[PATCH] cpm_uart: Initialize port.dev before it's used.

[Scott Wood <scottwood@freescale.com>]

Previously, this caused NULL to sometimes be passed as a device
to the DMA code.  With recent DMA changes, that now causes a BUG().

Signed-off-by: Scott Wood <scottwood@freescale.com>
---
This is a regression triggered by DMA changes; please apply for 2.6.30.

 drivers/serial/cpm_uart/cpm_uart_core.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/serial/cpm_uart/cpm_uart_core.c b/drivers/serial/cpm_uart/cpm_uart_core.c
index 5c6ef51..4dd00be 100644
--- a/drivers/serial/cpm_uart/cpm_uart_core.c
+++ b/drivers/serial/cpm_uart/cpm_uart_core.c
@@ -1339,13 +1339,13 @@ static int __devinit cpm_uart_probe(struct of_device *ofdev,
 
 	dev_set_drvdata(&ofdev->dev, pinfo);
 
+	/* initialize the device pointer for the port */
+	pinfo->port.dev = &ofdev->dev;
+
 	ret = cpm_uart_init_port(ofdev->node, pinfo);
 	if (ret)
 		return ret;
 
-	/* initialize the device pointer for the port */
-	pinfo->port.dev = &ofdev->dev;
-
 	return uart_add_one_port(&cpm_reg, &pinfo->port);
 }
 
-- 
1.5.6.rc1.6.gc53ad.dirty