Re: Rough notes from sys_membarrier() lightning BoF

From	Peter Zijlstra <peterz@infradead.org>
Newsgroups	linux.kernel
Subject	Re: Rough notes from sys_membarrier() lightning BoF
Date	2017-09-21 15:20 +0200
Message-ID	<us9yy-1tk-19@gated-at.bofh.it> (permalink)
References	<uqQoh-5MT-3@gated-at.bofh.it> <urPJw-5ky-7@gated-at.bofh.it> <urRLk-6BR-7@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

On Wed, Sep 20, 2017 at 06:13:50PM +0000, Mathieu Desnoyers wrote:

> > Also, can you elaborate on the PPC issue?  PPC appears to track
> > mm_cpumask more or less just like x86.  Is the issue just that this
> > tracking has no implied barriers?  If so, how does TLB flush on ppc
> > work?  It really does seem impressive to me that an architecture can
> > efficiently support munmap() but not an expedited private membarrier.
> 
> I'll leave this question to the PPC experts :)

IIRC PPC does not keep a tight mm_cpumask, it only sets bit, it never
clears bits. The atomic op required to set bits does not imply any
memory barrier on PPC.

TLB invalidation is a TLBI instruction, it sends TLBI broadcast packets
over the interconnect, it doesn't require IPIs like x86.

The only optimization PPC does is that if the mm_cpumask has only a
single bit set, it uses a TLBI instruction without broadcast, which is
cheaper.

Thread

Rough notes from sys_membarrier() lightning BoF "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-09-18 00:40 +0200
  Re: Rough notes from sys_membarrier() lightning BoF Alan Stern <stern@rowland.harvard.edu> - 2017-09-18 21:10 +0200
    Re: Rough notes from sys_membarrier() lightning BoF Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-09-18 21:10 +0200
    Re: Rough notes from sys_membarrier() lightning BoF "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-09-18 21:30 +0200
      Re: Rough notes from sys_membarrier() lightning BoF Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-09-18 21:40 +0200
        Re: Rough notes from sys_membarrier() lightning BoF "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2017-09-18 22:40 +0200
  Re: Rough notes from sys_membarrier() lightning BoF Andy Lutomirski <luto@kernel.org> - 2017-09-20 18:10 +0200
    Re: Rough notes from sys_membarrier() lightning BoF Andy Lutomirski <luto@kernel.org> - 2017-09-20 20:20 +0200
      Re: Rough notes from sys_membarrier() lightning BoF Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-09-20 22:00 +0200
    Re: Rough notes from sys_membarrier() lightning BoF Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-09-20 20:20 +0200
      Re: Rough notes from sys_membarrier() lightning BoF Peter Zijlstra <peterz@infradead.org> - 2017-09-21 15:20 +0200
        Re: Rough notes from sys_membarrier() lightning BoF Mathieu Desnoyers <mathieu.desnoyers@efficios.com> - 2017-09-21 20:10 +0200
      Re: Rough notes from sys_membarrier() lightning BoF Peter Zijlstra <peterz@infradead.org> - 2017-09-21 15:20 +0200
        Re: Rough notes from sys_membarrier() lightning BoF James Bottomley <James.Bottomley@HansenPartnership.com> - 2017-09-21 19:30 +0200
          Re: Rough notes from sys_membarrier() lightning BoF Peter Zijlstra <peterz@infradead.org> - 2017-09-22 11:40 +0200
        Re: Rough notes from sys_membarrier() lightning BoF Michael Ellerman <mpe@ellerman.id.au> - 2017-09-22 07:10 +0200

csiph-web