linux-kernel - Re: Fw: [lkp-developer] [sched,rcu] cf7a2dca60: [No primary change] +186% will-it-scale.time.involuntary_context

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 14 Dec 2016 10:54:25 +0100
From:   Michal Hocko <mhocko@...nel.org>
To:     "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
Cc:     linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        peterz@...radead.org
Subject: Re: Fw: [lkp-developer] [sched,rcu]  cf7a2dca60: [No primary change]
 +186% will-it-scale.time.involuntary_context_switches

On Tue 13-12-16 07:14:08, Paul E. McKenney wrote:
> Just FYI for the moment...
> 
> So even with the slowed-down checking, making cond_resched() do what
> cond_resched_rcu_qs() does results in a smallish but quite measurable
> degradation according to 0day.

So if I understand those results properly, the reason seems to be the
increased involuntary context switches, right? Or am I misreading the
data?
I am looking at your "sched,rcu: Make cond_resched() provide RCU
quiescent state" in linux-next and I am wondering whether rcu_all_qs has
to be called unconditionally and not only when should_resched failed few
times? I guess you have discussed that with Peter already but do not
remember the outcome.

Thanks for letting my know! 

> I will try some things to reduce the
> impact, but it is quite possible that we will need to live with both
> interfaces.

Thanks a lot for your time!
 
> 							Thanx, Paul
> 
> ----- Forwarded message from kernel test robot <ying.huang@...ux.intel.com> -----
> 
> Date: Mon, 12 Dec 2016 13:52:28 +0800
> From: kernel test robot <ying.huang@...ux.intel.com>
> TO: "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
> Cc: lkp@...org
> Subject: [lkp-developer] [sched,rcu]  cf7a2dca60: [No primary change] +186%
> 	will-it-scale.time.involuntary_context_switches
> 
> Greeting,
> 
> There is no primary kpi change in this test, below is the data collected through multiple monitors running background just for your information.
> 
> 
> commit: cf7a2dca6056544bb04a8f819fbbdb415bdb2933 ("sched,rcu: Make cond_resched() provide RCU quiescent state")
> https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git dev.2016.12.05c
> 
> in testcase: will-it-scale
> on test machine: 32 threads Intel(R) Xeon(R) CPU E5-2680 0 @ 2.70GHz with 64G memory
> with following parameters:
> 
> 	test: unlink2
> 	cpufreq_governor: performance
> 
> test-description: Will It Scale takes a testcase and runs it from 1 through to n parallel copies to see if the testcase will scale. It builds both a process and threads based test in order to see any differences between the two.
> test-url: https://github.com/antonblanchard/will-it-scale
> 
> 
> 
> Details are as below:
> -------------------------------------------------------------------------------------------------->
> 
> 
> To reproduce:
> 
>         git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git
>         cd lkp-tests
>         bin/lkp install job.yaml  # job file is attached in this email
>         bin/lkp run     job.yaml
> 
> testcase/path_params/tbox_group/run: will-it-scale/unlink2-performance/lkp-sb03
> 
> 15705d6709cb6ba6  cf7a2dca6056544bb04a8f819f  
> ----------------  --------------------------  
>          %stddev      change         %stddev
>              \          |                \  
>     116286                      114432        will-it-scale.per_process_ops
>      20902 ±  5%       186%      59731 ±  5%  will-it-scale.time.involuntary_context_switches
>       2694 ±  8%        61%       4344        vmstat.system.cs
>      10903 ± 99%     -1e+04        148 ±  5%  latency_stats.max.wait_on_page_bit.__migration_entry_wait.migration_entry_wait.do_swap_page.handle_mm_fault.__do_page_fault.do_page_fault.page_fault
>       3583 ± 38%      1e+04      14010 ± 51%  latency_stats.sum.ep_poll.SyS_epoll_wait.entry_SYSCALL_64_fastpath
>       4143 ± 24%      1e+04      14549 ± 51%  latency_stats.sum.ep_poll.SyS_epoll_wait.do_syscall_64.return_from_SYSCALL_64
>     271108 ± 71%     -2e+05      66364 ± 32%  latency_stats.sum.wait_on_page_bit.__migration_entry_wait.migration_entry_wait.do_swap_page.handle_mm_fault.__do_page_fault.do_page_fault.page_fault
>     834637 ±  8%        62%    1351381        perf-stat.context-switches
>      16449 ±  3%        54%      25349 ±  3%  perf-stat.cpu-migrations
>      25.94              35%      35.02        perf-stat.node-store-miss-rate%
>  2.534e+09              32%  3.335e+09        perf-stat.node-store-misses
>  1.002e+12               4%  1.043e+12        perf-stat.dTLB-stores
>   50923913               3%   52692115        perf-stat.iTLB-loads
>  1.696e+12                   1.745e+12        perf-stat.dTLB-loads
>  1.258e+12                   1.291e+12        perf-stat.branch-instructions
>  6.132e+12                   6.274e+12        perf-stat.instructions
>       0.37                        0.38        perf-stat.ipc
>       0.37              -3%       0.35        perf-stat.branch-miss-rate%
>      29.83              -4%      28.66        perf-stat.cache-miss-rate%
>  1.117e+10              -4%  1.071e+10        perf-stat.cache-misses
>  7.232e+09             -14%  6.187e+09        perf-stat.node-stores
-- 
Michal Hocko
SUSE Labs