linux-kernel - [PATCH tty v1 5/8] serial: 8250: lock port for rx

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: <20230525093159.223817-6-john.ogness@linutronix.de>
Date:   Thu, 25 May 2023 11:37:56 +0206
From:   John Ogness <john.ogness@...utronix.de>
To:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc:     Petr Mladek <pmladek@...e.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        linux-kernel@...r.kernel.org, Jiri Slaby <jirislaby@...nel.org>,
        Ilpo Järvinen <ilpo.jarvinen@...ux.intel.com>,
        Tony Lindgren <tony@...mide.com>,
        Lukas Wunner <lukas@...ner.de>,
        Matthias Schiffer <matthias.schiffer@...tq-group.com>,
        linux-serial@...r.kernel.org
Subject: [PATCH tty v1 5/8] serial: 8250: lock port for rx_dma() callback

The rx_dma() callback (omap_8250_rx_dma) accesses UART_IER. This
register is modified twice by each console write
(serial8250_console_write()) under the port lock. However, not
all calls to the rx_dma() callback are under the port lock.

Add the missing port locking around rx_dma() callback calls. Add
lockdep notation to the omap_8250_rx_dma().

Note that this is not fixing a real problem because:

1. Currently DMA is forced off for 8250_omap consoles.

2. The serial devices are resumed before console printing is enabled.

However, adding this locking allows for clean locking semantics
for the rx_dma() callback so that lockdep can be used to identify
possible problems in the future. It also simplifies synchronization
of UART_IER in general by not needing to rely on implementation
details such as 1 and 2.

Signed-off-by: John Ogness <john.ogness@...utronix.de>
---
 drivers/tty/serial/8250/8250_omap.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/drivers/tty/serial/8250/8250_omap.c b/drivers/tty/serial/8250/8250_omap.c
index c17d98161d5e..3cb9cfa62331 100644
--- a/drivers/tty/serial/8250/8250_omap.c
+++ b/drivers/tty/serial/8250/8250_omap.c
@@ -728,8 +728,11 @@ static int omap_8250_startup(struct uart_port *port)
 		priv->wer |= OMAP_UART_TX_WAKEUP_EN;
 	serial_out(up, UART_OMAP_WER, priv->wer);
 
-	if (up->dma && !(priv->habit & UART_HAS_EFR2))
+	if (up->dma && !(priv->habit & UART_HAS_EFR2)) {
+		spin_lock_irq(&port->lock);
 		up->dma->rx_dma(up);
+		spin_unlock_irq(&port->lock);
+	}
 
 	enable_irq(up->port.irq);
 
@@ -1003,6 +1006,9 @@ static int omap_8250_rx_dma(struct uart_8250_port *p)
 	unsigned long			flags;
 	u32				reg;
 
+	/* Port locked to synchronize UART_IER access against the console. */
+	lockdep_assert_held_once(&p->port.lock);
+
 	if (priv->rx_dma_broken)
 		return -EINVAL;
 
@@ -1736,8 +1742,11 @@ static int omap8250_runtime_resume(struct device *dev)
 	if (up && omap8250_lost_context(up))
 		omap8250_restore_regs(up);
 
-	if (up && up->dma && up->dma->rxchan && !(priv->habit & UART_HAS_EFR2))
+	if (up && up->dma && up->dma->rxchan && !(priv->habit & UART_HAS_EFR2)) {
+		spin_lock_irq(&up->port.lock);
 		omap_8250_rx_dma(up);
+		spin_unlock_irq(&up->port.lock);
+	}
 
 	priv->latency = priv->calc_latency;
 	schedule_work(&priv->qos_work);
-- 
2.30.2