Linux Filesystem, Storage, and Memory Management Summit, Day 1 [LWN.net]

Linux Filesystem, Storage, and Memory Management Summit, Day 1

Posted Apr 5, 2011 15:30 UTC (Tue) by willy (subscriber, #9762)
Parent article: Linux Filesystem, Storage, and Memory Management Summit, Day 1

> The struct mem_cgroup pointer, he things, is excessive; there will usually be a bunch of pages from a single file used in any given control group. So what is really needed is a separate structure to map from a control group to the address_space structure representing the backing store for a set of pages. Ideally, he would point to that structure (instead of struct address_space) in struct page, but that would require some filesystem API changes.

When I initially read this, I thought "Why not extend the address_space?"
I just clarified this with Johannes. There may be multiple cgroups within a single address space, so he's looking to introduce an additional indirection.

Seems fair enough to me.

(Log in to post comments)

Linux Filesystem, Storage, and Memory Management Summit, Day 1

Posted Apr 5, 2011 19:56 UTC (Tue) by hnaz (subscriber, #67104) [Link]

It would essentially describe one specific combination of an inode and a memory cgroup.

We want these objects also for another reason: memcg writeback. We can string them up and have a list of inodes of interest when pushing back the dirty limit of a specific memory cgroup.