On Tue 24-05-16 11:43:19, Vladimir Davydov wrote: > On Mon, May 23, 2016 at 07:44:43PM +0200, Michal Hocko wrote: > > On Mon 23-05-16 19:02:10, Vladimir Davydov wrote: > > > mem_cgroup_oom may be invoked multiple times while a process is handling > > > a page fault, in which case current->memcg_in_oom will be overwritten > > > leaking the previously taken css reference. > > > > Have you seen this happening? I was under impression that the page fault > > paths that have oom enabled will not retry allocations. > > filemap_fault will, for readahead. I thought that the readahead is __GFP_NORETRY so we do not trigger OOM killer. > This is rather unlikely, just like the whole oom scenario, so I haven't > faced this leak in production yet, although it's pretty easy to > reproduce using a contrived test. However, even if this leak happened on > my host, I would probably not notice, because currently we have no clear > means of catching css leaks. I'm thinking about adding a file to debugfs > containing brief information about all memory cgroups, including dead > ones, so that we could at least see how many dead memory cgroups are > dangling out there. Yeah, debugfs interface would make some sense. -- Michal Hocko SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>