axel Burri Subject: [RFC PATCH 1/6] btrfs-progs: splitcmd-gen.sh: create btrfs- binaries for selected subcommands Date: Wed, 29 Aug 2018 19:24:04 +0200 Message-Id: <20180829172409.18064-2-axel@tty0.ch> X-Mailer: git-send-email 2.16.4 In-Reply-To: <20180829172409.18064-1-axel@tty0.ch> References: <20180829172409.18064-1-axel@tty0.ch> Sender: linux-btrfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-btrfs@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP Create separate binaries for each subcommand ("btrfs foo bar"). Least invasive approach, generate c-files for each command: # ./splitcmd-gen.sh # make V=1 btrfs-subvolume-show # make V=1 btrfs-send # [...] Alternative approach: instead of including the c-file, link with obj in Makefile, e.g.: btrfs_subvolume_show_objects = cmds-subvolume.o btrfs_send_objects = cmds-send.o [...] This implies adaptions in cmds-subvolume.c (and others): -static int cmd_filesystem_show(int argc, char **argv) +int cmd_filesystem_show(int argc, char **argv) If they are defined non-static, we could probably simplify further and add `-Wl,-eentry` flags (changing entry point from "main" to "entry"). With this, and if handle_command_group() was declared in some library instead of btrfs.c, we would get rid of generated files completely. Signed-off-by: Axel Burri --- splitcmd-gen.sh | 70 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ splitcmd.c.in | 17 ++++++++++++++ 2 files changed, 87 insertions(+) create mode 100755 splitcmd-gen.sh create mode 100644 splitcmd.c.in diff --git a/splitcmd-gen.sh b/splitcmd-gen.sh new file mode 100755 index 00000000..4d2e0509 --- /dev/null +++ b/splitcmd-gen.sh @@ -0,0 +1,70 @@ +#!/bin/bash + +# +# Generate c-files for btrfs subcommands defined below +# + +# Notes on linux capabilities: +# +# btrfs-subvolume-show, btrfs-subvolume-list, btrfs-send: +# - CAP_FOWNER is only needed for O_NOATIME flag in open() system calls +# - why CAP_SYS_ADMIN? shouldn't CAP_DAC_READ_SEARCH be enough? +# +# btrfs-receive: +# - dependent on send-stream (see cmds-receive.c: "send_ops"): +# CAP_CHOWN, CAP_MKNOD, CAP_SETFCAP (for "lsetxattr") +# +# btrfs-filesystem-usage: +# - CAP_SYS_ADMIN is for BTRFS_IOC_TREE_SEARCH and BTRFS_IOC_FS_INFO +# in order to provide full level of detail, see btrfs-filesystem(8) + + +makefile_out="Makefile.install_setcap" + +splitcmd_list="" +setcap_lines="" + +function gen_splitcmd { + local name="$1" + local dest="${1}.c" + local cfile="$2" + local entry="$3" + local caps="$4" + echo "generating: ${dest} (cfile=${cfile}, entry=${entry})" + echo -e "/*\n * ${name}\n *\n * GENERATED BY splitcmd-gen.sh\n */\n" > $dest + sed -e "s|@BTRFS_SPLITCMD_CFILE_INCLUDE@|${cfile}|g" \ + -e "s|@BTRFS_SPLITCMD_ENTRY@|${entry}|g" \ + splitcmd.c.in >> $dest +} + +gen_splitcmd "btrfs-subvolume-show" \ + "cmds-subvolume.c" "cmd_subvol_show" \ + "cap_sys_admin,cap_fowner,cap_dac_read_search" + +gen_splitcmd "btrfs-subvolume-list" \ + "cmds-subvolume.c" "cmd_subvol_list" \ + "cap_sys_admin,cap_fowner,cap_dac_read_search" + +gen_splitcmd "btrfs-subvolume-snapshot" \ + "cmds-subvolume.c" "cmd_subvol_snapshot" \ + "cap_sys_admin,cap_fowner,cap_dac_override,cap_dac_read_search" + +gen_splitcmd "btrfs-subvolume-delete" \ + "cmds-subvolume.c" "cmd_subvol_delete" \ + "cap_sys_admin,cap_dac_override" + +gen_splitcmd "btrfs-send" \ + "cmds-send.c" "cmd_send" \ + "cap_sys_admin,cap_fowner,cap_dac_read_search" + +gen_splitcmd "btrfs-receive" \ + "cmds-receive.c" "cmd_receive" \ + "cap_sys_admin,cap_fowner,cap_chown,cap_mknod,cap_setfcap,cap_dac_override,cap_dac_read_search" + +gen_splitcmd "btrfs-filesystem-usage" \ + "cmds-fi-usage.c" "cmd_filesystem_usage" \ + "cap_sys_admin" + +gen_splitcmd "btrfs-qgroup-destroy" \ + "cmds-qgroup.c" "cmd_qgroup_destroy" \ + "cap_sys_admin,cap_dac_override" diff --git a/splitcmd.c.in b/splitcmd.c.in new file mode 100644 index 00000000..aa07af9a --- /dev/null +++ b/splitcmd.c.in @@ -0,0 +1,17 @@ +#include "@BTRFS_SPLITCMD_CFILE_INCLUDE@" + +/* + * Dummy object: used from second-level command groups (e.g. in + * "cmds-subvolume.c"), is never called in splitcmd executables. + */ +int handle_command_group(const struct cmd_group *grp, int argc, + char **argv) +{ + exit(1); +} + + +int main(int argc, char **argv) +{ + return @BTRFS_SPLITCMD_ENTRY@(argc, argv); +}