lwIP: Initial import by miri64 · Pull Request #3551 · RIOT-OS/RIOT

miri64 · 2015-08-03T20:35:29Z

Imports lwIP for RIOT as a pkg.

~~Depends on #3549.~~

Todos:

jnohlgard · 2015-08-03T21:14:41Z

Could you explain how this relates to netdev/netif and gnrc? Would this stack offer the same functionality as some parts of gnrc? Are there any differences in supported features etc?

miri64 · 2015-08-03T21:35:46Z

I port this stack mainly for comparability in my master thesis (there will follow a few more implementations over the next weeks ;-)). I will port lwIP's driver interface so that it will speak to netdev. As far as I know (but I did not delve deep enough yet to say for sure) the limitations compared to gnrc are:

~~no multi-interface support~~ (that's not true)
no 6LoWPAN (yet)
less modular

The main advantage might be it's size, but the final verdict will have to wait until I finished my master thesis ;-)

miri64 · 2015-10-25T19:51:49Z

Ported to and rebased upon new semaphore module

miri64 · 2015-11-24T22:03:02Z

Rebased to current master

miri64 · 2015-11-30T16:33:36Z

Tested with conn_ip and it worked \o/

OlegHahm · 2015-11-30T18:29:45Z

pkg/lwip/0001-Fix-warnings.patch

Why is this patch required?

As the patch message states: it fixes warnings, that would make it impossible to compile without W_ERROR=0

(and no, I don't remember the precise warning ;-P)

For the record: the warning was something like pkg/lwip/lwip/src/include/lwip/debug.h:71:32: error: suggest braces around empty body in an ‘if’ statement [-Werror=empty-body]

Ah right, and I thought then: why have a do { } while (0) around it anyway then?

This is a reminder for me to try it without this patch

Does not compile without the patch, so yes it is required.

Do you think there could be a way to get this patch upstream?

I can try...

The problem is that it is technically perfectly fine code, since the #ifndef throws an error anyway if LWIP_PLATFORM_ASSERT isn't defined and it's just our -Werror that makes problems here. In my experience with the community there I don't think this will come through.

miri64 · 2015-11-30T23:03:41Z

fixed a few assertion related issues

miri64 · 2015-11-30T23:39:10Z

Now I tested receiving of IPv6 frames, too ;-)

OlegHahm · 2015-12-01T00:56:32Z

pkg/lwip/include/arch/cc.h

Shouldn't we put parens around the x?

No, this is a macro to build structs platform indepentent.

OlegHahm · 2016-03-30T14:14:37Z

Check this: remove the ifneq and call make info-boards-supported again.

miri64 · 2016-03-30T14:19:17Z

First let me look what Murdock is saying about this.

miri64 · 2016-03-30T14:20:49Z

Check this: remove the ifneq and call make info-boards-supported again.

airfy-beacon arduino-due avsextrem cc2538dk fox frdm-k64f iotlab-m3 limifrog-v1 msba2 msbiot mulle nrf6310 nucleo-f103 nucleo-f303 nucleo-f334 nucleo-f401 nucleo-l1 openmote-cc2538 pba-d-01-kw2x pca10005 pttu remote saml21-xpro samr21-xpro stm32f0discovery stm32f3discovery stm32f4discovery udoo weio yunjia-nrf51822

OlegHahm · 2016-03-30T14:20:57Z

DRIVER keeps set to netdev2_tap in buildtest.

miri64 · 2016-03-30T14:23:46Z

Should be reset in https://github.com/RIOT-OS/RIOT/pull/3551/files#diff-b2af745a4a882b6e5cb2abb8b6ac785bR23 (l. 23). But anyway, can you try and see, what happens if the colon is removed before the equal sign for the DRIVER assignments?

OlegHahm · 2016-03-30T14:26:19Z

same result

OlegHahm · 2016-03-30T14:29:55Z

I think the problem is somehow related to the fact that $BOARD is always set to native while $DRIVER is set when running make buildtest.

If you do BOARD=iotlab-m3 make buildtest or BOARD=iotlab-m3 make info-boards-supported everything is fine.

miri64 · 2016-03-30T14:33:58Z

Then setting BOARD to e.g. iotlab-m3 in l.3 would do the trick?

OlegHahm · 2016-03-30T14:36:58Z

As a workaround for the broken FEATURES_REQUIRED/make buildtest thing, that would work, yes.

miri64 · 2016-03-30T14:41:16Z

For some reason the Makefile on Murdock is only evaluated after the build happens:

In file included from /data/riotbuild/drivers/at86rf2xx/at86rf2xx_netdev.c:33:0:
/data/riotbuild/drivers/include/at86rf2xx.h:139:5: error: unknown type name 'spi_t'
     spi_t spi;              /**< SPI bus the device is connected to */
     ^
/data/riotbuild/drivers/include/at86rf2xx.h:140:5: error: unknown type name 'spi_speed_t'
     spi_speed_t spi_speed;  /**< SPI speed to use */
     ^
/data/riotbuild/Makefile.base:60: recipe for target '/data/riotbuild/tests/lwip/bin/ek-lm4f120xl/at86rf2xx/at86rf2xx_netdev.o' failed
make[3]: *** [/data/riotbuild/tests/lwip/bin/ek-lm4f120xl/at86rf2xx/at86rf2xx_netdev.o] Error 1
make[3]: *** Waiting for unfinished jobs....
In file included from /data/riotbuild/drivers/at86rf2xx/include/at86rf2xx_internal.h:27:0,
                 from /data/riotbuild/drivers/at86rf2xx/at86rf2xx.c:31:
/data/riotbuild/drivers/include/at86rf2xx.h:139:5: error: unknown type name 'spi_t'
     spi_t spi;              /**< SPI bus the device is connected to */
     ^
/data/riotbuild/drivers/include/at86rf2xx.h:140:5: error: unknown type name 'spi_speed_t'
     spi_speed_t spi_speed;  /**< SPI speed to use */
     ^
/data/riotbuild/drivers/at86rf2xx/at86rf2xx.c: In function 'at86rf2xx_setup':
/data/riotbuild/drivers/at86rf2xx/at86rf2xx.c:48:5: error: implicit declaration of function 'spi_init_master' [-Werror=implicit-function-declaration]
     spi_init_master(dev->params.spi, SPI_CONF_FIRST_RISING, params->spi_speed);
     ^
/data/riotbuild/drivers/at86rf2xx/at86rf2xx.c:48:38: error: 'SPI_CONF_FIRST_RISING' undeclared (first use in this function)
     spi_init_master(dev->params.spi, SPI_CONF_FIRST_RISING, params->spi_speed);
                                      ^
/data/riotbuild/drivers/at86rf2xx/at86rf2xx.c:48:38: note: each undeclared identifier is reported only once for each function it appears in
cc1: all warnings being treated as errors
/data/riotbuild/Makefile.base:60: recipe for target '/data/riotbuild/tests/lwip/bin/ek-lm4f120xl/at86rf2xx/at86rf2xx.o' failed
make[3]: *** [/data/riotbuild/tests/lwip/bin/ek-lm4f120xl/at86rf2xx/at86rf2xx.o] Error 1
/data/riotbuild/Makefile.base:20: recipe for target 'ALL--/data/riotbuild/drivers/at86rf2xx' failed
make[2]: *** [ALL--/data/riotbuild/drivers/at86rf2xx] Error 2
make[2]: *** Waiting for unfinished jobs....
In file included from /data/riotbuild/pkg/lwip/contrib/lwip.c:26:0:
/data/riotbuild/drivers/include/at86rf2xx.h:139:5: error: unknown type name 'spi_t'
     spi_t spi;              /**< SPI bus the device is connected to */
     ^
/data/riotbuild/drivers/include/at86rf2xx.h:140:5: error: unknown type name 'spi_speed_t'
     spi_speed_t spi_speed;  /**< SPI speed to use */
     ^
In file included from /data/riotbuild/pkg/lwip/contrib/lwip.c:27:0:
/data/riotbuild/drivers/at86rf2xx/include/at86rf2xx_params.h:35:38: error: 'SPI_0' undeclared here (not in a function)
 #define AT86RF2XX_PARAM_SPI         (SPI_0)
                                      ^
/data/riotbuild/drivers/at86rf2xx/include/at86rf2xx_params.h:53:45: note: in expansion of macro 'AT86RF2XX_PARAM_SPI'
 #define AT86RF2XX_PARAMS_DEFAULT    {.spi = AT86RF2XX_PARAM_SPI, \
                                             ^
/data/riotbuild/drivers/at86rf2xx/include/at86rf2xx_params.h:69:5: note: in expansion of macro 'AT86RF2XX_PARAMS_DEFAULT'
     AT86RF2XX_PARAMS_DEFAULT,
     ^
/data/riotbuild/drivers/at86rf2xx/include/at86rf2xx_params.h:38:38: error: 'SPI_SPEED_5MHZ' undeclared here (not in a function)
 #define AT86RF2XX_PARAM_SPI_SPEED   (SPI_SPEED_5MHZ)
                                      ^
/data/riotbuild/drivers/at86rf2xx/include/at86rf2xx_params.h:54:51: note: in expansion of macro 'AT86RF2XX_PARAM_SPI_SPEED'
                                      .spi_speed = AT86RF2XX_PARAM_SPI_SPEED, \
                                                   ^
/data/riotbuild/drivers/at86rf2xx/include/at86rf2xx_params.h:69:5: note: in expansion of macro 'AT86RF2XX_PARAMS_DEFAULT'
     AT86RF2XX_PARAMS_DEFAULT,
     ^
/data/riotbuild/Makefile.base:60: recipe for target '/data/riotbuild/tests/lwip/bin/ek-lm4f120xl/lwip_contrib/lwip.o' failed
make[2]: *** [/data/riotbuild/tests/lwip/bin/ek-lm4f120xl/lwip_contrib/lwip.o] Error 1
make[2]: *** Waiting for unfinished jobs....
/data/riotbuild/Makefile.base:20: recipe for target 'ALL--/data/riotbuild/drivers' failed
make[1]: *** [ALL--/data/riotbuild/drivers] Error 2
make[1]: *** Waiting for unfinished jobs....
/data/riotbuild/Makefile.base:20: recipe for target 'ALL--/data/riotbuild/pkg/lwip/contrib' failed
make[1]: *** [ALL--/data/riotbuild/pkg/lwip/contrib] Error 2
/data/riotbuild/Makefile.include:253: recipe for target 'all' failed
make: *** [all] Error 2
Makefile:20: "START: DRIVER = '' FEATURES_REQUIRED = ''"
Makefile:40: "BEFORE CRITICAL: DRIVER = 'at86rf231' FEATURES_REQUIRED = ''"
Makefile:43: "IN CRITICAL: DRIVER = 'at86rf231' FEATURES_REQUIRED = ''"
Makefile:47: "AFTER CRITICAL: DRIVER = 'at86rf231' FEATURES_REQUIRED = 'periph_spi periph_gpio'"
Makefile:49: "END: DRIVER = 'at86rf231' FEATURES_REQUIRED = 'periph_spi periph_gpio'"

compare that to my output:

Makefile:20: "START: DRIVER = '' FEATURES_REQUIRED = ''"
Makefile:40: "BEFORE CRITICAL: DRIVER = 'at86rf231' FEATURES_REQUIRED = ''"
Makefile:43: "IN CRITICAL: DRIVER = 'at86rf231' FEATURES_REQUIRED = ''"
Makefile:47: "AFTER CRITICAL: DRIVER = 'at86rf231' FEATURES_REQUIRED = 'periph_spi periph_gpio'"
Makefile:49: "END: DRIVER = 'at86rf231' FEATURES_REQUIRED = 'periph_spi periph_gpio'"
There are unsatisfied feature requirements: periph_spi


EXPECT ERRORS!


Building application "lwip" for "ek-lm4f120xl" with MCU "lm4f120".

In file included from /home/martine/Repositories/RIOT-OS/RIOT/drivers/at86rf2xx/include/at86rf2xx_internal.h:27:0,
                 from /home/martine/Repositories/RIOT-OS/RIOT/drivers/at86rf2xx/at86rf2xx.c:31:
/home/martine/Repositories/RIOT-OS/RIOT/drivers/include/at86rf2xx.h:139:5: error: unknown type name 'spi_t'
     spi_t spi;              /**< SPI bus the device is connected to */
     ^
/home/martine/Repositories/RIOT-OS/RIOT/drivers/include/at86rf2xx.h:140:5: error: unknown type name 'spi_speed_t'
     spi_speed_t spi_speed;  /**< SPI speed to use */
     ^
/home/martine/Repositories/RIOT-OS/RIOT/drivers/at86rf2xx/at86rf2xx.c: In function 'at86rf2xx_setup':
/home/martine/Repositories/RIOT-OS/RIOT/drivers/at86rf2xx/at86rf2xx.c:48:5: error: implicit declaration of function 'spi_init_master' [-Werror=implicit-function-declaration]
     spi_init_master(dev->params.spi, SPI_CONF_FIRST_RISING, params->spi_speed);
     ^
/home/martine/Repositories/RIOT-OS/RIOT/drivers/at86rf2xx/at86rf2xx.c:48:38: error: 'SPI_CONF_FIRST_RISING' undeclared (first use in this function)
     spi_init_master(dev->params.spi, SPI_CONF_FIRST_RISING, params->spi_speed);
                                      ^
/home/martine/Repositories/RIOT-OS/RIOT/drivers/at86rf2xx/at86rf2xx.c:48:38: note: each undeclared identifier is reported only once for each function it appears in
cc1: all warnings being treated as errors
make[3]: *** [/home/martine/Repositories/RIOT-OS/RIOT/tests/lwip/bin/ek-lm4f120xl/at86rf2xx/at86rf2xx.o] Error 1
make[2]: *** [ALL--/home/martine/Repositories/RIOT-OS/RIOT/drivers/at86rf2xx] Error 2
make[1]: *** [ALL--/home/martine/Repositories/RIOT-OS/RIOT/drivers] Error 2
make: *** [all] Error 2

miri64 · 2016-03-30T14:42:28Z

Okay adapted for workaround now.

OlegHahm · 2016-03-30T14:53:54Z

It's still saying BOARD ?= native

miri64 · 2016-03-30T15:00:08Z

Derp...

OlegHahm · 2016-03-30T15:11:45Z

\o/ Murdock is finally green like Ireland!

miri64 · 2016-03-30T15:26:21Z

Then merge at will :-)

miri64 · 2016-03-30T15:36:52Z

Fixed assert issues and squashed immediately.

OlegHahm · 2016-03-30T15:54:24Z

ACK

miri64 · 2016-03-30T16:20:33Z

Murdock is happy.

OlegHahm · 2016-03-30T16:21:55Z

I'd say: the honor to push the button is yours!

miri64 · 2016-03-30T16:22:24Z

Thanks for the review :-)

OlegHahm · 2016-03-30T16:22:50Z

It was a pleasure!

miri64 assigned haukepetersen Aug 3, 2015

miri64 added this to the Release NEXT MAJOR milestone Aug 3, 2015

Kijewski mentioned this pull request Aug 21, 2015

sem: initial import of a lightweight semaphore layer #3549

Merged

miri64 force-pushed the pkg/port/lwip branch from cd68058 to bc6425a Compare October 25, 2015 19:50

miri64 force-pushed the pkg/port/lwip branch 6 times, most recently from f2d9e37 to b0f2873 Compare November 16, 2015 15:01

miri64 force-pushed the pkg/port/lwip branch from 2ac4bcb to f81323a Compare November 24, 2015 22:02

miri64 removed the State: waiting for other PR State: The PR requires another PR to be merged first label Nov 24, 2015

miri64 force-pushed the pkg/port/lwip branch 2 times, most recently from 6fbebaf to c0baa12 Compare November 30, 2015 16:32

OlegHahm reviewed Nov 30, 2015
View reviewed changes

OlegHahm reviewed Dec 1, 2015
View reviewed changes

OlegHahm mentioned this pull request Mar 30, 2016

make: buildest fails with board dependent application Makefiles #5207

Closed

miri64 force-pushed the pkg/port/lwip branch from a5b8066 to 2da31a5 Compare March 30, 2016 14:42

miri64 force-pushed the pkg/port/lwip branch from 2da31a5 to 9f8bbc8 Compare March 30, 2016 15:00

miri64 added 3 commits March 30, 2016 17:35

lwip: initial import of conn_ip wrapper

410f44f

lwip: initial import of conn_udp wrapper

5431df6

tests: intial import of lwIP test application

9b8417f

miri64 force-pushed the pkg/port/lwip branch from 9f8bbc8 to 9b8417f Compare March 30, 2016 15:35

miri64 merged commit 4dbc26f into RIOT-OS:master Mar 30, 2016

miri64 deleted the pkg/port/lwip branch March 30, 2016 16:22

Conversation

miri64 commented Aug 3, 2015

Uh oh!

jnohlgard commented Aug 3, 2015

Uh oh!

miri64 commented Aug 3, 2015

Uh oh!

miri64 commented Oct 25, 2015

Uh oh!

miri64 commented Nov 24, 2015

Uh oh!

miri64 commented Nov 30, 2015

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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

miri64 commented Nov 30, 2015

Uh oh!

miri64 commented Nov 30, 2015

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016

Uh oh!

OlegHahm commented Mar 30, 2016

Uh oh!

miri64 commented Mar 30, 2016