linux-kernel - [PATCH] 3.19-rc4: Fix race condition in i.MX serial port driver

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]

Message-ID: <trinity-2ff5fec6-196c-4ecc-9ec2-f1b223f0ea25-1421165228466@3capp-gmx-bs70>
Date:	Tue, 13 Jan 2015 17:07:08 +0100
From:	"Michael Doswald" <michael.doswald@....net>
To:	linux-serial@...r.kernel.org
Cc:	linux-kernel@...r.kernel.org
Subject: [PATCH] 3.19-rc4: Fix race condition in i.MX serial port driver

Under pressure, the imx.c device driver may transfer the same data multiple
times. 

Reason: 
Function imx_dma_tx checks for an active DMA transfer by calling
dmaengine_tx_status. The return value may indicate that no transfer is 
running while the dma_tx_callback function is still being executed. In 
this case the xmit->tail pointer may not be updated to the correct value 
when imx_dma_tx initiates a new DMA transfer, which will result in a DMA
transfer with the same base address as the previous one.

Patch:
Use existing dma_is_txing variable to guard DMA transfer initiation instead
of dmaengine_tx_status. Clear dma_is_txing at the end of the DMA callback
function therefore extending the critical section where the port lock is
held. Critical section extension is in line with the callback function 
of other drivers, e.g. serial-tegra.c (tegra_uart_tx_dma_complete).
 
 
Signed-off-by: Michael Doswald <michael.doswald@....net>
---
 drivers/tty/serial/imx.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
index 4c5e909..29e31b1 100644
--- a/drivers/tty/serial/imx.c
+++ b/drivers/tty/serial/imx.c
@@ -503,13 +503,10 @@ static void dma_tx_callback(void *data)
 
 	dma_unmap_sg(sport->port.dev, sgl, sport->dma_tx_nents, DMA_TO_DEVICE);
 
-	sport->dma_is_txing = 0;
-
 	/* update the stat */
 	spin_lock_irqsave(&sport->port.lock, flags);
 	xmit->tail = (xmit->tail + sport->tx_bytes) & (UART_XMIT_SIZE - 1);
 	sport->port.icount.tx += sport->tx_bytes;
-	spin_unlock_irqrestore(&sport->port.lock, flags);
 
 	dev_dbg(sport->port.dev, "we finish the TX DMA.\n");
 
@@ -518,8 +515,10 @@ static void dma_tx_callback(void *data)
 	if (waitqueue_active(&sport->dma_wait)) {
 		wake_up(&sport->dma_wait);
 		dev_dbg(sport->port.dev, "exit in %s.\n", __func__);
-		return;
 	}
+
+	sport->dma_is_txing = 0;
+	spin_unlock_irqrestore(&sport->port.lock, flags);
 }
 
 static void imx_dma_tx(struct imx_port *sport)
@@ -529,11 +528,9 @@ static void imx_dma_tx(struct imx_port *sport)
 	struct dma_async_tx_descriptor *desc;
 	struct dma_chan	*chan = sport->dma_chan_tx;
 	struct device *dev = sport->port.dev;
-	enum dma_status status;
 	int ret;
 
-	status = dmaengine_tx_status(chan, (dma_cookie_t)0, NULL);
-	if (DMA_IN_PROGRESS == status)
+	if (sport->dma_is_txing)
 		return;
 
 	sport->tx_bytes = uart_circ_chars_pending(xmit);
-- 
1.9.1


 
 
--
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/