To help a customer that complained about the time it takes for df to check one of their machines with nearly 22000 lines in /proc/self/mountinfo, my collegue came up with this patch that uses a hash table for seaching in filter_mount_list() and get_dev(). Data from said machine:
Without patch: real 0m2.273s user 0m0.768s sys 0m1.500s real 0m1.731s user 0m0.532s sys 0m1.188s With patch: real 0m1.286s user 0m0.052s sys 0m1.232s real 0m1.066s user 0m0.028s sys 0m1.032s Is this patch acceptable? Philipp
