On Mon, 2008-02-25 at 12:49 -0800, Luis Carlos Cobo wrote: > On Fri, 2008-02-22 at 17:21 +0100, Johannes Berg wrote: > > > + atomic_inc(&bss->users); > > > + atomic_inc(&bss->users); > > > > ? > > That was just allocated, shouldn't it just be atomic_set() then? And why > > two anyway? > > I was following the behavior in ieee80211_rx_bss_add(), where the same > thing is done. It is needed to to avoid the bss being freed on the next > ieee80211_rx_bss_put(), works as a reference count. Agree that > atomic_set() would be better. Ok, I didn't realise that it was the same in rx_bss_add(), I'm not too familiar with the code. > > > + memset(&iwe, 0, sizeof(iwe)); > > > + iwe.cmd = IWEVCUSTOM; > > > + sprintf(buf, "Mesh network (version %d)\n" > > > + "\t\t\tPath Selection Protocol ID: 0x%02X%02X%02X%02X\n" > > > + "\t\t\tPath Selection Metric ID: 0x%02X%02X%02X%02X\n" > > > + "\t\t\tCongestion Control Mode ID: 0x%02X%02X%02X%02X\n" > > > + "\t\t\tChannel Precedence: 0x%02X%02X%02X%02X", > > > > Maybe we should add separate items for this to make it easier to parse > > them? > > Ok, will include that too. Can you do things as follow-up patches that apply after the sta_info stuff I sent out today? I'd prefer to not respin the 11s and the sta_info patches. johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
