Re: Improve preempt-scheduling and x86 user access v3

2013-08-28 Thread Ingo Molnar
* Andi Kleen wrote: > Various optimizations related to CONFIG_PREEMPT_VOLUNTARY and x86 > uaccess Looks mostly good. Does this patchset change the number of cond_resched() preemption points on CONFIG_PREEMPT_VOLUNTARY=y, or is it a scheduling invariant? Thanks, Ingo -- To

Re: Improve preempt-scheduling and x86 user access v3

2013-08-28 Thread Ingo Molnar
* Andi Kleen a...@firstfloor.org wrote: Various optimizations related to CONFIG_PREEMPT_VOLUNTARY and x86 uaccess Looks mostly good. Does this patchset change the number of cond_resched() preemption points on CONFIG_PREEMPT_VOLUNTARY=y, or is it a scheduling invariant? Thanks,

Improve preempt-scheduling and x86 user access v3

2013-08-16 Thread Andi Kleen
Various optimizations related to CONFIG_PREEMPT_VOLUNTARY and x86 uaccess - Optimize copy_*_inatomic on x86-64 to handle 1-8 bytes without string instructions - Inline might_sleep and other preempt code to optimize various preemption paths This costs about 10k text size, but generates far

Improve preempt-scheduling and x86 user access v3

2013-08-16 Thread Andi Kleen
Various optimizations related to CONFIG_PREEMPT_VOLUNTARY and x86 uaccess - Optimize copy_*_inatomic on x86-64 to handle 1-8 bytes without string instructions - Inline might_sleep and other preempt code to optimize various preemption paths This costs about 10k text size, but generates far