This is the seventh post in a weekly series tracking the development of Tock, a safe multi-tasking operating system for microcontrollers.
The second revision of Imix board are back from the factory with a bunch of hardware fixes. We are working out some final hardware nits, writing drivers for the sensors (which are slightly different than the ones on the Firestorm) and working out the details with Stanford administration of how to distribute them. But everything is on track to be released by early November.
We made inroads on updates to the HIL traits.
@bradjc’s I2C HIL has progressed significantly and only waiting on a few fixes to the I2C slave implementation to merge.
@bradjc added support for using GPIO as a chip select on the SAM4L. This ended requiring a change to the SPI HIL where the the chip select type is associated to the SPI trait (blog post forthcoming just for that? Yes, I think so).
@KMarshland implemented the Alarm trait for the NRF51’s RTC controller which, unlike the Timer controller, can operate while the core is asleep.
@amilich implemented the UART trait for the NRF51.
The result of both of these changes is that the NRF51 can successfully run a base set of applications for on Tock.
@alevy refactored libtock to remove board specific dependencies as much as possible, separating driver interfaces into separate modules.
@ppannuto made aesthetic improvements to the userland make system as well as functional changes that make sure changes are tracked more accurately.
Interprocess communication (IPC)
We’ve begun prototyping an IPC system that would allow processes to expose driver-like functionality. @alevy has implemented a first version and described it on the mailing list. The IPC system leverages the same system calls as other drivers in the kernel, with some additional restrictions due to size and alignment requirements of the ARM MPU.