boards: add mcb2388

[benpicco]

Contribution description

The MCB2388 Evaluation Board is a development board for the LPC2388 MCU.

LPC2388 and LPC2387 are very similar, the only difference is that LPC2388 can also operate in USB host mode.

Since this is not used yet, I chose to just pretend it's an lpc2387.

So far, only the two UARTs, LEDs and LCD display are configured.

Testing procedure

I tried examples/default to blink the LEDs with SAUL.
tests/driver_hd44780 also displays text on the LCD.

Issues/PRs references

Was used for past and future cpu/lpc2387 PRs.

[aabadie]

Looks good in general but I have no hardware for testing. I just did a quick code review and found some things to change.

[maribu]

@leandrolanzieri, @aabadie: I think code-wise this is fine. Please speak up if you find something that I have overlooked.

@benpicco: I should be able to test this with an MSB-A2, provided I override the programmer configuration, right?

[benpicco]

I should be able to test this with an MSB-A2, provided I override the programmer configuration, right?

I'm not sure how useful that would be. Last time I checked the MSB-A2 didn't have an LCD display or 8 debug LEDs.

So you'd just be testing common code.

[maribu]

I'm not sure how useful that would be. Last time I checked the MSB-A2 didn't have an LCD display or 8 debug LEDs.

I would attach a logic analyzers to the pins of the MCU. But this helps little to verify that the configuration indeed matches the pins the LEDs & LCD are actually connected to...

[benpicco]

I ran all the tests, the only failures were:

Failures during compilation:
- [tests/lwip_sock_ip](tests/lwip_sock_ip/compilation.failed)
- [tests/lwip_sock_tcp](tests/lwip_sock_tcp/compilation.failed)
- [tests/lwip_sock_udp](tests/lwip_sock_udp/compilation.failed)

savannah.nongnu.org is currently down. Since this is all using 'virtual' network interfaces, I expect no board-specific surprises here anyway.

Failures during test:

• tests/cpp_ctors fixed by cpu/lpc2387: clean up lpc2387.ld, fixes tests/cpp_ctors #12830
• tests/driver_grove_ledbar - test uses GPIO_PIN(0,1) & GPIO_PIN(0,2) - GPIO_PIN(0,2) is also used for UART TX
• tests/driver_my9221 - dito
• tests/gnrc_ipv6_ext
• tests/gnrc_ipv6_ext_frag
• tests/gnrc_ipv6_nib_6ln
• tests/gnrc_rpl_srh
• tests/gnrc_sixlowpan
• tests/gnrc_sock_dns
• tests/gnrc_tcp
• tests/pkg_libcose
• tests/pkg_qdsa
• tests/pkg_relic - dito
• tests/pkg_tweetnacl - works with CFLAGS += -DTHREAD_STACKSIZE_MAIN=\(10*THREAD_STACKSIZE_DEFAULT\). (THREAD_STACKSIZE_DEFAULT is 512 on lpc2387 but 1024 on cortexm_common)
• tests/pthread_tls
• tests/shell
• tests/thread_cooperation works now
• tests/unittests - pkt_tests.sfprintf_float (tests/unittests/tests-printf_float/tests-printf_float.c 40) exp "2016.034900" was "-0.000000"
• tests/xtimer_periodic_wakeup - Min/max error: 1/15, no failue
  python3 dist/tools/compile_and_test_for_board/compile_and_test_for_board.py . 800,59s user 214,88s system 19% cpu 1:27:23,71 total

Will look into theses!

[benpicco]

Ran the tests again with the latest master.
Still some float errors (but those are unrelated to the board)

Failures during test:

• examples/micropython

Terminal program miniterm.py is required but not found in PATH. Aborting.

huh?

• tests/driver_grove_ledbar
• tests/driver_my9221

Those tests use the UART TX pin as GPIO for the driver - that won't work.

• tests/gnrc_ipv6_ext
• tests/gnrc_ipv6_ext_frag
• tests/gnrc_rpl_srh
• tests/gnrc_sock_dns
• tests/gnrc_tcp

This test requires root privileges.
It's constructing and sending Ethernet frames.

• tests/periph_timer_short_relative_set

2019-12-25 17:40:34,087 - INFO # interval 2 ok
2019-12-25 17:40:34,088 - INFO # ERROR: too long delay, aborted after 1001 (TEST_MAX_DIFF=1000)

fixed by #13003
(samr21-xpro fails after interval 8)

• tests/pkg_libfixmath_unittests

…
OK: fix16_from_str("0.99998") == 65535
OK: fix16_from_str("32767.99998") == fix16_maximum
OK: fix16_from_str("-32768.00000") == fix16_minimum

----Extended testing for whole range----
Value (fix16_t)-2147483648 gave -32768.00000, should be 0.00000
Value (fix16_t)-2147418111 gave -32766.99998, should be -98711338753
Value (fix16_t)-2147352574 gave -32765.99997, should be -15950035448
Value (fix16_t)-2147287037 gave -32764.99995, should be -41536550314
…

• tests/pkg_qdsa

qDSA_tests.test_qDSA_sign_verify (tests/pkg_qdsa/main.c 58) exp 0 was 1

• tests/shell

Uses socat which doesn't toggle the RTS lines.
Works when removing the 'RST' jumper to disable reset on UART.

• tests/unittests

pkt_tests.sfprintf_float (tests/unittests/tests-printf_float/tests-printf_float.c 40) exp "2016.034900" was "-0.000000"

I'll squash the ADC commits should #12911 get merged before this, otherwise I'll remove them and add them in the other PR.

[aabadie]

Looks pretty good.

@benpicco, do you have any idea on how to fix the remaining failing tests (tests/pkg_qdsa, tests/pkg_libfixmath-unittests, tests/unittests). It seems to me that float support doesn't work well on this platform.

[aabadie]

Code changes are good so I won't block this anymore.

@benpicco, would you mind opening an issue to list the remaining failing tests on this board so we can keep track of them.

ACK, please squash (and remove the ADC related commits)

[benpicco]

Thank you for the review!