> +/* plink is short for peer link */
> +struct mesh_plink {
Looking at this structure in more detail, a few more comments.
If you group "retries" and "ignore_timer" the structure will be four
bytes smaller (each of them uses one byte and is padded to four in the
current struct layout).
> + struct ieee80211_channel channel;
> + u32 channel_precedence;
These don't seem to be used? I was about to suggest that channel should
be a pointer instead of a copy of the data, especially since the
original data can be updated by the regulatory control code.
> +#define plink_inc(s) atomic_inc(&s->u.sta.mshstats.estab_plinks)
> +#define plink_dec(s) atomic_dec(&s->u.sta.mshstats.estab_plinks)
> +#define plink_capacity(s) (s->u.sta.mshcfg.dot11MeshMaxPeerLinks - \
> + atomic_read(&s->u.sta.mshstats.estab_plinks))
> +#define available_plinks(s) (min(plink_capacity(s), MESH_MAX_PLINKS - \
> + atomic_read(&s->u.sta.plinks)) > 0)
> +
> +#define activate_path(mpath) (mpath->flags |= \
> + (MESH_PATH_ACTIVE | MESH_PATH_RESOLVED))
Those should probably all be static inlines.
johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
