[PATCH net-next] wireless: test sscanf return values

[PATCH net-next] wireless: test sscanf return values

[Joe Perches]

At some point, it'd be good to make sscanf use __must_check
so make sure the net/ uses of sscanf use the return value.

Signed-off-by: Joe Perches <joe@perches.com>
---

Compiled, untested.

 net/mac80211/debugfs.c | 4 ++--
 net/wireless/core.c    | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/net/mac80211/debugfs.c b/net/mac80211/debugfs.c
index eeb0bbd..0e44be7 100644
--- a/net/mac80211/debugfs.c
+++ b/net/mac80211/debugfs.c
@@ -160,9 +160,9 @@ static ssize_t sta_tx_latency_stat_write(struct file *file,
 	tx_latency->n_ranges = n_ranges;
 	for (i = 0; i < n_ranges; i++) { /* setting bin ranges */
 		token = strsep(&bins, TX_LATENCY_BIN_DELIMTER_S);
-		sscanf(token, "%d", &tx_latency->ranges[i]);
 		/* bins values should be in ascending order */
-		if (prev_bin >= tx_latency->ranges[i]) {
+		if (sscanf(token, "%d", &tx_latency->ranges[i]) != 1 ||
+		    prev_bin >= tx_latency->ranges[i]) {
 			ret = -EINVAL;
 			kfree(tx_latency);
 			goto unlock;
diff --git a/net/wireless/core.c b/net/wireless/core.c
index 3af0ecf..69a350e 100644
--- a/net/wireless/core.c
+++ b/net/wireless/core.c
@@ -95,8 +95,8 @@ static int cfg80211_dev_check_name(struct cfg80211_registered_device *rdev,
 	ASSERT_RTNL();
 
 	/* prohibit calling the thing phy%d when %d is not its number */
-	sscanf(newname, PHY_NAME "%d%n", &wiphy_idx, &taken);
-	if (taken == strlen(newname) && wiphy_idx != rdev->wiphy_idx) {
+	if (sscanf(newname, PHY_NAME "%d%n", &wiphy_idx, &taken) == 1 &&
+	    taken == strlen(newname) && wiphy_idx != rdev->wiphy_idx) {
 		/* count number of places needed to print wiphy_idx */
 		digits = 1;
 		while (wiphy_idx /= 10)

RE: [PATCH net-next] wireless: test sscanf return values

[David Laight]

From: Joe Perches
> At some point, it'd be good to make sscanf use __must_check
> so make sure the net/ uses of sscanf use the return value.

Isn't it much safer to avoid sscanf() completely and use
a different function for converting numerics?

	David

Re: [PATCH net-next] wireless: test sscanf return values

[Joe Perches]

On Fri, 2015-02-27 at 10:35 +0000, David Laight wrote:
> From: Joe Perches
> > At some point, it'd be good to make sscanf use __must_check
> > so make sure the net/ uses of sscanf use the return value.
> 
> Isn't it much safer to avoid sscanf() completely and use
> a different function for converting numerics?

It's generally better to use something other than sscanf.
That doesn't mean sscanf isn't useful.