Bug 12237 - Default WiFi power-saving mode 'Maximum' is unreliable and performs poorly
: Default WiFi power-saving mode 'Maximum' is unreliable and performs poorly
Status: UNCONFIRMED
Product: Maemo 5 Community SSU
general
: unspecified
: N900 Maemo
: Unspecified enhancement with 2 votes (vote)
: ---
Assigned To: unassigned
: wifi-bugs
:
:
:
:
  Show dependency tree
 
Reported: 2011-05-19 05:56 UTC by Ashley Hooper
Modified: 2012-02-17 15:56 UTC (History)
2 users (show)

See Also:


Attachments


Note

You need to log in before you can comment on or make changes to this bug.


Description Ashley Hooper (reporter) 2011-05-19 05:56:14 UTC
SOFTWARE VERSION: 
20.2010.36-2 (also CSSU)

EXACT STEPS LEADING TO PROBLEM: 
(Explain in detail what you do (e.g. tap on OK) and what you see (e.g. message
Connection Failed appears))
1. Connect to a WiFi network for the first time (or manually set power-saving
mode for an existing connection to Maximum).
2. Note that performance is unreliable, yet improves in speed and reliability
(pages hang during loading, and often time out) if a continuous ping is left
running (presumably, this inhibits power-saving)
3. Switch to Intermediate power saving and notice the performance improves
without any need of continuous ping

EXPECTED OUTCOME: Power-saving doesn't cause my http requests to time out

ACTUAL OUTCOME: Http requests time out frequently and pages load very slowly
with Maximum power-saving, unless I run a continuous ping

REPRODUCIBILITY: always

EXTRA SOFTWARE INSTALLED: 
Lots

OTHER COMMENTS: 
I propose either fixing the Maximum power-saving option so that it does not
come at a cost in reliability/speed, or changing the default power-saving for
new connections to 'Intermediate'
Comment 1 Joerg Reisenweber 2011-05-19 17:23:28 UTC
I suggest to have a comprehensive list of ALL (recent and future) wifi APs that
don't support proper powersaving, and the way they can get detected. We then
can implement automatic setting of PSM into CSSU. Oh no wa can't as switching
PSM is a bit tricky and can't get done per AP, as you need to power down and
restart the wifi chip for a changed PSM to take effect.

INVALID (if you haven't guessed yet)
Comment 2 Andre Klapper maemo.org 2011-05-19 17:29:41 UTC
According to Nokia the UI for Maemo5 is frozen, hence WONTFIX.
Feel free to move it to Maemo 5 Community Updates product though, in case the
codebase is open...
Comment 3 Ashley Hooper (reporter) 2011-05-19 17:38:00 UTC
Okay, I accept Nokia's position on this.

Joerg, I don't know if you use an N900, but this happens to me with most APs I
use. I proposed two options, and you chose to pick one line of attack for one
of the options I proposed, ignore the other option (saner defaults), and
consequently mark it invalid - why?
Comment 4 Ashley Hooper (reporter) 2011-05-19 17:44:41 UTC
Thanks, Andre
Comment 5 Joerg Reisenweber 2011-05-19 17:52:13 UTC
(In reply to comment #3)
> Okay, I accept Nokia's position on this.
> 
> Joerg, I don't know if you use an N900, but this happens to me with most APs I
> use. I proposed two options, and you chose to pick one line of attack for one
> of the options I proposed, ignore the other option (saner defaults), and
> consequently mark it invalid - why?

Sorry, no offense intended. It's just this is an age old known problem, usually
attributed to borked APs, WFM (and a lot of others: 
[2011-05-19 16:29:49] <ShadowJK> I recently bought a wifi ap brand A and usb
wireless adapter brand A. PSM didn't work :)
[2011-05-19 16:29:59] <ShadowJK> but it did work with N900
) and there's just so much you can do about it as the management of PSM is
inside wifi chip fw anyway, which we can't possibly fix (even if it were N900's
fault which it isn't.

No objections about setting "default" to PSM medium - it applies to first boot
only in the end.

NB PSM is a global setting, even though it's hidden at end of the AP/connection
specific ones.

cheers and sorry for inappropriate answer
/j
Comment 6 Ashley Hooper (reporter) 2011-05-19 18:05:25 UTC
Are you sure you're not confusing TxPwr with Power Saving mode?

N900's Internet Connections applet in Settings allows per-AP power-saving mode
selection (logically).

Whereas if you elect to change the TxPwr (10mW or 100mW), the Internet
Connections applet warns you this will apply globally (all connections), and
that an immediate power down and reinitialisation of the card will ensue.

If I'm missing something, please forgive me, but all outward signs are that
power-saving mode is per-AP.
Comment 7 Joerg Reisenweber 2011-05-19 21:25:37 UTC
(In reply to comment #6)
> Are you sure you're not confusing TxPwr with Power Saving mode?

Indeed, you're right.
/j
Comment 8 Joerg Reisenweber 2011-05-19 21:45:30 UTC
so a scheme to try associating to unknown AP with low/no PSM, then figure if
the AP can do proper PSM, by either

* look up in a list of MAC (ranges), and/or
* some interactive thing possibly exploiting some server to find if wifi stalls
on higher PSM,

and then setting the config for this AP accordingly, this might actually make
some sense.

Icing on top would be a way to share the config for MAC ranges generated
locally by heuristics / probing / manual setting, and then confirmed by user,
to some central database.

Probably such a thing could get implemented as an installable pkg not needing
any patches to the core system (i.e. no need for cssu to deal with it).

/j
Comment 9 Ashley Hooper (reporter) 2011-05-20 04:21:02 UTC
Yes, I just thought that having Maximum power saving the default is quite
un-user-friendly, given how prevalent buggy APs are.

Even just changing the default would be a huge win, IMHO. It's one less thing I
have to do each time I use a new AP (and going through all those dialogs to
reach power management is a royal pain).

If further dev resources were available, a small (even CLI, with launcher) app
could bring down the interface and re-up it with maximum power-saving, run a
bunch of tests against the current AP and make a recommendation accordingly.

Or we could go further and implement what you have suggested.