Re: [PATCH RFC 1/2] brcmfmac: remove interface before notifying listener

From: Arend van Spriel
Date: Sat Jun 18 2016 - 15:26:59 EST


On 18-06-16 20:18, RafaÅ MiÅecki wrote:
> So far when receiving event about in-firmware-interface removal we were
> notifying our listener and afterwards we were removing Linux interface.
>

[snip]

>
> diff --git a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/fweh.c b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/fweh.c
> index 9da7a4c..5fd1886 100644
> --- a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/fweh.c
> +++ b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/fweh.c
> @@ -18,6 +18,7 @@
> #include "brcmu_wifi.h"
> #include "brcmu_utils.h"
>
> +#include "cfg80211.h"
> #include "core.h"
> #include "debug.h"
> #include "tracepoint.h"
> @@ -180,10 +181,16 @@ static void brcmf_fweh_handle_if_event(struct brcmf_pub *drvr,
> if (ifp && ifevent->action == BRCMF_E_IF_CHANGE)
> brcmf_fws_reset_interface(ifp);
>
> - err = brcmf_fweh_call_event_handler(ifp, emsg->event_code, emsg, data);

The reason for doing this first is because we are passing the ifp, which
is netdev_priv(ifp->ndev). In brcmf_remove_interface() we only
unregister the netdev, which will end up (after scheduling) in
brcmf_free_netdev() thus freeing the ifp. By moving the event handler
function ifp may be stale already.

> + if (ifp && ifevent->action == BRCMF_E_IF_DEL) {
> + bool rtnl_locked = brcmf_cfg80211_vif_event_armed(drvr->config);
> +
> + brcmf_remove_interface(ifp, rtnl_locked);

I guess rtnl_locked here means "rtnl_is_locked() by brcmfmac". It
actually does not matter who is holding the rtnl_lock. At least when it
is brcmfmac it is still a different task, ie. hostapd, iw, etc. Also
when brcmf_cfg80211_vif_event_armed() return false there may still be
some task holding the rtnl_lock.

Regards,
Arend