linux-kernel - [PATCH] hvc: ensure hvc_init is only ever called once in hvc

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-next>] [day] [month] [year] [list]

Date:	Tue, 14 Jan 2014 16:03:37 -0500
From:	Paul Gortmaker <paul.gortmaker@...driver.com>
To:	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Jiri Slaby <jslaby@...e.cz>
CC:	<linux-kernel@...r.kernel.org>,
	Paul Gortmaker <paul.gortmaker@...driver.com>,
	<stable@...r.kernel.org>, Rusty Russell <rusty@...tcorp.com.au>
Subject: [PATCH] hvc: ensure hvc_init is only ever called once in hvc_console.c

Commit 3e6c6f630a5282df8f3393a59f10eb9c56536d23 ("Delay creation of
khcvd thread") moved the call of hvc_init from being a device_initcall
into hvc_alloc, and used a non-null hvc_driver as indication of whether
hvc_init had already been called.

The problem with this is that hvc_driver is only assigned a value
at the bottom of hvc_init, and so there is a window where multiple
hvc_alloc calls can be in progress at the same time and hence try
and call hvc_init multiple times.  Previously the use of device_init
guaranteed that hvc_init was only called once.

This manifests itself as sporadic instances of two hvc_init calls
racing each other, and with the loser of the race getting -EBUSY
from tty_register_driver() and hence that virtual console fails:

    Couldn't register hvc console driver
    virtio-ports vport0p1: error -16 allocating hvc for port

Here we add an atomic_t to guarantee we'll never run hvc_init twice.

Cc: stable@...r.kernel.org  # v2.6.24+
Cc: Rusty Russell <rusty@...tcorp.com.au>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Fixes: 3e6c6f630a52 ("Delay creation of khcvd thread")
Reported-by: Jim Somerville <Jim.Somerville@...driver.com>
Tested-by: Jim Somerville <Jim.Somerville@...driver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@...driver.com>
---

[Greg: I realize tty-next is closed; given that this is an ancient
 bug, I don't see any real urgency here, so feel free to queue it
 whenever/whereever is most convenient - assuming it is an OK fix.]

 drivers/tty/hvc/hvc_console.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/tty/hvc/hvc_console.c b/drivers/tty/hvc/hvc_console.c
index 9eba119bcdd3..620da5f88e91 100644
--- a/drivers/tty/hvc/hvc_console.c
+++ b/drivers/tty/hvc/hvc_console.c
@@ -31,6 +31,7 @@
 #include <linux/list.h>
 #include <linux/module.h>
 #include <linux/major.h>
+#include <linux/atomic.h>
 #include <linux/sysrq.h>
 #include <linux/tty.h>
 #include <linux/tty_flip.h>
@@ -70,6 +71,9 @@ static struct task_struct *hvc_task;
 /* Picks up late kicks after list walk but before schedule() */
 static int hvc_kicked;
 
+/* hvc_init is triggered from hvc_alloc, i.e. only when actually used */
+static atomic_t hvc_needs_init __read_mostly = ATOMIC_INIT(-1);
+
 static int hvc_init(void);
 
 #ifdef CONFIG_MAGIC_SYSRQ
@@ -851,7 +855,7 @@ struct hvc_struct *hvc_alloc(uint32_t vtermno, int data,
 	int i;
 
 	/* We wait until a driver actually comes along */
-	if (!hvc_driver) {
+	if (atomic_inc_not_zero(&hvc_needs_init)) {
 		int err = hvc_init();
 		if (err)
 			return ERR_PTR(err);
-- 
1.8.5.2

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/