[lkml]   [2008]   [Apr]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [-mm] Add an owner to the mm_struct (v8)
Paul Menage wrote:
> On Sat, Apr 5, 2008 at 7:47 AM, Balbir Singh <> wrote:
>> Repeating my question earlier
>> Can we delay setting task->cgroups = &init_css_set for the group_leader, until
>> all threads have exited?
> Potentially, yes. It also might make more sense to move the
> exit_cgroup() for all threads to a later point rather than special
> case delayed group leaders.

Yes, that makes sense. I think that patch should be independent of this one
though? What do you think?

>> If the user is unable to remove a cgroup node, it will
>> be due a valid reason, the group_leader is still around, since the threads are
>> still around. The user in that case should wait for notify_on_release.
>> >
>> > To me, it seems that setting up a *virtual address space* cgroup
>> > hierarchy and then putting half your threads in one group and half in
>> > the another is asking for trouble. We need to not break in that
>> > situation, but I'm not sure it's a case to optimize for.
>> That could potentially happen, if the virtual address space cgroup and cpu
>> control cgroup were bound together in the same hierarchy by the sysadmin.
> Yes, I agree it could potentially happen. But it seems like a strange
> thing to do if you're planning to be not have the same groupings for
> cpu and va.

It's easier to set it up that way. Usually the end user gets the same SLA for
memory, CPU and other resources, so it makes sense to bind the controllers together.

>> I measured the overhead of removing the delay_group_leader optimization and
>> found a 4% impact on throughput (with volanomark, that is one of the
>> multi-threaded benchmarks I know of).
> Interesting, I thought (although I've never actually looked at the
> code) that volanomark was more of a scheduling benchmark than a
> process start/exit benchmark. How frequently does it have processes
> (not threads) exiting?

I could not find any other interesting benchmark for benchmarking fork/exits. I
know that volanomark is heavily threaded, so I used it. The threads quickly exit
after processing the messages, I thought that would be a good test to see the

> How many runs was that over? Ingo's recently posted volanomark tests
> against -rc7 showed ~3% random variation between runs.

I ran the test four times. I took the average of runs, I did see some variation
between runs, I did not calculate the standard deviation.

Warm Regards,
Balbir Singh
Linux Technology Center

 \ /
  Last update: 2008-04-05 19:51    [from the cache]
©2003-2014 Jasper Spaans. hosted at Digital Ocean