[Ksummit-2013-discuss] [ATTEND] insane cgroups, unfair rwlocks
Michel Lespinasse
walken at google.com
Sat Jul 20 02:07:11 UTC 2013
I would like to attend this year's Kernel Summit in Edinburgh.
Among the topics I would be interested to discuss:
(1) insane cgroups
This topic has been separately proposed by Jiri Kosina and Li Zefan already.
Performance isolation between processes running on a single machine is
one of the top requirements we have at Google. We are very concerned
that cgroup maintainers are threatening to simplify out of existence
some of the cgroups features we use. To be clear, our main concern is
not about interface changes per se - we could deal with simplified
interfaces if they retained their power - but with the fact that the
proposed new interfaces just can't do some of the things we are
currently using the old interfaces for. I think this should be
discussed at the kernel summit and I very much want Google to be
represented in this.
(2) unfair rwlocks
This is connected to Srivatsa S. Bhat's proposed topic, which he
thoughtfully copied me in (thanks!).
We sometimes hit issues that can be traced down to rwlock unfairness
(mostly with tasklist_lock); I have been taking an interest towards
solving these. Recently there has been more activity than usual about
locking schemes and I have been trying to steer people away from
introducing new unfair locks where possible. I'm actually not entirely
sure if this warrants a KS discussion (as I understand KS nowadays is
more about management than technical issues); however if such
discussions do take place at KS I would like to participate (and if
not, I am always interested to discuss this informally).
More general qualifications: I work in Google's kernel memory
management team. I tend to be active upstream mostly in memory
management areas, though I have also been participating in some
locking discussions over the past 6 months. I'm also currently in
charge of rebasing Google's kernel to a more recent upstream kernel
version, which has been taking most of my time over the last couple
months (so I have to admit I haven't been as active upstream as I
would like to).
--
Michel "Walken" Lespinasse
A program is never fully debugged until the last user dies.
More information about the Ksummit-2013-discuss
mailing list