spi: imx: Fix failure path leak on GPIO request error
authorTrent Piepho <tpiepho@impinj.com>
Fri, 27 Oct 2017 01:08:39 +0000 (18:08 -0700)
committerMark Brown <broonie@kernel.org>
Tue, 31 Oct 2017 11:20:23 +0000 (11:20 +0000)
If the code that requests any chip select GPIOs fails, the cleanup of
spi_bitbang_start() by calling spi_bitbang_stop() is not done.

Fix this by moving spi_bitbang_start() to after the code that requets
GPIOs.  The GPIOs are dev managed and don't need explicit cleanup.
Since spi_bitbang_start() is now the last operation, it doesn't need
to be cleaned up in the failure path.

CC: Shawn Guo <shawnguo@kernel.org>
CC: Sascha Hauer <kernel@pengutronix.de>
CC: Fabio Estevam <fabio.estevam@nxp.com>
CC: Mark Brown <broonie@kernel.org>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Trent Piepho <tpiepho@impinj.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
drivers/spi/spi-imx.c

index fe35aaea323b732fd2715e273dfd688eeac29b2f..5ddd32ba2521b2fd2bbf1aa2fd148bfd544657d6 100644 (file)
@@ -1613,11 +1613,6 @@ static int spi_imx_probe(struct platform_device *pdev)
        spi_imx->devtype_data->intctrl(spi_imx, 0);
 
        master->dev.of_node = pdev->dev.of_node;
-       ret = spi_bitbang_start(&spi_imx->bitbang);
-       if (ret) {
-               dev_err(&pdev->dev, "bitbang start failed with %d\n", ret);
-               goto out_clk_put;
-       }
 
        if (!spi_imx->slave_mode) {
                if (!master->cs_gpios) {
@@ -1641,6 +1636,12 @@ static int spi_imx_probe(struct platform_device *pdev)
                }
        }
 
+       ret = spi_bitbang_start(&spi_imx->bitbang);
+       if (ret) {
+               dev_err(&pdev->dev, "bitbang start failed with %d\n", ret);
+               goto out_clk_put;
+       }
+
        dev_info(&pdev->dev, "probed\n");
 
        clk_disable(spi_imx->clk_ipg);