spi: cadence-xspi: use for_each_available_child_of_node_scoped()
authorKousik Sanagavarapu <five231003@gmail.com>
Wed, 10 Apr 2024 13:01:16 +0000 (18:31 +0530)
committerMark Brown <broonie@kernel.org>
Wed, 10 Apr 2024 13:49:26 +0000 (14:49 +0100)
Refactor code for "is the node's child available?" check by using the
corresponding macro instead, which reads more clearly.

While at it, use scope-based cleanup instead of manual of_node_put()
calls when getting platform data through cdns_xspi_of_get_plat_data().

This removes the unnecessary "node_child" declaration out of the loop's
scope and auto cleans up "node_child" when it goes out of scope, even
when we return early due to error.

Signed-off-by: Kousik Sanagavarapu <five231003@gmail.com>
Link: https://msgid.link/r/20240410130205.179069-1-five231003@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
drivers/spi/spi-cadence-xspi.c

index 8648b8eb080dc9a2f77c22459016c523611ffac3..2209e9fc378fa94998f0c2fc152d5d472263e146 100644 (file)
@@ -486,20 +486,14 @@ static irqreturn_t cdns_xspi_irq_handler(int this_irq, void *dev)
 static int cdns_xspi_of_get_plat_data(struct platform_device *pdev)
 {
        struct device_node *node_prop = pdev->dev.of_node;
-       struct device_node *node_child;
        unsigned int cs;
 
-       for_each_child_of_node(node_prop, node_child) {
-               if (!of_device_is_available(node_child))
-                       continue;
-
+       for_each_available_child_of_node_scoped(node_prop, node_child) {
                if (of_property_read_u32(node_child, "reg", &cs)) {
                        dev_err(&pdev->dev, "Couldn't get memory chip select\n");
-                       of_node_put(node_child);
                        return -ENXIO;
                } else if (cs >= CDNS_XSPI_MAX_BANKS) {
                        dev_err(&pdev->dev, "reg (cs) parameter value too large\n");
-                       of_node_put(node_child);
                        return -ENXIO;
                }
        }