Open source Licence

[brotherdust]

@Fred : Thank you very much! Having the code answers many of the questions I have been thinking about asking. Great work to all. I know it's been a long haul. Keep up the good work!

[thibault]

Nice news!

[LoneTech]

Great news! I've cloned a copy and managed to build what appears to be a LoPy firmware.

Rough steps performed:
Get ESP32 compiler toolchain per instructions at https://github.com/pycom/pycom-esp-idf/blob/master/docs/linux-setup.rst (there exist instructions for other platforms too), adding commands xtensa-esp32-elf-gcc etc to PATH
Cloned Pycom's IDF and micropython repositories, setting export IDF_PATH=~/LoPy/pycom-esp-idf (adjust path to where you placed it)
Enter micropython repo directory for ESP32: cd pycom-micropython/esp32/
Workaround for unexpected Makefile behaviour: ln -s ../lib bootloader/lib (this shouldn't be needed, probably should change -Lbootloader/lib to -Llib in Makefile, alternatively make sure it builds bootloader variants of libraries)
Build firmware: make BOARD=LOPY LORA_BAND=USE_BAND_868
New firmware is saved in build/LOPY/release/appimg.bin

A warning produced by the build suggests that it did newline conversion on a PDF. That's going to end badly as it's a binary format, though it's a packaging issue and doesn't affect building the firmware itself.

Looking good so far, though it'll take a little while to find where I can touch it up. Expect a pull request or two in the future. :)

P.S. After curl --disable-epsv --ftp-pasv -T build/LOPY/release/appimg.bin ftp://micro:python@ESP_00C2C6//flash/sys/appimg.bin my LoPy said MicroPython e2cc46a-dirty on 2016-11-15; LoPy with ESP32 :)

[Fred]

@LoneTech Great stuff! Have fun!

[Xykon]

In case anyone finds it helpful I mirrored the html docs from esp-idf on my webserver at https://www.tiny-dev.com/esp32/

[livius]

@LoneTech
If you or someone have same guide for Windows :)

[Xykon]

@livius said in Open source Licence:

@LoneTech
If you or someone have same guide for Windows :)

https://www.tiny-dev.com/esp32/windows-setup.html

[livius]

It is incompleat
i do not see something like this in documentation
make BOARD=LOPY LORA_BAND=USE_BAND_868
as was posted by @LoneTech

and something like

After curl --disable-epsv --ftp-pasv -T build/LOPY/release/appimg.bin ftp://micro:python@ESP_00C2C6//flash/sys/appimg.bin my LoPy said MicroPython e2cc46a-dirty on 2016-11-15; LoPy with ESP32 :)

[Xykon]

make BOARD=LOPY LORA_BAND=USE_BAND_868

It's identical under all OS. Read Makefile and application.mk in the pycom-micropython/esp32 folder.

curl with these specific options is used as a command line ftp client. You can just as well use FileZilla following the instructions at https://docs.pycom.io/lopy/lopy/general.html#upgrading-the-firmware-over-the-air

[livius]

this should be in documentation. Reading script file is bad way for starters ;-)
but now it is clear

curl with these specific options is used as a command line ftp client
ah i see, he only copy image file throught ftp
i did not read whole line carefully - but it was so simple
when i saw curl i have stoped reading ;-)

[livius]

thank you for help
i now have make WIPI bin file
and i am ready to make some pull requests :)

[brotherdust]

@LoneTech Thanks for the instructions. I was able to successfully build on my Windows 10 box using Bash on Ubuntu on Windows. =) heh.

[brotherdust]

@brotherdust said in Open source Licence:

@LoneTech Thanks for the instructions. I was able to successfully build on my Windows 10 box using Bash on Ubuntu on Windows. =) heh.

So, yeah. It might have built successfully, but I can't get the LoPy to take it. Argh. Here's the CLI output:

$ curl --disable-epsv --ftp-pasv -T "build/LOPY/release/appimg.bin" ftp://micro:python@10.192.5.64//flash/sys/appimg.bin
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
 31  819k    0     0   31  256k      0   530k  0:00:01 --:--:--  0:00:01  534k
curl: (55) Send failure: Broken pipe

I think it's because the firmware file is too big. The one I compiled is ~820KB, but the official releases are weighing in around ~631KB. Is it possible this is the cause?

[LoneTech]

@brotherdust Unlikely; the firmware partitions are 1M large, and your transfer stopped at 256k. OTA updating isn't quite 100% for me; not sure if that readonly flash bug is reoccurring or what, but doing a serial download tends to clear it. Also, on the firmware versions with the watchdog bug OTA almost always triggers that.

[nickparrin]

Keep getting an error like this:

developer@developer:~/esp32/pycom-micropython/esp32$ make BOARD=LOPY LORA_BAND=USE_BAND_868
LINK build/LOPY/release/bootloader/bootloader.elf
xtensa-esp32-elf-gcc -nostdlib -Wl,-Map=build/LOPY/release/bootloader/bootloader.map -Wl,--no-check-sections -u call_user_start_cpu0 -Wl,-static -Wl,--undefined=uxTopUsedPriority -Wl,--gc-sections -T esp32.bootloader.ld -T esp32.rom.ld -T esp32.peripherals.ld -Wl,--start-group -Llib -L/home/developer/esp32/pycom-esp-idf/components/esp32/ld -Lbuild/LOPY/release/bootloader -L/home/developer/esp32/pycom-esp-idf/components/esp32/lib -Lbootloader -llog -lcore -lspi_flash build/LOPY/release/bootloader/bootloader.a -Wl,--end-group -Wl,-EL -o build/LOPY/release/bootloader/bootloader.elf
/home/developer/esp32/xtensa-esp32-elf/bin/../lib/gcc/xtensa-esp32-elf/4.8.5/../../../../xtensa-esp32-elf/bin/ld: cannot find -lcore
collect2: error: ld returned 1 exit status
application.mk:274: recipe for target 'build/LOPY/release/bootloader/bootloader.elf' failed
make: *** [build/LOPY/release/bootloader/bootloader.elf] Error 1

I set the IDF_PATH and added gcc to PATH. Tried with the ln command, and without by editing the Makefile directly

EDiT: fixed it thanks to daniel. I missed to import (inside pycom idf) the needed compontents:

• git submodule update --init

[livius]

Will be good to see port of this
https://github.com/micropython/micropython/pull/2091/files
and also enable time_us() which is currently commented out in the code
but i suppose it is limited by some esp32 IDF? and because of this is commented out?
https://github.com/pycom/pycom-micropython/blob/master/esp32/mods/modutime.c

[daniel]

@livius yes, machine.time_pulse_us() is very useful and we will implement it for the ESP32. A number of function in the time module are incomplete due to previous limitations of the IDF. The plan is to get back to that and implement them in early January.

Cheers,
Daniel

[livius]

@daniel
thank you for the info - especially about time frame

[livius]

@daniel
Can i ask what is IRAM_ATTR in functions declaration?

[daniel]

@livius that's needed in order to place functions inside the instruction RAM of the ESP32. Otherwise they will reside in flash, which means that they need to be fetched from the SPI flash everytime there's a cache miss. Time critical functions must be placed inside IRAM, as well as interrupt handlers.