Andrzej Kaczmarek
Andrzej Kaczmarek
You need to use latest newt tool (see https://github.com/apache/mynewt-newt/pull/423).
Perhaps someone has some better idea how to fix this... :)
tbh I do not really see any relation between mbox size and timer interrupt latency. Larger c2s mailbox means it's less likely that CMAC will spin waiting for free space...
So the only problem I could think of is reading s2c mailbox in an interrupt context causes some extra delays, although as I wrote above I think it should not...
One more thing to check: set only c2s mbox size to 1024 and leave s2c on default value. Large c2s mbox size guarantees that data can be pushed from CMAC...
@wes3 so could be that I had a hunch that larger s2c mbox size may introduce some timing issues, but tbh I do not recall what was my reasoning in...
I don't think we need this, there were significant optimizations done in CMAC code that seem to resolve pretty much all known issues, at least those I was aware of.
What configuration you use? Can you include some of the errors you see? I've just built host+controller with `-std=c99` without errors...
rebased on top of https://github.com/apache/mynewt-nimble/pull/1334
BSP is definitely not a good place to have this configured - it should be in some common package for MCU as otherwise syscfg has to be defined for each...