From: Hante Meuleman Date: Fri, 18 Sep 2015 22:08:11 +0200 Subject: [PATCH] brcmfmac: Workaround in change vif for wpa_supplicant support. Different wpa_supplicants have different behavior and expectations regarding the change_virtual_intf behavior. This patch implements a workaround for the different versions and possible brcmfmac configuration. Reviewed-by: Arend Van Spriel Reviewed-by: Pieter-Paul Giesberts Signed-off-by: Hante Meuleman Signed-off-by: Arend van Spriel Signed-off-by: Kalle Valo --- --- a/drivers/net/wireless/brcm80211/brcmfmac/cfg80211.c +++ b/drivers/net/wireless/brcm80211/brcmfmac/cfg80211.c @@ -776,6 +776,37 @@ brcmf_cfg80211_change_iface(struct wiphy s32 err = 0; brcmf_dbg(TRACE, "Enter, idx=%d, type=%d\n", ifp->bssidx, type); + + /* WAR: There are a number of p2p interface related problems which + * need to be handled initially (before doing the validate). + * wpa_supplicant tends to do iface changes on p2p device/client/go + * which are not always possible/allowed. However we need to return + * OK otherwise the wpa_supplicant wont start. The situation differs + * on configuration and setup (p2pon=1 module param). The first check + * is to see if the request is a change to station for p2p iface. + */ + if ((type == NL80211_IFTYPE_STATION) && + ((vif->wdev.iftype == NL80211_IFTYPE_P2P_CLIENT) || + (vif->wdev.iftype == NL80211_IFTYPE_P2P_GO) || + (vif->wdev.iftype == NL80211_IFTYPE_P2P_DEVICE))) { + brcmf_dbg(TRACE, "Ignoring cmd for p2p if\n"); + /* Now depending on whether module param p2pon=1 was used the + * response needs to be either 0 or EOPNOTSUPP. The reason is + * that if p2pon=1 is used, but a newer supplicant is used then + * we should return an error, as this combination wont work. + * In other situations 0 is returned and supplicant will start + * normally. It will give a trace in cfg80211, but it is the + * only way to get it working. Unfortunately this will result + * in situation where we wont support new supplicant in + * combination with module param p2pon=1, but that is the way + * it is. If the user tries this then unloading of driver might + * fail/lock. + */ + if (cfg->p2p.p2pdev_dynamically) + return -EOPNOTSUPP; + else + return 0; + } err = brcmf_vif_change_validate(wiphy_to_cfg(wiphy), vif, type); if (err) { brcmf_err("iface validation failed: err=%d\n", err); @@ -791,18 +822,6 @@ brcmf_cfg80211_change_iface(struct wiphy infra = 0; break; case NL80211_IFTYPE_STATION: - /* Ignore change for p2p IF. Unclear why supplicant does this */ - if ((vif->wdev.iftype == NL80211_IFTYPE_P2P_CLIENT) || - (vif->wdev.iftype == NL80211_IFTYPE_P2P_GO)) { - brcmf_dbg(TRACE, "Ignoring cmd for p2p if\n"); - /* WAR: It is unexpected to get a change of VIF for P2P - * IF, but it happens. The request can not be handled - * but returning EPERM causes a crash. Returning 0 - * without setting ieee80211_ptr->iftype causes trace - * (WARN_ON) but it works with wpa_supplicant - */ - return 0; - } infra = 1; break; case NL80211_IFTYPE_AP: