Round 10 (May 2015 to Aug 2015) projects:
Mentor:: Julia Lawall
Coccinelle is a program matching and transformation tool for C code that has been used extensively in contributing to the Linux kernel, for both code evolutions and bug fixes. Coccinelle is driven by specifications, known as semantic patches, that use a notation based on C code, and are this fairly easy to develop. Around 40 semantic patches are included with the Linux kernel source code, in scripts/coccinelle, and are used in the continuous testing service provided by Intel.
The use of out of date APIs is confusing for people who need to understand the code, and requires the maintenance of excess code. The goal of this project is to use Coccinelle to replace out of date API uses in the Linux kernel with more modern equivalents.
Media Controller Virtual Driver
Mentor:: Laurent Pinchart
The media subsystem has a kernel driver that emulates a V4L device in order to allow testing userspace applications without requiring hardware. We are, however, missing a driver that would emulate a Media Controller device. Such a driver would be very useful to test handling of complex media pipelines in userspace applications.
The goal of this project is to create a virtual Media Controller driver. No access to hardware is needed to perform this task, although a USB camera compatible with the USB Video Class could be useful to first experiment with the Media Controller API.
Modernize Linux Wireless drivers
Mentor: Jes Sorensen
The Linux wireless (WiFi) stack has migrated from the old obsolete WEXT API to the modern cfg80211/nl80211 API. The old WEXT API has been marked deprecated for years, but unfortunately a couple of drivers are still using it, making it impossible to remove the old code.
The goal of this project is to complete the conversion of the Orinoco wireless driver from using the WEXT API and make it a good cfg80211 citizen.
Full Dynamic Ticks
CPUs have to be interrupted through periodic timers to initiate tasks such as update of CPUs'load and runtime of tasks, perform preemption of running tasks if necessary and load balancing to even out system wide load. But when CPUs are idle, none of these tasks is required to be performed as there is no load on them. So periodic timer interrupts can be stopped on idle CPUs. This improves power efficiency of the kernel to a good extent. Thus an infrastructure called "Tickless Idle" which disables periodic timer interrupts when CPUs are idle, was added to the 2.6.21 release of the Linux kernel.
However some users required that periodic timer interrupts be disabled even on CPUs running single tasks. Typically high performance workloads (HPC) and real time tasks are performance critical and require to be run undisturbed as far as possible. For the benefit of such tasks, the Full Dynamic Ticks infrastructure was added to the 3.10 release of the Linux kernel. The goal of Full Dynamic Ticks is to enable CPUs to run in tickless mode as long as they are running single tasks. As a consequence it manages to isolate such CPUs to run performance critical user tasks without interruptions from the kernel. Over the months much effort has gone into making this stable. However there is more work to be done.
The goal of this project is to tackle specific issues that are currently in way of running CPUs in Full Dynamic Tick mode effectively.
IIO dummy driver
The Industrial I/O subsystem is intended to provide support for devices that in some sense are analog to digital or digital to analog converters. Some devices that fall in this category are: accelerometers, gyroscopes, light sensors, etc.
Linux staging already holds a dummy IIO driver which is very useful as documentation on how to write a driver. Also it is very useful for testing user space code and rapidly prototyping new interfaces.
The goal of this project is to move IIO dummy driver out of staging. The major challenge in order to make this ready is interrupt emulation. The way evgen module generates interrupts greatly restricts their use and is far from clean. There is also work on the device creation part. Current approach where we use a module parameter controlling how many instances are created it's pretty ugly. Last thing we need to add support for more devices, allowing userspace applications to do smoke testing without have the actual hardware.
If you like this project and want to apply make sure you check the tasks page.