From mboxrd@z Thu Jan 1 00:00:00 1970 From: Frederic Weisbecker Subject: Re: [GIT PULL v2] timers updates for 3.15 Date: Thu, 13 Mar 2014 16:33:04 +0100 Message-ID: <20140313153301.GA25296@localhost.localdomain> References: <1394642174-14315-1-git-send-email-fweisbec@gmail.com> <20140313110510.GA10424@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mail-ea0-f179.google.com ([209.85.215.179]:40078 "EHLO mail-ea0-f179.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753632AbaCMPdJ (ORCPT ); Thu, 13 Mar 2014 11:33:09 -0400 Content-Disposition: inline In-Reply-To: <20140313110510.GA10424@gmail.com> Sender: linux-arch-owner@vger.kernel.org List-ID: To: Ingo Molnar Cc: Thomas Gleixner , LKML , Archs , Marcelo Tosatti , Peter Zijlstra , Rik van Riel , Huiqingding On Thu, Mar 13, 2014 at 12:05:10PM +0100, Ingo Molnar wrote: > > * Frederic Weisbecker wrote: > > > Ingo, Thomas, > > > > Please pull the timers/cputime-fix-steal-v3 branch that can be found at: > > > > git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git > > timers/cputime-fix-steal-v3 > > > > Only the last patch changed to address Peterz concern on arch Kbuild > > layout. This got sorted out by running the script from b119fa61d440f > > to alpabetically sort the entries. > > > > --- > > The main purpose of this set is to fix a bug on full dynticks configs > > where steal time accounting appears to be zero in /proc/stat even > > after a few seconds of competing guests running busy loops in a same > > host CPU. It's not a regression though as it was there since the > > beginning with full dynticks. > > > > So patch [4/6] ("cputime: Fix jiffies based cputime assumption on > > steal accounting") is the most important patch of the series. The rest > > is mostly preparatory work to fix the bug plus various cleanups. > > > > Thanks, > > Frederic > > --- > > > > Frederic Weisbecker (6): > > cputime: Fix nsecs_to_cputime() return type cast > > cputime: Default implementation of nsecs -> cputime conversion > > cputime: Bring cputime -> nsecs conversion > > cputime: Fix jiffies based cputime assumption on steal accounting > > sched: Remove needless round trip nsecs <-> tick conversion of steal time > > arch: Remove stub cputime.h headers > > > > > > arch/alpha/include/asm/Kbuild | 7 ++++--- > > arch/alpha/include/asm/cputime.h | 6 ------ > > arch/cris/include/asm/Kbuild | 3 ++- > > arch/cris/include/asm/cputime.h | 6 ------ > > arch/frv/include/asm/Kbuild | 5 +++-- > > arch/frv/include/asm/cputime.h | 6 ------ > > arch/m32r/include/asm/Kbuild | 5 +++-- > > arch/m32r/include/asm/cputime.h | 6 ------ > > arch/microblaze/include/asm/Kbuild | 5 +++-- > > arch/microblaze/include/asm/cputime.h | 1 - > > arch/mn10300/include/asm/Kbuild | 3 ++- > > arch/mn10300/include/asm/cputime.h | 1 - > > arch/score/include/asm/Kbuild | 4 ++-- > > arch/score/include/asm/cputime.h | 6 ------ > > arch/x86/include/asm/Kbuild | 1 + > > arch/x86/include/asm/cputime.h | 1 - > > Does this tree also address the heavy conflicts with tip:core/locking? Ah I didn't hear about that. Thomas told me there was a conflict issue but he couldn't tell me more about it. > If not then we should probably carry this in a separate branch that > merges tip:core/locking and tip:timers/core. So I propose you something even more simple. The choice of tip:timers/core as a base was actually just about topic. But there is no dependency on it. (And actually sched/core would have been a better choice for a base if any). So in order to fix the conflict and minimize the dependencies, I just rebased the patches on top of tip:core/locking only. No merge on top of that. It seems to work pretty well. The pullable result is in sched/cputime on my tree. Let me know if that's ok for you. Thanks.