linux-kernel - Re: data-race in cgroup_get_tree / proc_cgroup

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <YwuySgH4j6h2CGvk@slm.duckdns.org>
Date:   Sun, 28 Aug 2022 08:22:02 -1000
From:   Tejun Heo <tj@...nel.org>
To:     Gabriel Ryan <gabe@...columbia.edu>
Cc:     Christian Brauner <brauner@...nel.org>,
        Abhishek Shah <abhishek.shah@...umbia.edu>,
        linux-kernel@...r.kernel.org, andrii@...nel.org, ast@...nel.org,
        bpf@...r.kernel.org, cgroups@...r.kernel.org, daniel@...earbox.net,
        hannes@...xchg.org, john.fastabend@...il.com, kafai@...com,
        kpsingh@...nel.org, lizefan.x@...edance.com,
        netdev@...r.kernel.org, songliubraving@...com, yhs@...com
Subject: Re: data-race in cgroup_get_tree / proc_cgroup_show

On Mon, Aug 22, 2022 at 01:04:58PM -0400, Gabriel Ryan wrote:
> Hi Christian,
> 
> We ran a quick test and confirm your suggestion would eliminate the
> data race alert we observed. If the data race is benign (and it
> appears to be), using WRITE_ONCE(cgrp_dfl_visible, true) instead of
> cmpxchg in cgroup_get_tree() would probably also be ok.

I don't see how the data race can lead to anything but would the following
work?

Thanks.

diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index ffaccd6373f1e..a90fdba881bdb 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -2172,7 +2172,7 @@ static int cgroup_get_tree(struct fs_context *fc)
 	struct cgroup_fs_context *ctx = cgroup_fc2context(fc);
 	int ret;
 
-	cgrp_dfl_visible = true;
+	WRITE_ONCE(cgrp_dfl_visible, true);
 	cgroup_get_live(&cgrp_dfl_root.cgrp);
 	ctx->root = &cgrp_dfl_root;
 
@@ -6056,7 +6056,7 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns,
 		struct cgroup *cgrp;
 		int ssid, count = 0;
 
-		if (root == &cgrp_dfl_root && !cgrp_dfl_visible)
+		if (root == &cgrp_dfl_root && !READ_ONCE(cgrp_dfl_visible))
 			continue;
 
 		seq_printf(m, "%d:", root->hierarchy_id);


-- 
tejun