netdev - Re: [syzbot] KASAN: use-after-free Read in p9_req

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: <YvyD053bdbGE9xoo@codewreck.org>
Date:   Wed, 17 Aug 2022 14:59:47 +0900
From:   asmadeus@...ewreck.org
To:     syzbot <syzbot+de52531662ebb8823b26@...kaller.appspotmail.com>
Cc:     davem@...emloft.net, edumazet@...gle.com, ericvh@...il.com,
        kuba@...nel.org, linux-kernel@...r.kernel.org,
        linux_oss@...debyte.com, lucho@...kov.net, netdev@...r.kernel.org,
        pabeni@...hat.com, syzkaller-bugs@...glegroups.com,
        v9fs-developer@...ts.sourceforge.net
Subject: Re: [syzbot] KASAN: use-after-free Read in p9_req_put

syzbot having a fresh look at 9p?

Well at least that one should be easy enough, the following (untested)
probably should work around that issue:

-----
>From 433138e5d36a5b29b46b043c542e14b9dc908460 Mon Sep 17 00:00:00 2001
From: Dominique Martinet <asmadeus@...ewreck.org>
Date: Wed, 17 Aug 2022 14:49:29 +0900
Subject: [PATCH] 9p: p9_client_create: use p9_client_destroy on failure

If trans was connected it's somehow possible to fail with requests in
flight that could still be accessed after free if we just free the clnt
on failure.
Just use p9_client_destroy instead that has proper safeguards.

Reported-by: syzbot+de52531662ebb8823b26@...kaller.appspotmail.com
Signed-off-by: Dominique Martinet <asmadeus@...ewreck.org>

diff --git a/net/9p/client.c b/net/9p/client.c
index 5bf4dfef0c70..da5d43848600 100644
--- a/net/9p/client.c
+++ b/net/9p/client.c
@@ -948,7 +948,7 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
 
 	err = parse_opts(options, clnt);
 	if (err < 0)
-		goto free_client;
+		goto out;
 
 	if (!clnt->trans_mod)
 		clnt->trans_mod = v9fs_get_default_trans();
@@ -957,7 +957,7 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
 		err = -EPROTONOSUPPORT;
 		p9_debug(P9_DEBUG_ERROR,
 			 "No transport defined or default transport\n");
-		goto free_client;
+		goto out;
 	}
 
 	p9_debug(P9_DEBUG_MUX, "clnt %p trans %p msize %d protocol %d\n",
@@ -965,7 +965,7 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
 
 	err = clnt->trans_mod->create(clnt, dev_name, options);
 	if (err)
-		goto put_trans;
+		goto out;
 
 	if (clnt->msize > clnt->trans_mod->maxsize) {
 		clnt->msize = clnt->trans_mod->maxsize;
@@ -979,12 +979,12 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
 		p9_debug(P9_DEBUG_ERROR,
 			 "Please specify a msize of at least 4k\n");
 		err = -EINVAL;
-		goto close_trans;
+		goto out;
 	}
 
 	err = p9_client_version(clnt);
 	if (err)
-		goto close_trans;
+		goto out;
 
 	/* P9_HDRSZ + 4 is the smallest packet header we can have that is
 	 * followed by data accessed from userspace by read
@@ -997,12 +997,8 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
 
 	return clnt;
 
-close_trans:
-	clnt->trans_mod->close(clnt);
-put_trans:
-	v9fs_put_trans(clnt->trans_mod);
-free_client:
-	kfree(clnt);
+out:
+	p9_client_destroy(clnt);
 	return ERR_PTR(err);
 }
 EXPORT_SYMBOL(p9_client_create);
-----

I'll test and submit to Linus over the next few weeks.

I had a quick look at the other new syzbot warnings and:
 - 'possible deadlock in p9_req_put' is clear enough, we can just drop
the lock before running through the cancel list and I don't think
that'll cause any problem as everything has been moved to a local list
and that lock is abused by trans fd for its local stuff. I'll also send
that after quick testing.
----
>From c46435a4af7c119bd040922886ed2ea3a2a842d7 Mon Sep 17 00:00:00 2001
From: Dominique Martinet <asmadeus@...ewreck.org>
Date: Wed, 17 Aug 2022 14:58:44 +0900
Subject: [PATCH] 9p: trans_fd/p9_conn_cancel: drop client lock earlier

syzbot reported a double-lock here and we no longer need this
lock after requests have been moved off to local list:
just drop the lock earlier.

Reported-by: syzbot+50f7e8d06c3768dd97f3@...kaller.appspotmail.com
Signed-off-by: Dominique Martinet <asmadeus@...ewreck.org>

diff --git a/net/9p/trans_fd.c b/net/9p/trans_fd.c
index e758978b44be..60fcc6b30b46 100644
--- a/net/9p/trans_fd.c
+++ b/net/9p/trans_fd.c
@@ -205,6 +205,8 @@ static void p9_conn_cancel(struct p9_conn *m, int err)
 		list_move(&req->req_list, &cancel_list);
 	}
 
+	spin_unlock(&m->client->lock);
+
 	list_for_each_entry_safe(req, rtmp, &cancel_list, req_list) {
 		p9_debug(P9_DEBUG_ERROR, "call back req %p\n", req);
 		list_del(&req->req_list);
@@ -212,7 +214,6 @@ static void p9_conn_cancel(struct p9_conn *m, int err)
 			req->t_err = err;
 		p9_client_cb(m->client, req, REQ_STATUS_ERROR);
 	}
-	spin_unlock(&m->client->lock);
 }
 
 static __poll_t
----

 - but I don't get the two 'inconsistent lock state', the hint says it's
possibly an interrupt while the lock was held but that doesn't seem to
be the case from the stack trace (unless we leaked the lock, at which
point anything goes)
I'd need to take time to look at it, feel free to beat me to these.

--
Dominique