Re: [Xen-devel] [PATCH v3 0/6] x86/time: PVCLOCK_TSC_STABLE_BIT support

[Joao Martins <joao.m.martins@xxxxxxxxxx>]

[Missed CC-ing the maintainers in the cover letter, my apologies]

On 08/24/2016 01:43 PM, Joao Martins wrote:
> Hey!
> 
> This is v3 on the pvclock TSC stable bit series.
> 
> Complete changelog on individual patches but overall is addressing
> Jan's comments, plus some other changes regarding the recent monotonicity
> improvements on x86/time.
> 
> Series is divided as follows:
> 
>  R      * Patch 1: Small refactor around init_platform_time to reuse
>                    initialization code when switching to TSC.
>  U      * Patch 2: Adds a new clocksource based on TSC
>  U,U    * Patch 3, 4: Adjustments for patch 5
>  U      * Patch 5: Implements the PVCLOCK_TSC_STABLE_BIT
>  N      * Patch 6: Document new clocksource
> 
> [ R := Reviewed-by ;; U := Updated ;; N := New ]
> 
> I kept the series the same but a fundamental difference from previous
> versions is that I stop clocksource=tsc from being used at all if hotplug is
> possible. To facilitate the review I kept it on Patch 5 as originally posted,
> whereas clocksource is added in Patch 2. But if preferred I can merge these 
> two.
> 
> The main benefit of this series is two-fold:
>  
>  1. Provide the guarantee of monotonic results on xen own system time as seen
>  by any cpu when using TSC as clocksource.
> 
>  2. Provide this same guarantee to guests and thus set the
>  TSC_STABLE_BIT (both FreeBSD and Linux support it) which then allows guests 
> to
>  skip expensive monotonicity check between PV CPU time infos. Plus, on Linux
>  specifically this also means that it could support vDSO which greatly 
> increases
>  performance (x10) for gettimeofday and clock_gettime since it would no
>  longer need to do the system call to get a reliable snapshot of system time.
>  For a reference on my laptop the speed of gettimeofday under xen pvclock is 
>  ~2 Mops/sec (Million ops per sec) whereas with vDSO it's on the range
>  of ~22 Mops/sec on <= 4.4 kernels and ~37 Mops on >= 4.5.
>  
>  Doing a long running time warp test for the past days on a dual-socket 
> Haswell
>  machine and I haven't yet seen time going backwards.
> 
> Thanks!
> Joao
> 
> Joao Martins (6):
>   x86/time: refactor init_platform_time()
>   x86/time: implement tsc as clocksource
>   x86/time: streamline platform time init on plt_update()
>   x86/time: refactor read_platform_stime()
>   x86/time: implement PVCLOCK_TSC_STABLE_BIT
>   docs: update clocksource option
> 
>  docs/misc/xen-command-line.markdown |   6 +-
>  xen/arch/x86/platform_hypercall.c   |   3 +-
>  xen/arch/x86/time.c                 | 226 
> +++++++++++++++++++++++++++++++-----
>  xen/include/asm-x86/time.h          |   1 +
>  4 files changed, 205 insertions(+), 31 deletions(-)
> 

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel