There is a flag called CPU_FREQ_STAT, which

exports CPU frequency statistics information through sysfs
      file system.

More info available here: cpufreq driver kconfig

Then, one can guess constant I/O operation due to the statistic information exporting can cause memory overhead or any decrease of the performance and battery life.

Would this affirmation be correct? If not, why?

1 Answer 1


sysfs is a virtual filesystem. It doesn't really exist on disk, so there is no (disk) I/O. Nor is there any I/O at all, even virtual, except when something reads the file. It's just a kernel API that's being exposed to userspace through open/read/write/close instead of, e.g., adding another syscall.

There is probably a tiny overhead. It surely takes a trivial bit of memory to hold the counters, a trivial amount of CPU time to update them, and increases the kernel image size by a trivial amount.

OTOH, if frequency scaling is used on your machine, turning that off will greatly reduce your ability to investigate its behavior—and lowering CPU frequency at the right time typically has a major effect on both performance and battery life.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .