Newer
Older
From: =?UTF-8?q?Rafa=C5=82=20Mi=C5=82ecki?= <zajec5@gmail.com>
Date: Thu, 9 Jul 2015 00:07:59 +0200
Subject: [PATCH] brcmfmac: workaround bug with some inconsistent BSSes state
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
---
--- a/drivers/net/wireless/brcm80211/brcmfmac/cfg80211.c
+++ b/drivers/net/wireless/brcm80211/brcmfmac/cfg80211.c
@@ -662,8 +662,36 @@ static struct wireless_dev *brcmf_cfg802
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
u32 *flags,
struct vif_params *params)
{
+ struct net_device *dev;
struct wireless_dev *wdev;
+ /*
+ * There is a bug with in-firmware BSS management. When adding virtual
+ * interface brcmfmac first tells firmware to create new BSS and then
+ * it creates new struct net_device.
+ *
+ * If creating/registering netdev(ice) fails, BSS remains in some bugged
+ * state. It conflicts with existing BSSes by overtaking their auth
+ * requests.
+ *
+ * It results in one BSS (addresss X) sending beacons and another BSS
+ * (address Y) replying to authentication requests. This makes interface
+ * unusable as AP.
+ *
+ * To workaround this bug we may try to guess if register_netdev(ice)
+ * will fail. The most obvious case is using interface name that already
+ * exists. This is actually quite likely with brcmfmac & some user space
+ * scripts as brcmfmac doesn't allow deleting virtual interfaces.
+ * So this bug can be triggered even by something trivial like:
+ * iw dev wlan0 delete
+ * iw phy phy0 interface add wlan0 type __ap
+ */
+ dev = dev_get_by_name(&init_net, name);
+ if (dev) {
+ dev_put(dev);
+ return ERR_PTR(-EEXIST);
+ }
+
brcmf_dbg(TRACE, "enter: %s type %d\n", name, type);
switch (type) {
case NL80211_IFTYPE_ADHOC: