On Fri, Feb 6, 2015 at 6:25 AM, Karanbir Singh mail-lists@karan.org wrote:
Hi,
lots of people in various SIGs and user side projects are keen on bringing up a mainline kernel, that tracks upstream closer than the distro kernel does.
Kernel integration and maintenance of someone else's code tree is a lot of work. Been there, done that, have the scar tissue and the forced resignation of a kernel team. I have stories about that, but the team was basically backporting desired changes from newer kernels into their "stable" old kernel. It was the equivalent of shoving RHEL 6 kernels into an RHEL 7 operating system for "stability" and to "perserve optimizatoins" Anyway: this kind of effort can get very tricky, very fast, when related dependencies get involved. This is especially true for filesystems and hardware integration. And the resources needed can be very difficult to predict, especially when kernel compilation is sensitive to gcc and glibc and mkfs and NFS and dbus and USB and audio, etc., etc.
Kernel development also badly needs test hardware, at least for regression testing. Where would this come from?