The track will deal with issues around unified memory management for systems (initially felt on ARM-based SoCs) where not all devices have an IOMMU available, and where cache management and cache coherency may be atypical. The effort is underway and is being coordinated by Linaro, with support from component contributors and maintainers. The goal is to provide necessary allocation (e.g., physically contiguous buffers), DMA mapping and buffer sharing services to device drivers with maximum leverage of existing API and code. This track will serve as a check point to review what's been done and what's left to do.
Project page and conference slides are here
Developers from the components, kernel developers, and SoC vendor developers.
The following is an abbreviated list of participants in the mini-summit around this topic in May:
Also invited but unable to travel at the time:
It would be good to have additional interested parties related to graphics and video processing devices.
Proposal added by Jesse Barker jesse.barker@linaro.org