In Linux it is often said that every device is a file. To most of us a file is something just there. The kernel boys and girls actually tweak the bits right down to the hardware to make the magic of a file somewhere becoming accessible to the system, mounting the file.
Last week, one developer found a recent version of Linux mounted files considerably slower than it usually did. He noticed because he mounted a lot of file-systems. When careful timing was done the difference was measured. Indeed mounts were an order of magnitude slower. By repeating the measurements for several versions and finally versions with and without certain patches/changes, the cause was found. Then discussion broke out about how to fix both issues, why the change had been made and how to do it differently.
In the process, coach Linus stepped in having noticed some perspective was missing. He quickly turned the pack around to solve the “real problem” and Linux now is as fast as ever and other problems have disappeared. Yeah team!
“It seems the recent kernel is slower mounting hard disk than older kernels.”
Indeed, GNU/Linux is a cooperative product of the world with everyone contributing what they can and getting the job done. It’s the right way to do IT. In fact, I rarely mount much because I use my system for hours and only mount between session but I think it’s important that someone actually cares if IT slows down for any reason, unlike that other OS when M$ actually benefits when their OS slows down through use and folks buy a new one…
BTW, “here’s the comparison …
0.013s 3.7.0-rc7 + synchronize_sched_expedited()
0.350s 3.7.0-rc7 + Oleg’s patch.”