Gpio Interrupt fix for the CC2538

[AnonMall]

Fixed the issue where the gpio interrupt is not properly acknowledged.

[smlng]

please squash (and rebase if necessary?), otherwise ACK!

[aabadie]

Does this PR provides a fix for #6650 ? Then it should be merged asap.

[smlng]

partly - the issue is more complex, #6773 provides the other part, but is more a workaround than a fix. The problem is that the gpio periph of cc2538 was partly adapted to periph_gpio driver of RIOT, but i2c, spi, and uart are still using the gpio driver from TI. Hence, rather than applying #6773 workaround, I'd work on rewritting the named 3 periph drivers ...

Nevertheless, this PR fixes an issue when using interrupts on gpios

[smlng]

But still: DO NOT MERGE THIS, just to have in the release. We had lots of trouble lately with prematurely merged periph drivers - this need very careful testing!

[smlng]

needs testing, DO NOT MERGE!

[aabadie]

DO NOT MERGE!

no problem, it won't be in the release if it's not ACK'ed, merged and backported. Keep calm ;)

[AnonMall]

sorry i am a little new to squashing... just fixing some things. Please bear with me.

[smlng]

take your time, consider the following:

1. setup RIOT-OS/RIOT as an upstream to your repo, see here
2. update your master to upstream master:

git fetch upstream
git checkout master
git pull --rebase upstream master

3. now update your branch and push (force, is needed as you rewrite history of your branch):

git checkout  gpioIntFix
git rebase master
git push -f

4. in case you have commit you want to squash use interactive rebase, see here too

git rebase -i master

[miri64]

In fact you can just git fetch upstream and rebase to upstream/master (git rebase upstream/master or git rebase -i upstream/master for interactive mode). edit: shouldn't spewing improvements full of errors themselves ^^"

[AnonMall]

fnished rebase and should be squashed... as it is only one file, further rebasing shouldn't be nescessary i guess. Thank you both for helping out.

[smlng]

perfect! I'll test.

[smlng]

tested this PR on board remote-revb and it fixes the following issue present on master:

• GPIO init as interrupt works as expected
• but when interrupt fires, it does not return correctly
• for tests/periph_gpio this means no shell prompt after interrupt

With this PR interrupt handling returns and shell prompt is available again.

Nice job @AnonMall, thanks! ACK - and GO as soon as Murdock is happy.

[miri64]

Maybe, if it fixes some stuff we should reconsider backporting into the release. What do you think @aabadie?

[aabadie]

I think it should be backported (since the mentioned bugs are quite important)

[smlng]

@AnonMall can you provide the backport? Basically make a copy of the branch and rebase onto tag 2017.07-RC1, then make a new pull request against 2017.07-branch (instead of master).

[edit]: and prefix the PR title with [backport] to make that clear.

[AnonMall]

I can do that. But that also means you want me to rebase it against 2017.07-branch instead of master right?

[AnonMall]

Okay done. #7314