Linux Filesystem, Storage, and Memory Management Summit, Day 2

There were a couple aspects to that work. One is to reduce mmap_sem hold times - I have done some work in that area, which got in 2.6.37 and 2.6.38. There are still a few situations where mmap_sem is held while we want for a disk access, but it's much less common than it was last year.

The other direction Nick proposed last year was to have smaller granularity for the mmap_sem - AFAIK there has been no major work in that direction.