[Tinyos Core WG] Meeting: 7/11/07

[Martin Turon]

T2 Core Notes 2007/7/11

David Gay, Phil Levis, Ben G., Martin Turon, Vlado, Fred Chang

PL: First item status updates

DG: Tracking down Mica2 log bug (AT45)

PL: It would be useful to extend the unit tests to catch that in the
future.

DG: Already in test/storage.  Need to check in

BG: Serial stack - checking for overruns.

PL: Back ground when operating at 115.2k baud Telos drops many packets
	Theory is interrupt latency is the cause.  What if that is not
the 
	root cause?
	
BG: If it is something else, it would be harder to design solution.
	Can't imagine what else it would be.
	
PL: Large atomic sections perhaps?

V: MSP430 single phase printf debugging library.  
Wanted to use UARTByte for that, but it is implemented as a busy 
loop with an unprotected flag from other async event users.
Some tests of printf lib from Kevin switched to different UART.
Found bug in MSP430 timer section that allows tasks to run at 4.00 MHz.
Setting target from 4MHz binary to 4.096MHz caused error.  Tmote sky
has external resistor that changed temperature stability of DCO.
Factored out DCO code before simple calibration.  Introduced new
components
in TelosB that are commented out, but would remove the code before 
calibration.  

What is target for telosb platforms?  Is it designed to handle all
telosb
clones?  Or can tmote sky platform be separate for handling of these
external
resistors?

PL: Tmote would be a separate platform that inherits telosb with minor 
code difference.  Make a separate platform after release.

V: Pretty stable without the flag.  Looked on scope under hot air.  
Not that bad.

PL: Net2 is moving forward RE: physical layer information.  White link
bit
has been implemented for cc2420.  Code is on a branch.  Keeping CT code
as now, while exploring new interfaces to make this happen.

PL: Next on agenda is testing.  We have test results for eyes, micaz,
and 
mica2.  Ping joe/jonathon for telosb. 

DG: mica2 test status - logging bug isolated.  Weird behavior in cc1000
LPL
stack.  Easy fix is to revert to previous stack before David Moss' 
optimized power changes.

PL: 2.0.1 user reported failure after 2 days.

DG: This fix solves failure after 5 seconds.  Not sure about 2 days.

PL: Could be Anti-theft bug or CPT bug

DG: Anti-theft doesn't do much - no arrays or anything

PL: Core testing vs. net2 testing.  Two applications may not have 
every platform.  Tough to test multi-hop on an eyes node for example.

DG: Ideally most work is at application independent level.  You can't 
test every combination, need to leverage HIL abstractions and test
those.

PL: Stack worked fine with collection and dissemination, but putting
them 
together didn't work.  Propose that first time we get tests, test them
cross-platform, then move to working group level.  First Deluge test is
on all platforms to do sanity check beyond authors.  

PL: Other testing issues?  All tests are due tomorrow.  
Next item is TEP status.  TEP 1 finalized.  

DG: TEP 102 ready to send back to Steve.  Sent on July 2.  Phil to ping
him.

PL: Next milestone all TEPs below 120 finalized after 2.0.2 release.
The
list of TEPs is in the agenda with current status.  TEP 104 Alan W. sent
comments for Gil to respond to.  One issue is 5 TEPs have been sent to
the
steering committee that have yet to have shephards.  Our milestone may
be 
difficult to meet if review process doesn't move forward.  Should we
push
harder, or push the milestone.  

V: Open the circle of who can be a shephard.  Steering commitee only
limits
pool of volunteers.

PL: Steering comittee names them, but they can come from anywhere.

V: Proactively search for shephards on tiny-devel and propose them 
to steering committee. 

PL: Shephards where proposed, but steering committee never got back to
them.
So we can also take existing names and interact with them directly.  TEP
109 
has been blocking on this for 6 months.

PL: Last item: once 2.0.2 is out, 2.1 is next step.  Some discussion on 
definition of 2.1.  3 things: 
1) changes to packet interfaces for better checking and different wiring
2) use of nesC 1.2.9 for standard BitUtils and newer compilers
3) Remove #ifdef and improve structure
Others?

DG: working cc1000 LPL stack.  DMA perhaps.

V: Sensor boards.  Meta data, radio control.

PL: Radio X cannot support this functionality.  We need a statement for
what 
to do in this situation.  

V: Whatever net2 needs if there are substantial changes to HAL/HIL, it
would 
be duty of core to handle that.

MT: Some thoughts: how well is power management working on ARM
platforms?

DG: Phil B. is the person to ask

MT: How easily could threads be added for advanced platforms such as
ARM?

DG: Current code wouldn't work with a thread model.

PL: Threads will appear eventually, so lets think through how they would

appear in T2.

DG: The current code model will have issues if threads are underneath.

PL: Not conviced of that.  Not changing all code would be good.

DG: Arguably you should reimplement system abstractions in that case.

FC: Thinking about how to do energy management with threads.   

PL: USC is interested in threads.  They want to load code dynamically.  
Threads simplifies that.  Down calls only makes dynamic linking much
easier.

MT: What are the changed assumptions for threads?

DG: Tasks assume they are never interrupted -- they don't protect busy
flags with
atomic for example.

V: Question of religion, perhaps we should postpone discussion until we
are ready
to address it in detail.  Porting all the code to threading will be a
nightmare.

PL: Other agenda items for 2.1?

V: Simplify and automate the release process.

PL: Lib printf in all the tests that prints SUCCESS/FAIl.  Ok.

DG: Fixed the random storage bug.

PL: Other items for which we should write TEPs?  

DG: Platform description ones.

V: Assumptions.  Reserved timers.

PL: Arch Rock said they would like to see a "how to create a platform"
TEP.

V: On vacation next 4 meetings.  Talk to you in August.

PL: Things slowing down.  Perhaps meeting every week not needed.

DG: Sure, until we start having magic thread discussions... :)

Bye everybody!

Martin 
__________________________________________________
Martin Turon  |  Crossbow Technology, Inc.

-----Original Message-----
From: tinyos-2.0wg-bounces at Mail.Millennium.Berkeley.EDU
[mailto:tinyos-2.0wg-bounces at Mail.Millennium.Berkeley.EDU] On Behalf Of
David Gay
Sent: Wednesday, July 11, 2007 8:59 AM
To: Philip Levis
Cc: TinyOS Core WG
Subject: Re: [Tinyos Core WG] Meeting: 7/11/07

Wednesday, July 11, 2007, 09:30 AM US Pacific Time
Bridge: 2
Passcode: 9052418

David
_______________________________________________
Tinyos-2.0wg mailing list
Tinyos-2.0wg at Mail.Millennium.Berkeley.EDU
https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-2.0
wg