core: mbox: initial commit by kaspar030 · Pull Request #4919 · RIOT-OS/RIOT

kaspar030 · 2016-02-28T13:19:33Z

Alternative to #4552.

differences:

a quarter of the code size
more than two times faster

WIP:

documentation missing

(~~timeout functions missing~~) postponed

miri64 · 2016-02-28T14:07:08Z

Not tested, but at least code-wise it looks simpler then my implementation. However: I would prefer putting the non-mbox related changes to core into a separate commit.

OlegHahm · 2016-02-28T14:27:14Z

This is based on #4557, right?

kaspar030 · 2016-02-28T15:28:13Z

This is based on #4557, right?

Ah, yes...

OlegHahm · 2016-02-29T15:30:30Z

core/include/mbox.h

+};
+
+void mbox_put(mbox_t *mbox, msg_t *msg);
+int mbox_tryput(mbox_t *mbox, msg_t *msg);


For API consistency: mbox_try_put()

miri64 · 2016-03-06T20:17:46Z

I really would like to have this soonish (though lwIP is already winning time-wise despite my crappy-mbox API :P)

kaspar030 · 2016-03-14T10:18:10Z

@authmillenon Sorry, I found a flaw in the logic. Will have to fix that first.

miri64 · 2016-03-14T10:20:54Z

Kk. Thesis evaluation is done anyways. Mentioned this PR as an optimization possibility, but used my mbox implementation for the actual comparison.

miri64 · 2016-03-31T16:05:26Z

Fixes #4342

miri64 · 2016-03-31T16:05:36Z

(please rebase)

kaspar030 · 2016-04-10T00:21:36Z

rebased

I un-marked WIP as it should be usable now. Let's add timeout versions later.

kaspar030 · 2016-04-10T12:17:31Z

As this is very unintrusive, we could think about merging for the release, marking it as experimental?

miri64 · 2016-06-07T10:13:42Z

core/include/thread.h

-#if defined(MODULE_CORE_MSG) || defined(MODULE_CORE_THREAD_FLAGS)
-    void *wait_data;                /**< used by msg and thread flags   */
+#if defined(MODULE_CORE_MSG) || defined(MODULE_CORE_THREAD_FLAGS) \
+    || defined(MODULE_MBOX)


How does that module gets defined?

nowhere. fixed.

miri64 · 2016-06-10T20:49:24Z

core/include/mbox.h

+#endif
+
+/** Static initializer for mbox objects */
+#define MBOX_INIT(queue, queue_size) {{0}, {0}, CIB_INIT(queue_size), queue, NULL}


Also needs a dynamic initializer e.g.:

static inline void mbox_init(mbox_t *mbox, msg_t queue, unsigned int queue_size) { mbox_t m = MBOX_INIT(queue, queue_size); *mbox = m; }

so you can (re-)initialize the variable at other points than variable definition

kaspar030 · 2016-07-14T12:54:26Z

addressed comments
rebased

emmanuelsearch · 2016-07-14T13:08:14Z

@miri64 ACK still stands?

miri64 · 2016-07-14T13:16:18Z

core/include/mbox.h

+
+#ifndef MODULE_CORE_MBOX
+#error mbox depends on the module core_mbox!
+#endif


IMHO this is superfluous. The linker will yield an error anyway.

miri64 · 2016-07-14T13:16:53Z

@miri64 ACK still stands?

Apart from comment above, yes :-)

kaspar030 · 2016-07-14T13:31:31Z

addressed & squashed

emmanuelsearch · 2016-07-14T13:48:06Z

Looks good from what I can see.
It does not break anything, since it's new feature.
So nothing speaks against an ACK.
Hence: ACK

miri64 · 2016-07-14T15:20:16Z

AVR and MSP430 are butthurt about something.

miri64 · 2016-07-18T14:00:11Z

Ping?

kaspar030 · 2016-07-18T14:20:43Z

Hm, that seems to be a compiler problem. My arch avr-gcc (5.3.0) compiles the code just fine, the one used by murdock doesn't.

kaspar030 · 2016-07-18T14:21:23Z

rebased

kaspar030 · 2016-07-18T15:04:00Z

Turns out there was an excess initializer. Makes me wonder why the other compilers didn't choke on that.

amended a fix

kaspar030 · 2016-07-18T15:55:58Z

two ACKs -> go

kaspar030 added State: WIP State: The PR is still work-in-progress and its code is not in its final presentable form yet Area: core Area: RIOT kernel. Handle PRs marked with this with care! Type: new feature The issue requests / The PR implemements a new feature for RIOT labels Feb 28, 2016

kaspar030 mentioned this pull request Feb 28, 2016

mbox: initial import of a thread-independent IPC #4552

Closed

OlegHahm added the State: waiting for other PR State: The PR requires another PR to be merged first label Feb 28, 2016

miri64 mentioned this pull request Feb 29, 2016

core: make messaging optional #3851

Merged

OlegHahm reviewed Feb 29, 2016
View reviewed changes

OlegHahm assigned miri64 Mar 1, 2016

miri64 mentioned this pull request Mar 8, 2016

doc: conn_udp: document that send and receives needs to be called from the same thread #4993

Closed

miri64 added this to the Release 2016.07 milestone Mar 23, 2016

miri64 mentioned this pull request Mar 31, 2016

[RFC] thread-decoupled messaging #4342

Closed

kaspar030 force-pushed the mbox branch from 4dbb917 to 3c54b97 Compare April 10, 2016 00:04

kaspar030 removed State: WIP State: The PR is still work-in-progress and its code is not in its final presentable form yet State: waiting for other PR State: The PR requires another PR to be merged first labels Apr 10, 2016

kaspar030 added CI: ready for build If set, CI server will compile all applications for all available boards for the labeled PR CI: needs squashing Commits in this PR need to be squashed; If set, CI systems will mark this PR as unmergable labels Apr 10, 2016

kaspar030 force-pushed the mbox branch from bde68a6 to 99295bb Compare April 10, 2016 11:16

miri64 assigned Kijewski, miri64 and OlegHahm and unassigned Kijewski and OlegHahm Jun 6, 2016

miri64 mentioned this pull request Jun 7, 2016

gnrc_netreg: introduce helper macro/function for entry init #5524

Merged

miri64 reviewed Jun 7, 2016
View reviewed changes

miri64 mentioned this pull request Jun 7, 2016

gnrc_netreg: mbox and arbitrary callback support #5526

Merged

miri64 reviewed Jun 10, 2016
View reviewed changes

kaspar030 force-pushed the mbox branch from a1c9ed7 to 0839972 Compare July 14, 2016 12:53

miri64 reviewed Jul 14, 2016
View reviewed changes

kaspar030 force-pushed the mbox branch from 0839972 to 95affb2 Compare July 14, 2016 13:25

cib: add unsafe operations

9bfea94

kaspar030 force-pushed the mbox branch from 95affb2 to 8e9c5f7 Compare July 18, 2016 14:21

core: mbox: introduce thread decoupled message queues

3a6f950

kaspar030 force-pushed the mbox branch from 8e9c5f7 to 3a6f950 Compare July 18, 2016 15:03

kaspar030 merged commit 66710ef into RIOT-OS:master Jul 18, 2016

kaspar030 deleted the mbox branch July 18, 2016 15:56

miri64 mentioned this pull request Feb 14, 2017

ps: add state name for STATUS_MBOX_BLOCKED #6601

Merged

Conversation

kaspar030 commented Feb 28, 2016

Uh oh!

miri64 commented Feb 28, 2016

Uh oh!

OlegHahm commented Feb 28, 2016

Uh oh!

kaspar030 commented Feb 28, 2016

Uh oh!

OlegHahm Feb 29, 2016

Choose a reason for hiding this comment

Uh oh!

miri64 commented Mar 6, 2016

Uh oh!

kaspar030 commented Mar 14, 2016

Uh oh!

miri64 commented Mar 14, 2016

Uh oh!

miri64 commented Mar 31, 2016

Uh oh!

miri64 commented Mar 31, 2016

Uh oh!

kaspar030 commented Apr 10, 2016

Uh oh!

kaspar030 commented Apr 10, 2016

Uh oh!

miri64 Jun 7, 2016

Choose a reason for hiding this comment

Uh oh!

kaspar030 Jul 14, 2016

Choose a reason for hiding this comment

Uh oh!

miri64 Jun 10, 2016

Choose a reason for hiding this comment

Uh oh!

kaspar030 Jul 14, 2016

Choose a reason for hiding this comment

Uh oh!

kaspar030 commented Jul 14, 2016

Uh oh!

emmanuelsearch commented Jul 14, 2016

Uh oh!

miri64 Jul 14, 2016

Choose a reason for hiding this comment

Uh oh!

miri64 commented Jul 14, 2016

Uh oh!

kaspar030 commented Jul 14, 2016

Uh oh!

emmanuelsearch commented Jul 14, 2016

Uh oh!

miri64 commented Jul 14, 2016

Uh oh!

miri64 commented Jul 18, 2016

Uh oh!

kaspar030 commented Jul 18, 2016

Uh oh!

kaspar030 commented Jul 18, 2016

Uh oh!

kaspar030 commented Jul 18, 2016

Uh oh!

kaspar030 commented Jul 18, 2016

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants