boards/arduino-mkr1000: add initial support (without wifi) #6666

aabadie · 2017-02-28T14:45:42Z

This PR adds a partial support for the Arduino MKR1000 board.

This board uses the same MCU as the Arduino Zero : a SAMD21G18A. But apparently, one has to keep the bootloader at the beginning of the flash memory : as a consequence I added a custom linker script.

The debug UART is also not available via the USB port so one needs an FTDI plugged on the RX/TX pins (pins number 13/14) on the board to access it.

Last thing (but not least) : to flash the board via USB, one needs to use Bossa (like for Arduino-due). For the moment, I have to double press the reset button each time I want to flash the board. This is not very convenient and it's not necessary with the Arduino IDE.

Adding an SPI driver for the ATWINC1500 Wifi module will also be required if one wants to add WiFi to this board.

aabadie · 2017-03-31T10:46:43Z

@haukepetersen, interested in testing this one ?

aabadie · 2017-04-14T14:48:58Z

I'm planning to add the wifi driver soon using a package from WiFi101 repository. But I prefer adding this in a follow-up PR. Are there people interested in testing this PR so we can merge it fast ?

kbumsik · 2017-05-02T12:43:06Z

Hi, I came from #6985 . I am currently merging your branch with my work. I have a question before making a PR to you branch, because I have no experience with programming with BOSSA: Why does Linkerscript's flash address start with 0x2000? Can't BOSSA access to system memory, or you want keep arduino bootloader?

aabadie · 2017-05-03T08:21:42Z

@kbumsik I merged your PR. Thanks ! I you could give your test result of this PR (flash, ADC, PWM, UART, etc) that would be great and we could merge it fast.
I'm also looking forward your PR with the ATWINC1500 driver :)

kbumsik · 2017-05-03T08:31:25Z

@aabadie Thanks for a quick feedback :) I will work on ATWINC1500 driver targeting both MKR1000 and WINC1500 XPlained Pro.
For testing, I can confirm that example/default UART terminal example, with pin TX: D14, RX: D13, works perfectly for now. I will update more testing result later.

kbumsik · 2017-05-06T08:40:04Z

This needs rebase and a fix for ADC after #7009.

aabadie · 2017-05-10T11:24:57Z

@kaspar030, I would go for option 3. Seems to be the more reasonable for me: point 1 break "we support latest Ubuntu LTS" as you said and point 2 breaks full cppcheck support (sam0.h is not a vendor file but RIOT code).

haukepetersen · 2017-05-11T10:20:09Z

boards/arduino-mkr1000/Makefile.include

+    include $(RIOTMAKE)/tools/jlink.inc.mk
+
+    # set default serial port depending on operating system
+    PORT_LINUX ?= /dev/ttyUSB0


This should be ttyACM0 for independent of the programmer, right? So IMHO these two lines should be taken out of this if block.

What do you mean with independent of the programmer ?
The board USB port is /dev/ttyACM0 but I don't know how it behaves if ones uses the JLINK programmer which I didn't test. If @kbumsik set it to /dev/ttyUSB0 there's probably a good reason.

I copied Makefile from stm32f4 discovery. Just like the case in stm32f4 discovery, /dev/ttyUSB0 is for UART serial converter. So you are right - this is independent to the JLink. I added this when I didn't know how bossac works.

Oh, ok, I understand now (sorry, for the delay...!)

haukepetersen · 2017-05-11T10:22:27Z

boards/arduino-mkr1000/Makefile.include

+    # when BOSSA is used, use a different flash map
+    # refer https://github.com/shumatech/BOSSA for this programmer
+    export PROGRAMMER = bossa
+    export CPU_MODEL = samd21g18a_mkr1000


Why a different CPU model here? If it is because of the different linkerscript, then it would be the cleaner solution to override the linkerscript variable here directly!

I was not aware of this variable, I'll change that as you suggest

haukepetersen · 2017-05-11T10:25:01Z

boards/arduino-mkr1000/dist/openocd.cfg

@@ -0,0 +1,7 @@
+source [find interface/cmsis-dap.cfg]


Does this board support being flashed using openocd through CMSIS-DAP? If so, this option is not usable as there is no option the board's Makefile.include... So I'd say either remove this file or add the needed options to the Makefile.include

It is not. We can just remove ir.

haukepetersen · 2017-05-11T10:25:51Z

boards/arduino-mkr1000/include/board.h

+ */
+#define XTIMER              TIMER_0
+#define XTIMER_CHAN         (0)
+/** @} */


no need for this block, as these are the default values. Just remove them.

haukepetersen · 2017-05-11T10:27:31Z

cpu/sam0_common/include/sam0.h

 #elif defined(CPU_MODEL_SAMD21G17AU)
  #include "vendor/samd21/include/samd21g17au.h"
-#elif defined(CPU_MODEL_SAMD21G18A)
+#elif defined(CPU_MODEL_SAMD21G18A) || defined(CPU_MODEL_SAMD21G18A_MKR1000)


as stated above, there should be no need for this varaible.

haukepetersen · 2017-05-11T10:35:31Z

About the Murdock issue: sam0.h should be ignored in any case, as this is basically vendor header. My proposed solution: #7038. Does this fix the timeout?

haukepetersen · 2017-05-11T10:38:35Z

One more question: I was not able to flash the board, as bossac gives me some GLIB errors:

/home/hauke/dev/riot/RIOT/boards/arduino-mkr1000/dist/flash.sh 
/home/hauke/dev/riot/RIOT/boards/arduino-mkr1000/dist/bossac: relocation error: /home/hauke/dev/riot/RIOT/boards/arduino-mkr1000/dist/bossac: symbol _ZNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEE9_M_createERmm, version GLIBCXX_3.4.21 not defined in file libstdc++.so.6 with link time reference
make: *** [flash] Error 127

Does anyone have a quick solution to this?

aabadie · 2017-05-11T16:36:46Z

Does anyone have a quick solution to this?

The bossa binary was built on my Linux using gcc 6.2. I probably have a too recent glibc and this is not compatible with yours. I'll rebuild it using an older Ubuntu (16.04)

aabadie · 2017-05-11T17:14:05Z

@haukepetersen, except your bossa issue, your other comments are addressed.

haukepetersen · 2017-05-12T09:35:08Z

Changes look good. I still can't flash my board (played around with some bossac versions and compiled it myself for a try), but it seems in the end I bricked my board... It doesn't register as /dev/ttyACM0 anymore...

About the bossa issue: I think it would make much sense (as a next step) to include it in form of a Makefile as part of the RIOT tools folder and get rid of the binaries for the arduino sam based boards after all. This way, users of these boards can build their bossac themselves and we get around shipping binaries (which was always sub-optimal). What do you think?

haukepetersen

Untested ACK (as my board seems bricked)

kbumsik · 2017-05-12T09:42:09Z

@haukepetersen , I am not sure it is your case, once you flash the MKR1000 using Bossa, it normally won't go back to the bootloader mode so you won't see it as tty device. If you want to flash it again, you may double press the reset button and it will go to the bootloader mode. The board will tell you it is in the bootloader mode by slowly blinking the green LED.

haukepetersen · 2017-05-12T10:07:56Z

Sounds like a plan :-) Will give it a try...

aabadie · 2017-05-12T11:41:49Z

About the bossa issue: I think it would make much sense (as a next step) to include it in form of a Makefile as part of the RIOT tools folder

+1 good idea

aabadie · 2017-05-12T11:47:07Z

@haukepetersen, regarding the "bricked" board, as @kbumsik said, just double press the reset button before flashing. I initially added a comment about that in flash script
Thinking of it, the stty line is probably useless (originally copy pasted from arduino-due)

haukepetersen · 2017-05-12T11:58:54Z

boards/arduino-mkr1000/Makefile.include

+    # when BOSSA is used (default), use a different flash map
+    # refer https://github.com/shumatech/BOSSA for this programmer
+    export PROGRAMMER = bossa
+    export LINKER_SCRIPT ?= $(RIOTCPU)/$(CPU)/ldscripts/$(CPU_MODEL)_mkr1000.ld


The path to the linkerscript must be ~~$(RIOTCPU)/sam0/ldscripts/$(CPU_MODEL)_mkr1000.ld~~
EDIT:
$(RIOTCPU)/sam0_common/ldscripts/$(CPU_MODEL)_mkr1000.ld

indeed, missed that change

haukepetersen · 2017-05-16T07:58:25Z

Looks good to me now, though the issue with the bossa tool remains. We should simply include a global build script/Makefile in RIOT to build that tool - and then use that for the mkr1000, the due, and udoo. Maybe I find time to look into this... But until then, I am ok with merging this PR -> please squash.

…oards

aabadie · 2017-05-16T08:15:54Z

squashed and rebased.

haukepetersen · 2017-05-16T08:30:27Z

Murdock is happy -> let's go!

aabadie added Platform: ARM Platform: This PR/issue effects ARM-based platforms Type: new feature The issue requests / The PR implemements a new feature for RIOT labels Feb 28, 2017

aabadie added this to the Release 2017.04 milestone Feb 28, 2017

aabadie assigned haukepetersen Feb 28, 2017

aabadie force-pushed the mkr1000 branch from 348c2b4 to 9cb123e Compare February 28, 2017 16:38

aabadie assigned kYc0o Feb 28, 2017

aabadie force-pushed the mkr1000 branch from 093aede to c62734a Compare February 28, 2017 19:25

aabadie force-pushed the mkr1000 branch 2 times, most recently from 1149bd0 to f6a3d83 Compare March 10, 2017 17:11

aabadie force-pushed the mkr1000 branch from f6a3d83 to b41d0d3 Compare March 31, 2017 10:45

aabadie force-pushed the mkr1000 branch 3 times, most recently from c674ecf to 8c8133d Compare April 10, 2017 06:09

aabadie mentioned this pull request Apr 10, 2017

boards/arduino-mkrzero: add initial support #6881

Merged

aabadie force-pushed the mkr1000 branch from 8c8133d to 47221c1 Compare April 14, 2017 14:46

aabadie force-pushed the mkr1000 branch 2 times, most recently from 956807c to 08fe0e2 Compare April 18, 2017 12:12

kaspar030 modified the milestones: Release 2017.04, Release 2017.07 Apr 21, 2017

aabadie added the Community: Hack'n'ACK candidate This PR is a candidate for review and discussion during one of RIOT's monthly Hack'n'ACK parties label Apr 25, 2017

aabadie mentioned this pull request May 1, 2017

Segger GDB with TUI has bad newlines. #6985

Closed

kbumsik mentioned this pull request May 3, 2017

MKR1000: Updated periph config and SWD flashing aabadie/RIOT#3

Merged

haukepetersen requested changes May 11, 2017

View reviewed changes

aabadie force-pushed the mkr1000 branch from 3e9b1c4 to c14be8c Compare May 11, 2017 17:12

haukepetersen approved these changes May 12, 2017

View reviewed changes

haukepetersen requested changes May 12, 2017

View reviewed changes

aabadie force-pushed the mkr1000 branch 2 times, most recently from 6cd8024 to 6158fac Compare May 13, 2017 17:43

haukepetersen approved these changes May 16, 2017

View reviewed changes

aabadie and others added 4 commits May 16, 2017 10:14

cpu/sam0_common: add support for MCU used in Arduino MKR1000 board

d768f5b

boards/arduino-mkr1000: initial support

f97b449

test/unittests: add arduino-mkr1000 for ARM and unsufficient memory b…

4c9866d

…oards

boards/arduino-mkr1000: Updated periph config and SWD flashing (#3)

67d7346

aabadie force-pushed the mkr1000 branch from 6158fac to 67d7346 Compare May 16, 2017 08:15

haukepetersen merged commit 8f8c7e3 into RIOT-OS:master May 16, 2017

aabadie deleted the mkr1000 branch May 16, 2017 09:07

kbumsik mentioned this pull request May 17, 2017

Makefiles: How to add subdirectories in an application directory? #7070

Closed

kbumsik mentioned this pull request Jun 1, 2017

cpu/samd21: Added CPU_MODEL_SAMD21G18A interrupt config #7125

Merged

kbumsik mentioned this pull request Oct 8, 2017

WIP: driver/winc1500: Initial port for Wi-Fi module against NETDEV_ETH. #7433

Closed

4 tasks

boards/arduino-mkr1000: add initial support (without wifi) #6666

boards/arduino-mkr1000: add initial support (without wifi) #6666

Uh oh!

Conversation

aabadie commented Feb 28, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

aabadie commented Mar 31, 2017

Uh oh!

aabadie commented Apr 14, 2017

Uh oh!

kbumsik commented May 2, 2017

Uh oh!

aabadie commented May 3, 2017

Uh oh!

kbumsik commented May 3, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kbumsik commented May 6, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

aabadie commented May 10, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kbumsik May 11, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

haukepetersen commented May 11, 2017

Uh oh!

haukepetersen commented May 11, 2017

Uh oh!

aabadie commented May 11, 2017

Uh oh!

aabadie commented May 11, 2017

Uh oh!

haukepetersen commented May 12, 2017

Uh oh!

haukepetersen left a comment

Choose a reason for hiding this comment

Uh oh!

kbumsik commented May 12, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

haukepetersen commented May 12, 2017

Uh oh!

aabadie commented May 12, 2017

Uh oh!

aabadie commented May 12, 2017

Uh oh!

haukepetersen May 12, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

haukepetersen commented May 16, 2017

Uh oh!

aabadie commented May 16, 2017

Uh oh!

haukepetersen commented May 16, 2017

aabadie commented Feb 28, 2017 •

edited

Loading

kbumsik commented May 3, 2017 •

edited

Loading

kbumsik commented May 6, 2017 •

edited

Loading

kbumsik May 11, 2017 •

edited

Loading

kbumsik commented May 12, 2017 •

edited

Loading

haukepetersen May 12, 2017 •

edited

Loading