Re: [PATCH v12 1/2] cfg80211: add 'flags' to cfg80211_rx_mgmt()

[Johannes Berg <johannes@sipsolutions.net>]

On Tue, 2013-06-25 at 16:42 +0300, Vladimir Kondratiev wrote:

> +++ b/include/uapi/linux/nl80211.h
> @@ -1436,6 +1436,8 @@ enum nl80211_commands {
>   *	allowed to be used with the first @NL80211_CMD_SET_STATION command to
>   *	update a TDLS peer STA entry.
>   *
> + * @NL80211_ATTR_RXMGMT_FLAGS: flags for cfg80211_rx_mgmt, u32

That's a little insufficient, don't you think? cfg80211_rx_mgmt()
doesn't even really make sense to userspace, you should point to the
nl80211 event and maybe even reference this new attribute in the command
description.

> +/**
> + * enum nl80211_rxmgmt_flags - flags for cfg80211_rx_mgmt
> + *
> + * @NL80211_RXMGMT_FLAG_REPLIED: frame was replied by device/driver.
> + */
> +enum nl80211_rxmgmt_flags {
> +	NL80211_RXMGMT_FLAG_REPLIED = (1 << 0),

No need for parentheses.

>  bool cfg80211_rx_mgmt(struct wireless_dev *wdev, int freq, int sig_mbm,
> -		      const u8 *buf, size_t len, gfp_t gfp)
> +		      const u8 *buf, size_t len, u32 flags, gfp_t gfp)

I'm starting to wonder if this function has too many arguments?
Thoughts? Maybe it should get a struct or so? Not really sure ... could
just be a worthwhile cleanup afterwards?

johannes