From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 144E711707 for ; Mon, 21 Aug 2023 19:52:24 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 8C7B0C433C8; Mon, 21 Aug 2023 19:52:23 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linuxfoundation.org; s=korg; t=1692647543; bh=buo1gXP+6ENs24Tr2jw2oL4Zh6jCjQRo6G9GFD6psB4=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=thSgdgWVLyvM5ldmX4DyqkSeGPCC6l/beB86VKgjG92en5MXx+q3xFmDeBVyuWkEO 4Mqz0QitKuhsZ7cJcfj+C4TvYq1+rjP5I0Obu0GOeCy11tUImxioc3x0CR8PBoZO65 4rD6j0bvwf44Bl/ZNlCD57YIdtPIQyNyT2mlPgRE= From: Greg Kroah-Hartman To: stable@vger.kernel.org Cc: Greg Kroah-Hartman , patches@lists.linux.dev, BassCheck , Tuo Li , Andreas Gruenbacher , Sasha Levin Subject: [PATCH 6.1 051/194] gfs2: Fix possible data races in gfs2_show_options() Date: Mon, 21 Aug 2023 21:40:30 +0200 Message-ID: <20230821194125.052528047@linuxfoundation.org> X-Mailer: git-send-email 2.41.0 In-Reply-To: <20230821194122.695845670@linuxfoundation.org> References: <20230821194122.695845670@linuxfoundation.org> User-Agent: quilt/0.67 Precedence: bulk X-Mailing-List: patches@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From: Tuo Li [ Upstream commit 6fa0a72cbbe45db4ed967a51f9e6f4e3afe61d20 ] Some fields such as gt_logd_secs of the struct gfs2_tune are accessed without holding the lock gt_spin in gfs2_show_options(): val = sdp->sd_tune.gt_logd_secs; if (val != 30) seq_printf(s, ",commit=%d", val); And thus can cause data races when gfs2_show_options() and other functions such as gfs2_reconfigure() are concurrently executed: spin_lock(>->gt_spin); gt->gt_logd_secs = newargs->ar_commit; To fix these possible data races, the lock sdp->sd_tune.gt_spin is acquired before accessing the fields of gfs2_tune and released after these accesses. Further changes by Andreas: - Don't hold the spin lock over the seq_printf operations. Reported-by: BassCheck Signed-off-by: Tuo Li Signed-off-by: Andreas Gruenbacher Signed-off-by: Sasha Levin --- fs/gfs2/super.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-) diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c index 9d27aa8bd2bc6..44c564f0bc622 100644 --- a/fs/gfs2/super.c +++ b/fs/gfs2/super.c @@ -981,7 +981,14 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root) { struct gfs2_sbd *sdp = root->d_sb->s_fs_info; struct gfs2_args *args = &sdp->sd_args; - int val; + unsigned int logd_secs, statfs_slow, statfs_quantum, quota_quantum; + + spin_lock(&sdp->sd_tune.gt_spin); + logd_secs = sdp->sd_tune.gt_logd_secs; + quota_quantum = sdp->sd_tune.gt_quota_quantum; + statfs_quantum = sdp->sd_tune.gt_statfs_quantum; + statfs_slow = sdp->sd_tune.gt_statfs_slow; + spin_unlock(&sdp->sd_tune.gt_spin); if (is_ancestor(root, sdp->sd_master_dir)) seq_puts(s, ",meta"); @@ -1036,17 +1043,14 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root) } if (args->ar_discard) seq_puts(s, ",discard"); - val = sdp->sd_tune.gt_logd_secs; - if (val != 30) - seq_printf(s, ",commit=%d", val); - val = sdp->sd_tune.gt_statfs_quantum; - if (val != 30) - seq_printf(s, ",statfs_quantum=%d", val); - else if (sdp->sd_tune.gt_statfs_slow) + if (logd_secs != 30) + seq_printf(s, ",commit=%d", logd_secs); + if (statfs_quantum != 30) + seq_printf(s, ",statfs_quantum=%d", statfs_quantum); + else if (statfs_slow) seq_puts(s, ",statfs_quantum=0"); - val = sdp->sd_tune.gt_quota_quantum; - if (val != 60) - seq_printf(s, ",quota_quantum=%d", val); + if (quota_quantum != 60) + seq_printf(s, ",quota_quantum=%d", quota_quantum); if (args->ar_statfs_percent) seq_printf(s, ",statfs_percent=%d", args->ar_statfs_percent); if (args->ar_errors != GFS2_ERRORS_DEFAULT) { -- 2.40.1