Skip to contentSkip to conversation

pinctrl: pinconf-generic: More fwnode related conversions

PATCH In review Andy Shevchenko 2 participants 5 messages Mar 17 – Mar 19

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover pinctrl: pinconf-generic: More fwnode related conversions

While I have noticed some inconsistencies in the pinconf-generic code, I also found one minor issue and a room to convert the code to use fwnode APIs instead of being too OF-centric. This might help in the future to use similar data structures in software nodes.

Andy Shevchenko (3):
pinctrl: pinconf-generic: Fully validate 'pinmux' property
pinctrl: pinconf-generic: Validate fwnode instead of device node
pinctrl: pinconf-generic: Convert ..._parse_dt_pinmux() to fwnode API

 drivers/pinctrl/pinconf-generic.c | 37 ++++++++++++++++++-------------
 1 file changed, 22 insertions(+), 15 deletions(-)
Signature · 1 line
2.50.1
#1 lore raw

PATCH 1/3 pinctrl: pinconf-generic: Fully validate 'pinmux' property

The pinconf_generic_parse_dt_pinmux() assumes that the 'pinmux' property is not empty when present. This might be not true. With that, the allocator will give a special value in return and not NULL which lead to the crash when trying to access that (invalid) memory. Fix that by fully validating 'pinmux' value, including its length.

Fixes 7112c05fff83 ("pinctrl: pinconf-generic: Add API for pinmux propertity in DTS file")
---
 drivers/pinctrl/pinconf-generic.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/pinctrl/pinconf-generic.c +6 -1
index 61b5b3fb94ce..d0b825ff52db 100644
@@ -325,12 +325,17 @@ int pinconf_generic_parse_dt_pinmux(struct device_node *np, struct device *dev,
 		return -ENOENT;
 	}
 
+	npins_t = prop->length / sizeof(u32);
+	if (npins_t == 0) {
+		dev_info(dev, "pinmux property doesn't have entries\n");
+		return -ENODATA;
+	}
+
 	if (!pid || !pmux || !npins) {
 		dev_err(dev, "parameters error\n");
 		return -EINVAL;
 	}
 
-	npins_t = prop->length / sizeof(u32);
 	pid_t = devm_kcalloc(dev, npins_t, sizeof(*pid_t), GFP_KERNEL);
 	pmux_t = devm_kcalloc(dev, npins_t, sizeof(*pmux_t), GFP_KERNEL);
 	if (!pid_t || !pmux_t) {
Signature · 1 line
2.50.1
#2 lore raw

PATCH 2/3 pinctrl: pinconf-generic: Validate fwnode instead of device node

In reply to Andy Shevchenko

Currently we convert device node to fwnode in the pinconf_generic_parse_dt_config() and then validate the device node. This is confusing order. Instead, assign fwnode and validate it.

Fixes e002d162654b ("pinctrl: pinconf-generic: Use only fwnode API in parse_dt_cfg()")
---
 drivers/pinctrl/pinconf-generic.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/pinctrl/pinconf-generic.c +3 -2
index d0b825ff52db..0ed57720ae61 100644
@@ -377,12 +377,13 @@ int pinconf_generic_parse_dt_config(struct device_node *np,
 				    unsigned long **configs,
 				    unsigned int *nconfigs)
 {
-	struct fwnode_handle *fwnode = of_fwnode_handle(np);
 	unsigned long *cfg;
 	unsigned int max_cfg, ncfg = 0;
+	struct fwnode_handle *fwnode;
 	int ret;
 
-	if (!np)
+	fwnode = of_fwnode_handle(np);
+	if (!fwnode)
 		return -EINVAL;
 
 	/* allocate a temporary array big enough to hold one of each option */
Signature · 1 line
2.50.1
#3 lore raw

PATCH 3/3 pinctrl: pinconf-generic: Convert ..._parse_dt_pinmux() to fwnode API

In reply to Andy Shevchenko

Convert pinconf_generic_parse_dt_pinmux() to fwnode API. This makes code cleaner and potentially reusable for some other types of fwnodes, such as software nodes.

---
 drivers/pinctrl/pinconf-generic.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)
drivers/pinctrl/pinconf-generic.c +14 -13
index 0ed57720ae61..88eacd1aa2de 100644
@@ -312,20 +312,19 @@ int pinconf_generic_parse_dt_pinmux(struct device_node *np, struct device *dev,
 				    unsigned int **pid, unsigned int **pmux,
 				    unsigned int *npins)
 {
+	struct fwnode_handle *fwnode = of_fwnode_handle(np);
 	unsigned int *pid_t;
 	unsigned int *pmux_t;
-	struct property *prop;
 	unsigned int npins_t, i;
-	u32 value;
 	int ret;
 
-	prop = of_find_property(np, "pinmux", NULL);
-	if (!prop) {
+	ret = fwnode_property_count_u32(fwnode, "pinmux");
+	if (ret < 0) {
 		dev_info(dev, "Missing pinmux property\n");
-		return -ENOENT;
+		return ret;
 	}
 
-	npins_t = prop->length / sizeof(u32);
+	npins_t = ret;
 	if (npins_t == 0) {
 		dev_info(dev, "pinmux property doesn't have entries\n");
 		return -ENODATA;
@@ -342,14 +341,16 @@ int pinconf_generic_parse_dt_pinmux(struct device_node *np, struct device *dev,
 		dev_err(dev, "kalloc memory fail\n");
 		return -ENOMEM;
 	}
+
+	ret = fwnode_property_read_u32_array(fwnode, "pinmux", pmux_t, npins_t);
+	if (ret) {
+		dev_err(dev, "get pinmux value fail\n");
+		goto exit;
+	}
+
 	for (i = 0; i < npins_t; i++) {
-		ret = of_property_read_u32_index(np, "pinmux", i, &value);
-		if (ret) {
-			dev_err(dev, "get pinmux value fail\n");
-			goto exit;
-		}
-		pmux_t[i] = value & 0xff;
-		pid_t[i] = (value >> 8) & 0xffffff;
+		pid_t[i] = pmux_t[i] >> 8;
+		pmux_t[i] = pmux_t[i] & 0xff;
 	}
 	*pid = pid_t;
 	*pmux = pmux_t;
Signature · 1 line
2.50.1
maintainer
#4 lore raw
In reply to Andy Shevchenko
Quoted 9 lines · Andy Shevchenko

While I have noticed some inconsistencies in the pinconf-generic code, I also found one minor issue and a room to convert the code to use fwnode APIs instead of being too OF-centric. This might help in the future to use similar data structures in software nodes.

Andy Shevchenko (3):
pinctrl: pinconf-generic: Fully validate 'pinmux' property
pinctrl: pinconf-generic: Validate fwnode instead of device node
pinctrl: pinconf-generic: Convert ..._parse_dt_pinmux() to fwnode API

Patches applied!

Yours,
Linus Walleij