cgroup's support for IO resource provision was missing a major piece - the block layer couldn't tell to which cgroup a given writeback IO should be attributed. All non-direct write IOs generated through write(2) and dirtied mmap pages were charged against the root cgroup defeating the purpose of IO provisioning. cgroup is finally growing writeback support which will not only improve IO provisioning but also make memory provisioning behave correctly with respect to the IO pressure in the cgroup. This presentation examines what was broken and how the new cgroup writeback support closes the gap.
Tejun has been working on various aspects of Linux kernel since 2005 and is currently maintaining percpu memory allocator, control groups, and workqueue. He currently works as a software engineer for Red Hat.