<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Tue, Apr 8, 2014 at 9:28 PM, Dan Williams <span dir="ltr"><<a href="mailto:dcbw@redhat.com" target="_blank">dcbw@redhat.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
<div class="">Note that mwifiex does have this information, it just doesn't expose it<br></div>
right now. It does receive an 'nf' with each packet, but apparently<br>
only chooses to use that information when processing management packets.<br>
It also simply calculates an RSSI from the NF and the SNR of the packet.<br></blockquote><div><br></div><div>Fair point.</div><div><br></div><div>My perspective, though, is that the Linux kernel API doesn't require drivers to</div>
<div>expose this information. So, concerns about mwifiex aside, we don't really have</div><div>a good reason to expect that all drivers will provide noise floor measurements.</div><div>(Note also that some drivers that normally provide measurements will sometimes</div>
<div>fail to do so.)</div><div><br></div><div>Further, wpa_supplicant already has the plumbing to keep track of whether or not</div><div>we received noise floor measurements. I'd like to use that plumbing to improve</div>
<div>BSS selection in cases where either a) the driver doesn't provide the measurements</div><div>at all, or b) the driver has some transient issue which prevents it from providing</div><div>some measurements.</div><div>
<br></div><div>How's that sound?</div><div><br></div></div></div></div>