Just wondering if you are considering to switch to vanilla kernel at some point.

The app on the sam4l shouldn't do any manipulation so that we can easily extend the python code.

The signpost/software/edison/toFlash.tar.gz image seems to not work for me.

I installed git lfs:

sudo port install git-lfs
git clone https://github.com/lab11/signpost.git
cd signpost/software/edison                                                                                    
git lfs track toFlash.tar.gz                                                                                    
git lfs pull toFlash.tar.gz  

The image size seems correct and the file seems OK:

bash-3.2$ ls -l toFlash.tar.gz                                                                                            
-rw-r--r--  1 cxh  staff  158338761 Feb 15 11:39 toFlash.tar.gz                                                           
bash-3.2$ tar -ztvf *.gz | head                                                                                           
drwxrwxrwx  0 root   root        0 Nov 22 01:15 toFlash/                                                                  
-rw-rw-rw-  0 root   root  4194304 Nov 22 01:09 toFlash/edison_ifwi-dbg-01-dfu.bin                                        
-rw-rw-rw-  0 root   root  4194304 Nov 22 01:09 toFlash/edison_ifwi-dbg-00-dfu.bin                                        
-rw-rw-rw-  0 root   root      742 Nov 22 01:09 toFlash/filter-dfu-out.js                                                 
-rw-rw-rw-  0 root   root  4194468 Nov 22 01:09 toFlash/edison_ifwi-dbg-01.bin                                            
-rw-rw-rw-  0 root   root  4194304 Nov 22 01:09 toFlash/edison_ifwi-dbg-03-dfu.bin                                        
drwxrwxrwx  0 root   root        0 Nov 22 01:09 toFlash/helper/                                                           
drwxrwxrwx  0 root   root        0 Nov 22 01:09 toFlash/helper/images/                                                    
-rw-rw-rw-  0 root   root   315857 Nov 22 01:09 toFlash/helper/images/Edison-arduino-blink-led.png                        
-rw-rw-rw-  0 root   root   298970 Nov 22 01:09 toFlash/helper/images/Edison-breakout-board.png                           
bash-3.2$ 

When I use the Intel Edison Setup Application to install this file, I see it validate the image, and then there is a message about dfs (not sure about that) and then I get a window that says:

Uh-oh! Your Intel Edison did not update, This is likely because your Intel Edison did not reconnect properly, You'll need to download the newer firmware and manually update the Intel Edison Board. Detailed instructions are at https://software.intel.com/en-us/flashing-your-firmware-edison

I'm able to download the latest image from Intel and install that.

Should signpost/software/edison/toFlash.tar.gz work on an bare Intel Edison that is just connected to a SparkFun Base Kit for Intel® Edison KIT-14105 ?

• Hole sizes too small for radar module
• TX/RX silkscreen backwards
• PCB board name label too small
• MSGEQ7 schematic
• SAM4L switching power supply doesn't work

Thomas added a summon app which can be viewed at j2x.us/signp. This issue is for tracking updates that are needed for it. @tzachari

• Add time and date to the Controller view
• Add the UC Berkeley logo to the signpost
• Either add the UCSD logo to the Air Quality view or rename to UCSD Air Quality
• Make SIGNPOST name bigger. Might have to move it

To have a hope of tight time synchronization, modules are probably going to need to wire PPS to a timer/counter (TC) pin, so that it can trigger a counter start. This might not really help, because the SAM4L doesn't have a 32bit counter that can take an external trigger anyways, but it's something to think about.

We should spec out the actual quality of the PPS signal that modules can expect and include that in documentation

So far just nits but a place to collect things.

• Add a space on silkscreen to write in an identifier (i.e. Backplane no. 12)
• Move R46 and R47 to be near R48. (Or vice-versa and move the test points too)
• Software resettable fuses for modules
  • Moved to the power module's responsibility
• TVS diodes on all module I/O ?
  • I'm not convinced this is needed, can revisit if stuff actually breaks
• Deal with USB Hub Reset problem
  
  • Resolved by adding GPIO extender for reset line control. Debug backplane may still add PoR circuitry, but not really needed for signpost proper

• Numbers in silkscreen for modules
• Move part labels off of vias (korea is ok when the vias are small, these vias are big)
• Fix U30 package (and the other 5)
• Minimize plane area (?)
• USB data muxes have wrong control signals. Should be fed with MODn_USB_!EN, not MODn_ISO_USB
• Also, it seems like the unused port should have pull downs: https://e2e.ti.com/support/interface/digital_interface/f/130/p/367935/1295211#1295211
• Pull !BACKPLANE_RESET high instead of low
  • Add jumper spot for !backplane_reset to VCC
• Fix stubs on USB pull-down resistors (see 4.1 of http://www.usb.org/developers/docs/hs_usb_pdg_r1_0.pdf)
• Update power module for battery / solar attachment
• Add pullup !OVR for unused USB port

List for tracking problems discovered with Rev B of the controller

• Swap C12 and C25 locations

Issue to track fixes for controller RevA

• There are two pins labeled USB_D+ and none USB_D-

• Switch module 1 and 0 programming silkscreen (light is still correct)

• Shows the I2C bus
• Shows that there is an USB to edison
• Abstract the controller, storage, power supply
• Less detail than the existing arch diagram

I believe the current implementation relies solely on CPU timing for randomness (specifically for the ECDH key exchange). That's a problem because it's insecure and because we don't get to use our fancy TRNG driver for the SAM4L.

Instead, the signpost user libs can simply define

#define MBEDTLS_ENTROPY_HARDWARE_ALT

int mbedtls_hardware_poll( void *data,
                           unsigned char *output, size_t len, size_t *olen );

to call the underlying TRNG driver.

While this code is all public, I didn't see a license file or anything specified in the README. It'd be good to clarify this.

Things in libs don't get re-compiled when they change.

• USB !BUSPWR is wrong: Currently tied to GND, which states that the hub and all downstream devices are powered from the upstream port. Should be tied high, stating that the hub has an "external" power supply. In both modes, the hub still will perform overcurrent and poweroff protections.
  • Symptom: This limits the hub to 500mA max total draw, as each downstream port has a minimum 100mA allotment (USB spec), so the hub disables higher ports (cannot use USB in Module 5 or Module 7 slot)
  • Remedy: Cut trace at pin 10 (!BUSPWR) and short pins 10 and 9.

• U19 pin 1 indicator
• Better precision resistors for solar charger divider (Input and output regulation)
• More space monitors and sense resistors
• More space around charger/regulators
• U27 wrong package
• D3 wrong package
• Add reset button on watchdog

Need to select which LD file to use based on whether you make program or make flash

Tracking things to fix for the next revision:

• Grounded mounting holes
• Silkscreen labels on header pins
• Change to 0402 ferrite bead
• Remove extraneous pressure sensors. We can't even purchase all of them anymore

In the audio_module software/kernel/boards/audio_module/src/main.rs file the line
sam4l::pm::setup_system_clock(sam4l::pm::SystemClockSource::ExternalOscillator, 16000000); causes the baud rate to be 115200.
Changing the speed to 48 MHz
sam4l::pm::setup_system_clock(sam4l::pm::SystemClockSource::ExternalOscillator, 48000000); causes the baud rate to be 38400. Unsure as to why

• Better isolation for things touching the Linux power domain
• PPS connection to Storage Master
• Output Enable pulldown on logic level converter
• PWRBTN# net needs isolation from SAM4L. Doesn't currently allow Edison to run without powering the SAM4Ls
• Keep more space around Edison mounting holes

Sending "hello" to controller results in crash. Possibly due to expectations of incoming packet field lengths.

It looks like Travis isn't actually catching when our applications fail in Signpost. See https://travis-ci.org/lab11/signpost#L1168. It looks like some applications are still including firestorm.h and have been failing, but travis thinks all tests have passed.

@ppannuto Could you take a look at this at some point?

Below are some updates I'd like to see in the Edison Image:

• Install Node and npm. Node 6.9.5 is preferred. See Edison Node Notes. Accessors will require Node.

• Set the date. When I installed the default Intel-provided image, it seemed like the date was set properly, or at least close to the current time. I ran into problems using wget and npm because the date was set to 1/1/2000 and the certs were not valid.

• Set up a sbuser with the appropriate keys, see Set up the sbuser account on the Edison. Downloading accessors requires the sbuser account.

There is no big rush on these, just when we get around to updating the Edison image, it would be good to consider these.

When Theo was running through the Signpost tutorial, he found that the signpost-debug-radio pip package seems to have an issue and can't find _version.py. We should try to replicate this and fix it. Leaving this issue for now so that I remember it.

• R8 and C7 Labels are switched (which is bad if you want a working regulator)

Modify the paste layer on the ICs to have a smaller square in the center of the pad instead of the entire pad.

• Move temperature sensor off of ground plane

We want to improve radio module reliability. This has two prongs: detecting errors and resetting cleanly

1. We want to be able to detect errors on the Nucleum and iM880B. For the Nucleum, the ble_error function seems like it should be able to do this. However, while testing I could not seem to get this function to be run... For the iM880, we should read its responses. The console getauto function can probably do this adequately. Upon detecting an error, the app_watchdog_reset_app function can be called to reboot the module.
2. We want the radio board the reliably reset even if the Nucleum or iM880B have gotten into a bad state. The power gate on the Nucleum is toggled at start, getting it into a reliable state. Something needs to be done for the iM880B too. Using its power gate, however, seemed to lead to system problems and was being tested when the iM880 on my radio board gave up the ghost, leading to worries on my part that it might have been doing bad things for the board. @adkinsjd suggested there is a reset line that can be used instead.

• Missing silk labels for:
  • 
• C54/C72 are decapping the wrong thing (sigh..)
  • No, they were right, Eagle was just tricking me. VCC is pins 9 and 46 (or, 9*2 in the schematic), but decap was correctly on 46. Le sigh
• Isolation override switches have momentary short
  • Addressed by just letting the LEDs float when they're off instead of grounding them

• Add LED for fake radio module.

In the Bonus: Running two applications concurrently section of software/docs/TutorialSession.md, the apps/tock_examples/hello_app/build/storm/app.bin path is incorrect.

May need changes to the template/Makefile

It's unclear to new users what's located in which directory, and especially where the Tock submodule is located. Currently we point to Tock's getting started instructions, which assume you're in the Tock submodule to run commands. We should write a map of the repository and do something to make getting started make more sense to new users.

This probably goes hand-in-hand with defining an I2C message bus format.

In tock, I am still having to manually adjust my baud rate up to 125000 to get an actual baud rate of 115200.

With watchdog timer firing at 250 ms, app watchdog timer firing at 30 s, and app timer firing at 6 s, every 235 s the app seemingly stops getting a callback for the 6 s app timer. 30 s the app watchdog fires and resets the MCU.

Disabling the HW watchdog and watchdog timer (the 250 ms one) seems to make this problem go away. Changing the app timer interval does not seem to have any effect.