search

ARMmbed / core-util

DEPRECATED: Mbed 3 utilities library
Other
12 stars 17 forks source link

Move IRQ state management into mbed-util #25

Closed bremoran closed 8 years ago

bremoran commented 8 years ago

This should address the PR in https://github.com/ARMmbed/core-util/pull/23 is a more reusable way. Now, minar should be able to use its IRQ disable/enable logic from this source.

0xc0170 commented 8 years ago

LGTM

@rgrover happy as a replacement for #23?

A side question - Shall we fix via separate PR - mbed namespace in this repository?

bremoran commented 8 years ago

@rgrover Should be fixed in latest commit.

rgrover commented 8 years ago

is irq_state the right name for these files? how about irq.[ch]? or interrupt.[ch]?

rgrover commented 8 years ago

why do we have CriticalSectionLock and irq_state as separate concepts? what has been achieved here in addition to the already pending pull request: https://github.com/ARMmbed/core-util/pull/23

bremoran commented 8 years ago

@rgrover This PR is explicitly to remove duplicated code. The CriticalSectionLock you proposed had code duplicated from minar-platform. irq_state moves the minar-platform code into a space that can be used by both minar and CriticalSectionLock.

Once this has been approved, the next step will be a patch to minar which uses the irq_state. Following that, I will push a minor revision to minar-platform which removes the irq_state APIs.

rgrover commented 8 years ago

why don't we consolidate irq_state into CriticalSectionLock? why are they separate concepts? Isn't CriticalSectionLock the more generic concept?

bremoran commented 8 years ago

@rgrover Feel free to file a PR for that.

rgrover commented 8 years ago

@bremoran why are you introducing irq_state here? Why not leave this within CriticalSectionLock as is being suggested by pull #23?

hugovincent commented 8 years ago

+1 to consolidation.

bremoran commented 8 years ago

@rgrover Please see my comment here: https://github.com/ARMmbed/core-util/pull/23#issuecomment-141401754

Are you planning to rewrite minar to use CriticalSectionLock?

bremoran commented 8 years ago

@hugovincent -1 for consolidation. It doesn't make sense to me to force users into using an object oriented model that depends on object lifetimes.

hugovincent commented 8 years ago

@bremoran I'm not saying we should necessarily consolidate on the thing you don't like, I'm saying we shouldn't have two almost identical mechanisms with different semantics. It might be that to consolidate we have to build a third mechanism and then replace the first two.

bremoran commented 8 years ago

@hugovincent I think you misunderstand me. I'm saying that the CriticalSectionLock has some nice semantics for certain use-cases. I also can see how a procedural API has nice semantics for other use-cases. What I don't see is a good reason not to have both APIs available.