Ticket #881 (closed enhancement: fixed)

Opened 5 years ago

Last modified 5 years ago

Get measured noise data from the HAL instead of assuming -95dBm.

Reported by: scottraynel@gmail.com Assigned to: mrenzmann
Priority: major Milestone: version 0.9.3
Component: madwifi: driver Version: trunk
Keywords: Cc:
Patch is attached: 1 Pending:

Description

The current HAL exports two functions which appear to have gone unnoticed for quite some time.

This patch makes use of those functions and gives measured noise data from the HAL to the driver and stack instead of simply assuming -95dBm throughout.

Currently the noise level is retrived on every rx interrupt and stored in the softc for the device. It is then passed to the 802.11 stack during the rx tasklet where it is used by the wireless extensions code. It is also used by the monitor mode functions to record the current noise level.

Because the descriptors only have an RSSI value in them and not a noise value, there is some error. The noise can not be retrived per frame, only per interrupt, and we cannot be sure if the RSSI was calculated using the noise level we just retrieved or some "older" value.

The other problem is scanning. Currently we do not keep a list of channels and their associated noise levels, we only keep a "current noise" value, which will end up being the last noise value retrieved. This means that all scan results will have the same noise level rather than individual levels for each channel. Presumably we could keep a list of channels, and then before we switch to the next channel to scan, get the noise and store it. The scan result list can then be made up using this information. For now however, scans simply report -95dBm as they always have.

In summary, this patch does the following:

  • Gets the measured noise level per rx interrupt from the HAL
  • Passes the noise level to wireless extensions (iwconfig, etc)
  • Use this level to calculate signal strength of frames in monitor mode.
  • Does not apply to scanning (working on it).

It would be nice to get some people testing this patch and providing feedback so that we could eventually have real noise information being exported by madwifi.

Attachments

madwifi_measured_noise.patch (9.4 kB) - added by scottraynel@gmail.com on 09/14/06 06:08:19.
Use measured noise levels instead of assuming -95 dBm

Change History

09/14/06 06:08:19 changed by scottraynel@gmail.com

  • attachment madwifi_measured_noise.patch added.

Use measured noise levels instead of assuming -95 dBm

09/14/06 06:12:11 changed by kelmo

Just for completeness sake, can you please Sign-off the patch? Thanks

09/14/06 06:14:27 changed by scottraynel@gmail.com

Signed-off-by: Scott Raynel <scottraynel@gmail.com>

09/14/06 21:03:29 changed by mentor

nice

09/15/06 03:32:09 changed by mrenzmann

  • status changed from new to assigned.
  • owner set to mrenzmann.
  • version set to trunk.
  • component changed from madwifi: other to madwifi: driver.
  • milestone set to version 0.9.3.

Very nice, thanks for that. The patch looks good. Unless there are any objections, we should commit it to trunk. There should be enough time for testing and eventually backing it out before the next release, since we currently have two blockers (the new HAL and a reported security issue in Ad-hoc mode).

Comments?

09/15/06 03:43:43 changed by kelmo

Done in r1713. Let the testing begin.

10/06/06 21:37:25 changed by mrenzmann

  • status changed from assigned to closed.
  • resolution set to fixed.

No reports so far, so let's close this ticket for now. If someone stumbles about problems in this regard, a new ticket should be opened, I propose.

10/20/06 14:19:54 changed by e. s.

Wow! Was searching exactly for that... Great!

In a diploma thesis red that the value is constant because the noise level of the card itself is given - the level of the noise the card produces. But actually that is wrong! You prove it!


Add/Change #881 (Get measured noise data from the HAL instead of assuming -95dBm.)