staging: rtl8192u: Replace GFP_ATOMIC with GFP_KERNEL in ieee80211_ccmp_init

ieee80211_ccmp_init() is never called in atomic context.

ieee80211_ccmp_init() is only set as ".init" in
struct ieee80211_crypto_ops.
The call chains ending up at "->init" function are:
[1] ->init() <- ieee80211_wpa_set_encryption() <-
	ieee80211_wpa_supplicant_ioctl()
[2] ->init() <- ieee80211_wx_set_encode_ext_rsl() <-
	r8192_wx_set_enc_ext()
[3] ->init() <- ieee80211_wx_set_encode_rsl() <-
	r8192_wx_set_enc()

ieee80211_wpa_supplicant_ioctl(), r8192_wx_set_enc_ext() and
r8192_wx_set_enc() call mutex_lock(), which indicates these functions
are not called in atomic context.

Despite never getting called from atomic context,
ieee80211_ccmp_init() calls kzalloc() with GFP_ATOMIC,
which does not sleep for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
which can sleep and improve the possibility of sucessful allocation.

This is found by a static analysis tool named DCNS written by myself.
And I also manually check it.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/staging/rtl8192u/ieee80211/ieee80211_crypt_ccmp.c

@@ -66,7 +66,7 @@ static void *ieee80211_ccmp_init(int key_idx)
 {
 	struct ieee80211_ccmp_data *priv;
 
-	priv = kzalloc(sizeof(*priv), GFP_ATOMIC);
+	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
 	if (!priv)
 		goto fail;
 	priv->key_idx = key_idx;