[RFC PATCH 0/5] cputime: Generic virtual based cputime accounting v2

2012-08-03 Thread Frederic Weisbecker
Hi, The changes since v1 (https://lkml.org/lkml/2012/7/27/286) are: - change naming from "user hooks" to "code domain tracking". - use of __this_cpu_* APIs instead of __get_cpu_* There is still a dependency against Paul's rcu/idle branch and the vtime consolidation set:

[RFC PATCH 0/5] cputime: Generic virtual based cputime accounting v2

2012-08-03 Thread Frederic Weisbecker
Hi, The changes since v1 (https://lkml.org/lkml/2012/7/27/286) are: - change naming from user hooks to code domain tracking. - use of __this_cpu_* APIs instead of __get_cpu_* There is still a dependency against Paul's rcu/idle branch and the vtime consolidation set:

[RFC PATCH 0/5] cputime: Generic virtual based cputime accounting

2012-07-27 Thread Frederic Weisbecker
So, This is a proposition to handle the cputime accounting without the tick to prepare for stopping the tick further idle. I have managed to reuse and generalize the kernel/user boundary hooks used by RCU. This way we can minimize the changes in archs that don't support virtual cputime and we

[RFC PATCH 0/5] cputime: Generic virtual based cputime accounting

2012-07-27 Thread Frederic Weisbecker
So, This is a proposition to handle the cputime accounting without the tick to prepare for stopping the tick further idle. I have managed to reuse and generalize the kernel/user boundary hooks used by RCU. This way we can minimize the changes in archs that don't support virtual cputime and we